henry e. allison-essays on kant-oxford university press (2012).pdf

Essays on Kant

This page intentionally left blank

Essays on Kant

Henry E. Allison

1

3Great Clarendon Street, Oxford, OX2 6DP,United Kingdom

Oxford University Press is a department of the University of Oxford.It furthers the University’s objective of excellence in research, scholarship,and education by publishing worldwide. Oxford is a registered trade mark ofOxford University Press in the UK and in certain other countries

# in this volume Henry E. Allison 2012

The moral rights of the author have been asserted

First Edition published in 2012

Impression: 1

All rights reserved. No part of this publication may be reproduced, stored ina retrieval system, or transmitted, in any form or by any means, without theprior permission in writing of Oxford University Press, or as expressly permittedby law, by licence or under terms agreed with the appropriate reprographicsrights organization. Enquiries concerning reproduction outside the scope of theabove should be sent to the Rights Department, Oxford University Press, at theaddress above

You must not circulate this work in any other formand you must impose this same condition on any acquirer

British Library Cataloguing in Publication DataData available

Library of Congress Cataloging in Publication DataData available

ISBN 978–0–19–964703–3 (Hbk)ISBN 978–0–19–964702–6 (Pbk)

Printed in Great Britain byMPG Books Group, Bodmin and King’s Lynn

Links to third party websites are provided by Oxford in good faith andfor information only. Oxford disclaims any responsibility for the materialscontained in any third party website referenced in this work.

To the memory of my father, John P. Allison

Contents

Acknowledgments ixNote on sources and key to abbreviations and translations xi

Introduction 1

Part I

1. Commentary on Section Nine of the Antinomy of Pure Reason 15

2. Where Have all the Categories Gone? Reflections on Longuenesse’sReading of Kant’s Transcendental Deduction 31

A Response to a Response: An Addendum to “Where Have all theCategories Gone?” 43

3. Kant and the Two Dogmas of Rationalism 49

4. Transcendental Realism, Empirical Realism, and Transcendental Idealism 67

Part II

5. We Can Act Only Under the Idea of Freedom 87

6. On the Very Idea of a Propensity to Evil 99

7. Kant’s Practical Justification of Freedom 110

8. The Singleness of the Categorical Imperative 124

9. Kant on Freedom of the Will 137

Part III

10. Is the Critique of Judgment “Post-Critical”? 165

11. Reflective Judgment and the Application of Logic to Nature: Kant’sDeduction of the Principle of Purposiveness as an Answer to Hume 177

12. The Critique of Judgment as a “True Apology” for Leibniz 189

13. Kant’s Antinomy of Teleological Judgment 201

Part IV

14. The Gulf between Nature and Freedom and Nature’s Guarantee ofPerpetual Peace 217

15. Kant’s Conception of Aufklärung 229

16. Teleology and History in Kant: The Critical Foundations of Kant’sPhilosophy of History 236

17. Reason, Revelation, and History in Lessing and Kant 254

Bibliography 274Index 283

viii CONTENTS

Acknowledgments

I wish to thank the editors and publishers for their permission to reprint the followingessays:“Commentary on Section Nine of the Antinomy of Pure Reason,” Kritik der reinen

Vernunft: Ein kooperativer Kommentar, ed. M. Willaschek and J. Mohr, Frankfurt:Vittorio Klostermann (1998), 465–90, reprinted with the kind permission of AkademieVerlag.“Where Have all the Categories Gone? Reflections on Longuenesse’s Reading of

Kant’s Transcendental Deduction,” Inquiry, 43 (2000), 67–80, reprinted with the kindpermission of Taylor & Francis Group.“Kant and the Two Dogmas of Rationalism,” in A Companion to Rationalism, ed.

Alan Nelson, Oxford: Blackwell Publishing (2005), 343–59, reprinted with the kindpermission of John Wiley & Sons Ltd.“Transcendental Realism, Empirical Realism and Transcendental Idealism,”Kantian

Review, 11 (2006), 1–28.“We Can Act Only Under the Idea of Freedom,” (Presidential Address), The

Proceedings and Addresses of the American Philosophical Association, 71 (1997), 39–50.“The Very Idea of a Propensity to Evil,” Journal of Value Inquiry, 36 (2002), 337–48,

reprinted with the kind permission of Springer Science + Business Media B.V.“Kant’s Practical Justification of Freedom,” forthcoming in Kant on Practical Justifica-

tion, ed. Sorin Baiasu and Mark Timmons.“The Singleness of the Categorical Imperative,” Proceedings of the Eleventh Interna-

tional Kant Congress, Pisa 2010, ed. Stefano Bacin, Alfredo Ferrarin, Claudio La Rocca,and Margit Ruffing, New York: Walter de Gruyter (2012).“Kant on Freedom of the Will,” in The Cambridge Companion to Kant and Modern

Philosophy, ed. Paul Guyer, Cambridge: Cambridge University Press (2006), 381–415.“Is the Critique of Judgment ‘Post-Critical’?,” in The Reception of Kant’s Critical

Philosophy: Fichte, Schelling, & Hegel, ed. Sally Sedgwick, Cambridge: CambridgeUniversity Press (2000), 78–92.“The Critique of Judgment as a ‘True Apology’ for Leibniz,” in Kant und die

Berliner Aufklärung, Akten des IX. Internationalen Kant-Kongresses, ed. V. Gerhardt,R.-P. Horstmann, and R. Schumacher, New York: Walter de Gruyter (2001),286–99.“Reflective Judgement and the Application of Logic to Nature: Kant’s Deduction of

the Principle of Purposiveness as an Answer to Hume,” in Strawson and Kant, ed. Hans-Johann Glock, Oxford: Oxford University Press (2003), 169–83.

“Kant’s Antinomy of Teleological Judgment,” The Southern Journal of Philosophy, 30Supplement (1991), 25–42.

“The Gulf between Nature and Freedom and Nature’s Guarantee of PerpetualPeace,” in Proceedings of the Eighth International Kant Congress Memphis 1995, ed. HokeRobinson, Milwaukee: Marquette University Press (1996), 37–49.

“Kant’s Conception of Aufklärung,” in The Proceedings of the Twentieth World Congressof Philosophy, Modern Philosophy, Volume 7, ed. Mark D. Gedney, Philosophy Docu-mentation Center, Bowling Green State University (2000), 35–44.

“Teleology and History in Kant: The Critical Foundations of Kant’s Philosophy ofHistory,” in Kant’s Idea for a Universal History with a Cosmopolitan Aim: A Critical Guide,ed. Amelie Oksenberg Rorty and James Schmidt, Cambridge: Cambridge UniversityPress (2009), 24–45.

“Reason, Revelation, and History in Lessing and Kant,” Internationales Jahrbuch desDeutschen Idealismus (2009), 35–57.In addition, I wish to thank Greg Taylor for the preparation of the index.

x ACKNOWLEDGMENTS

Note on sources and key toabbreviations and translations

Apart from the Critique of Pure Reason, where references are to the standard A and B pagination,all references to Kant are to the volume and page of Kant’s gesammelte Schriften (KGS), herausge-geben von der Deutschen (formerly Konliglichen Preuissischen) Akademie der Wissenschaften,29 volumes [Berlin: Walter de Gruyter (and predecessors), 1902ff.] and second (where applica-ble) to the volume and pagination in The Cambridge Edition of the Works of Immanuel Kant,Cambridge: Cambridge University Press.

A Anthropologie in pragmatischer Hinsicht [KGS 7, 177–33] Anthropology froma Pragmatic Point of View, translated by Robert B. Louden in Anthropology, Historyand Education (231–429), The Cambridge Edition of the Works of Immanuel Kant,edited by Günter Zoller and Robert B. Louden, Cambridge: CambridgeUniversity Press (2007).

AF Anthropologie Friedländer [KGS 25, 469–728].

BA Beantwortung der Frage: Was ist Aufklärung [KGS 8, 33–42] “An Answer to theQuestion: What is Enlightenment?” translated by Mary Gregor, PracticalPhilosophy (16–22), The Cambridge Edition of the Works of Immanuel Kant,Cambridge: Cambridge University Press (1996).

Br Kant’s Briefwechsel [KGS 10–12] Correspondence, translated and edited by ArnulfZweig, The Cambridge Edition of the Works of Immanuel Kant, Cambridge:Cambridge University Press (1999).

EE Erste Einleitung in die Kritik der Urteilskraft [KGS 20] “First Introduction to theCritique of the Power of Judgment,” translated by Paul Guyer and EricMatthews, in Critique of the Power of Judgment, The Cambridge Edition of the Works ofImmanuel Kant (3–51), edited by Paul Guyer, Cambridge: Cambridge UniversityPress (2000).

EF Zum ewigen Frieden. Ein philosopischer Entwurf [KGS 8, 341–86] Toward PerpetualPeace: A Philosophical Project, translated by Mary J. Gregor in Practical Philosophy(317–51), The Cambridge Edition of the Works of Immanuel Kant, edited by MaryJ. Gregor, Cambridge: Cambridge University Press (1996).

GMS Grundlugung zur Metaphysik der Sitten [KGS 4, 385–463] Groundwork for theMetaphysics of Morals.

GTP Über den Gebrauch teleologischer Principien in der Philosophie [KGS 8, 157–84] “Onthe use of teleological principles in philosophy,” translated by Günter Zoller, inAnthropology, History and Education (192–218), The Cambridge Edition of the Worksof Immanuel Kant (107–20), edited by Günter Zoller and Robert B. Louden,Cambridge: Cambridge University Press (2007).

IAG “Idee zu einer allgemeinen Geschichte in weltbürgerlicher Absicht” [KGS 8, 15–31]“Idea of a universal history from a cosmopolitan point of view,” translated byAllen Wood, in Anthropology, History and Education, The Cambridge Edition of theWorks of Immanuel Kant (107–20), edited by Günter Zoller and RobertB. Louden, Cambridge: Cambridge University Press (2007).

ID De Mundi sensibilis atque intelligibilis forma et Principiis [KGS 2, 385–419] Concerningthe Form and Principles of the Sensible and Intelligible World (The Inaugural Dissertation)in The Cambridge Edition of the Works of Immanuel Kant, Theoretical Philosophy,1755–1770 (377–426), translated and edited by David Walford and RalfMeerbote, Cambridge: Cambridge University Press (1992).

JL Jäsche Logik [KGS 9, 1–150] The Jäsche Logic, translated by J. Michael Young, inLectures on Logic, The Cambridge Edition of the Works of Immanuel Kant, edited byJ. Michael Young (527–640), Cambridge: Cambridge University Press (1992).

KpV Kritik der praktischen Vernunft [KGS 5, 1–163] Critique of Practical Reason, translatedby Mary J. Gregor, in Practical Philosophy, The Cambridge Edition of the Works ofImmanuel Kant (139–271), edited by Mary J. Gregor, Cambridge: CambridgeUniversity Press (1996).

KrV Kritik der reinen Vernunft [KGS 3–4] Critique of Pure Reason, translated and editedby Paul Guyer and Allen Wood, The Cambridge Edition of the Works of ImmanuelKant, Cambridge: Cambridge University Press (1998).

KU Kritik der Urtheilskraft [KGS 5, 165–485]Critique of the Power of Judgment, translatedby Paul Guyer and Eric Matthews, The Cambridge Edition of the Works of ImmanuelKant, edited by Paul Guyer, Cambridge: Cambridge University Press (2000).

MAM “Muthsmasslicher Anfang der Menschengeschichte” [KGS 8, 107–23]“Conjectural beginning of human history,” translated by Allen Wood, inAnthropology, History and Education, The Cambridge Edition of the Works of ImmanuelKant (163–75), edited by Günter Zoller and Robert B. Louden, Cambridge:Cambridge University Press (2007).

MAN Metaphysische Anfangsgründe der Naturwissenschaft [KGS 4, 465–565] MetaphysicalFoundations of Natural Science, translated by Michael Freedman, in TheoreticalPhilosophy after 1781, 183–270, The Cambridge Edition of the Works of ImmanuelKant, edited by Henry Allison and Peter Heath, Cambridge: CambridgeUniversity Press, (2002).

M L1 Metaphysik L1, [KGS 28, 195–301] partially translated by Karl Ameriks and SteveNaragon, in Lectures on Metaphysics, The Cambridge Edition of the Works of ImmanuelKant (42–106), edited by Karl Ameriks and Steve Naragon, Cambridge:Cambridge University Press, (1997).

Met M Metaphysik Mrongovius [KGS 29, 745–940], translated by Karl Ameriks and SteveNaragon, in Lectures on Metaphysics, The Cambridge Edition of the Works of ImmanuelKant (109–286), edited by Karl Ameriks and Steve Naragon, Cambridge:Cambridge University Press (1997).

Moral M2 Moral Mrongovius II [KGS 29, 595–642].

xii NOTE ON SOURCES AND KEY TO ABBREVIATIONS AND TRANSLATIONS

MPC Moralphilosophie Collins [KGS 27, 241–471] translated by Peter Heath inLectures on Ethics (41–222), The Cambridge Edition of the Works of Immanuel Kant,edited by Peter Heath and J. B. Schneewind, Cambridge: Cambridge UniversityPress (1997).

MS Metaphysik der Sitten [KGS 6, 202–493] The Metaphysics of Morals (365–603),translated by Mary J. Gregor in Practical Philosophy, The Cambridge Edition of theWorks of Immanuel Kant (365–603), edited by Mary J. Gregor, Cambridge:Cambridge University Press (1996).

ND Principiorum primorum cognitionis metaphysicae nova dilucidatio [KGS 1, 385–46],NewElucidation of the First Principles of Metaphysical Cognition, Theoretical Philosophy,1775–1770, (1–45) translated by David Walford in collaboration with RalfMeerbote, Cambridge: Cambridge University Press (1992).

PRO Prolegomena zu einer jeden künftigen Metaphysik, die als Wissenschaft wird auftertenkonnen [KGS 4, 253–383], Prolegomena to any future metaphysics that will be able tocome forward as science, translated by Gary Hatfield, in Theoretical Philosophy after1781 (53–169), The Cambridge Edition of the Works of Immanuel Kant, edited byHenry Allison and Peter Heath, Cambridge: Cambridge University Press (2002).

R Reflexionen [KGS 15–19].

RGV Die Religion innerhalb der Grenzen der blossen Vernunft [KGS 6, 57–215] Religionwithin the boundaries of mere reason, translated by Allen Wood and George DiGiovani in Religion and Rational Theology (55–213), edited by Allen Wood andGeorge Di Giovani, The Cambridge Edition of the Works of Immanuel Kant,Cambridge: Cambridge University Press (1996).

RSV Recension von Schulz’s Versuch einer Anleitung zur Sittenlehre für alle Menschen,ohne Unterschied der Religion, nebst einem Anhange von den Todesstraffen [KGS 8,9–15] “Review of Schulz’s Attempt at an introduction to a doctrine of morals forall human beings regardless of different religions,” translated by Mary J. Gregor inPractical Philosophy (7–10), The Cambridge Edition of the Works of Immanuel Kant,edited by Mary J. Gregor, Cambridge: Cambridge University Press (1996).

SF Der Streit der Facultäten [KGS 7: 5–116] The Conflict of the Faculties, translated andedited by Allen Wood and George Di Giovani in Religion and Rational Theology(237–327), edited by Allen Wood and George Di Giovani, The Cambridge Editionof the Works of Immanuel Kant), Cambridge: Cambridge University Press (1996).

TP Über den Gemeinspruch: Das mag in der Theorie richtig sein, taugt aber nicht für diePraxis [KGS 8, 273–313] “On the common saying: That may be correct intheory, but it is of no use in practice”, translated by Mary J. Gregor in PracticalPhilosophy (279–309), The Cambridge Edition of the Works of Immanuel Kant, editedby Mary J. Gregor, Cambridge: Cambridge University Press (1996).

UE Über eine Entdeckung, nach der alle neue Kritik der reinen Vernunft durch eine ältereentbehrlich gemacht warden soll [KGS 8, 185–251] On a discovery whereby any newcritique of pure reason is to be made superfluous by an older one, translated by HenryAllison in Theoretical Philosophy after 1781 (283–336), The Cambridge Edition of the

NOTE ON SOURCES AND KEY TO ABBREVIATIONS AND TRANSLATIONS xiii

Works of Immanuel Kant, edited by Henry Allison and Peter Heath, Cambridge:Cambridge University Press (2002).

VEF Vorarbeiten zum ewigen Frieden [KGS 23, 152–92] Preliminaries to Perpetual Peace.

VMS Vorarbeiten zu Die Metaphysik der Sitten [KGS 23, 207–330] Preliminaries to theMetaphysics of Morals.

WHD “Was heisst: Sich im Denken orientiren?” [KGS 8, 131–47] “What does it meanto orient oneself in thinking?” translated by Allen Wood, in Religion and RationalTheology (1–18), The Cambridge Edition of the Works of Immanuel Kant, Cambridge:Cambridge University Press (1996).

xiv NOTE ON SOURCES AND KEY TO ABBREVIATIONS AND TRANSLATIONS

Introduction

This volume contains a collection of seventeen of my essays on various aspects of Kant’sphilosophy, all but one of which were written or published after 1996.1 Each of theseessays has been edited and revised for this volume. In most cases the revisions arerelatively minor, consisting mainly in the addition of notes and modest stylisticchanges. In others, the changes are more substantive and are aimed at removing or atleast lessening unclarity in the original version. In the most important of these I haveflagged the changes in a note so that a reader who wishes to can compare the revisedwith the original version. I have, however, resisted the temptation to rewrite orradically change any of the essays, since that would defeat the aim of providing arecord of my work in the shape that it has taken over the past fifteen plus years.Nevertheless, there is one completely new item, namely, an addendum to my criticalreflections on Beatrice Longuenesse’s important book (Kant and the Capacity to Judge) inwhich I respond to her previously published response to my original critique.Although these essays cover the full range of my work on Kant, most of them

revolve around three themes: the nature of transcendental idealism, freedom of thewill, and the purposiveness of nature, which, not coincidentally, I regard as the threefundamental conceptions in Kant’s “critical” philosophy. The first two of these havebeen the central foci of my work on Kant since its inception and were the subjectmatter of my first two systematic books on Kant and collection of essays.2 While I havenot radically altered my views on either issue, I believe that the essays in this volumedealing with them contain a significant development or at least clarification of theseviews. In some of them this takes the form of an attempt to place Kant’s views in abroader historical context by means of a comparison with those of other thinkersranging from Leibniz, Wolff, Crusius, and Hume to Kant’s idealistic successors. Thethird of these themes is central to the third Critique and has been a major concern of mymore recent work, which found its systematic expression in Kant’s Theory of Taste(2001a). But whereas that book is concerned primarily with Kant’s aesthetics, the essays

1 This was the year of the publication of my previous collection of essays on Kant (1996a). The exceptionis my “Kant’s Antinomy of Teleological Judgment” (1991a). Despite its earlier date I have decided to includeit in this collection because it is the only writing of mine devoted to this topic and it complements a numberof the other essays dealing with various aspects of Kant’s views on teleology.

2 These are Allison (1983) [revised and enlarged edition (2004)] and (1990).

in this collection dealing with purposiveness and the reflective power of judgment areall concerned with the extra-aesthetic dimensions of these conceptions, which includesKant’s accounts of empirical concepts, induction, and teleology.The essays are divided into four parts and, with one exception, are ordered chrono-

logically within each part.3 The first part, which consists of four essays and the above-mentioned addendum, is concerned with topics in Kant’s theoretical philosophy,specifically, the first Critique. The second is made up of five essays dealing withKant’s practical philosophy, with an emphasis on the concept of freedom. The thirdis composed of four essays involving topics from the third Critique. The final partcontains four essays that discuss a variety of topics, including Kant’s political philoso-phy, his philosophy of history, and philosophy of religion. These are likewise related, ifsometimes indirectly, to the third Critique, inasmuch as they involve in various waysKant’s conception of the purposiveness of nature (including human nature). In whatfollows I discuss each of the essays in turn.

IThe first essay dealing with Kant’s theoretical philosophy, “Commentary on SectionNine of the Antinomy of Pure Reason,” was originally intended as a chapter coveringthis section in a co-operative commentary on the first Critique. Since Kant’s aim in thissection is to provide a “critical” resolution of each of the four antinomies, his argumentis extremely complex, and following the ground rules of the commentary, I endeavorto cover the account as a whole. But inasmuch as Kant’s main interest is with theresolution of the third antinomy, this receives the bulk of attention.Kant’s concern in the portion of the text devoted to the resolution of the third

antinomy is to demonstrate the implications of this resolution for the free will problem.As part of a conflict between cosmological claims regarding the sensible world as awhole, the antinomy proper concerns the apparent conflict between the demand ofreason for completeness in the explanation of phenomena, which is affirmed in thethesis and requires the postulation of an “absolute beginning,” and the claim of theantithesis that there can be no such beginning because every cause must itself have itsantecedent cause, ad infinitum. Kant’s resolution consists in the assertion that theseclaims are compatible, since the denial of unconditioned causality applies only withinthe course of nature, leaving conceptual space for an “intelligible cause” that meets thedemand of the thesis.4

Although at first glance the connection between this issue and the free will problemseems contrived, Kant argues for their connection by claiming that the “practicalconcept of freedom,” which is that of free will as ordinarily understood, includes thenotion of absolute spontaneity, which is also the content of the transcendental idea of

3 The exception is essay 13, which for thematic reasons is grouped last in Part III.4 For my fullest and most recent analysis of this antinomy see Allison (2004), 376–84.

2 ESSAYS ON KANT

freedom that was appealed to in the thesis of the antinomy. Kant further claims that,because of this inclusion, the free will problem requires a transcendental rather than apsychological (or physiological) resolution. I analyze this claim, which involves thedenial of the adequacy of an entirely naturalistic approach to the question of free will,and argue that Kant is best read as offering a conceptual thesis regarding what is builtinto the thought of rational agency rather than a metaphysical thesis regarding aninaccessible noumenal self.The second essay in the first part, “Where Have all the Categories Gone? Reflec-

tions on Longuenesse’s Reading of Kant’s Transcendental Deduction,” contains mycritique of Beatrice Longuenesse’s interpretation of the B-Deduction in Kant and theCapacity to Judge and was originally presented at an author meets critic session of theWestern Division of the American Philosophical Association in New Orleans in 1999.I have attempted to attain (at least for the time being) the last word on the matter byincluding an addendum in which I respond to Longuenesse’s initial response to mycritique.5 I wish to emphasize, however, that this is a friendly critique, since herremarkable book has been an important influence on my own work and she hasdone me the honor of stating that my interpretation of the B-Deduction in the originalversion of Kant’s Transcendental Idealism (1983) was the closest to hers in the literature.This relatively wide area of agreement has, however, led me to attach especialimportance to our disagreement, particularly since the latter involves something ascentral to Kant’s thought as the Transcendental Deduction. As I argue in my critique,despite the many virtues of Longuenesse’s interpretation, I do not think that she hasprovided a satisfactory account of how Kant argues for the necessity of the categories inthe two parts of the B-Deduction and how the two parts were intended to yield asingle proof. And, as I note in my addendum, this remains the case after her response tomy original critique.The third essay of this part: “Kant and the Two Dogmas of Rationalism,” was

written for a collection of essays on classical rationalism.6 I used the occasion to give asystematic expression to a line of thought that was suggested to me by Philip Kitcher’sremark that in the first Critique Kant came close to anticipating the central thesis ofQuine’s “Two Dogmas of Empiricism.” Taking this as my point of departure, I arguethat Kant actually came closer to providing a critique of what may be called the “twodogmas of rationalism.” These are the propositions that in every true proposition thepredicate is contained in the concept of the subject and that sensible cognition isreducible in principle (if not for the human understanding) to purely conceptual orintellectual cognition. The former is often referred to as the “predicate-in-notionprinciple” and I have termed the latter the “reducibility principle.” I claim not onlythat they are mirror images of Quine’s more famous dogmas and that they reciprocallyimply one another, but that Kant’s critique of these dogmas provides an invaluable

5 For Longuenesse’s response, see Longuenesse (2000). 6 Allison (2005).

INTRODUCTION 3

framework for understanding the fundamental differences between his “critical” phi-losophy and classical rationalism (including but not limited to the philosophy ofLeibniz). In particular, it shows that the fundamental issue for Kant is not how syntheticjudgments are possible a priori but how they are possible at all.The final essay of the first part, “Transcendental Realism, Empirical Realism, and

Transcendental Idealism,” is an attempt to clarify my views on the nature of Kant’stranscendental idealism and to respond to some of my critics. Although its centralclaim, namely, that the key to understanding Kant’s idealism is to view it in contrast tothe transcendental realism to which he opposes it and that this leads to a recognition ofthe methodological or meta-philosophical (as opposed to metaphysical) nature of thisidealism is not new, I believe that the discussion goes beyond my earlier treatments ofthis topic in at least four respects.First, it develops a line of thought that I introduced in the revised version of Kant’s

Transcendental Idealism (2004), according to which transcendental realism is identifiedwith any view that regards the proper objects of human cognition as things or objects ingeneral. This appears to conflict with Kant’s own tendency to characterize such realismin terms of its identification of “mere appearances,” that is, objects of possible experi-ence, with things in themselves; but by examining the largely unexplored relationshipbetween the concept of an object in general, that is, an object as such or qua object, anda thing (or object) considered as it is in itself, I argue that there is no conflict and that thisway of framing the issue leads to a better understanding of what Kant is really up to withhis limitation of cognition to objects of possible experience. Second, I analyze Kant’sempirical realism more carefully than I had in the past and argue that, far from being aform of phenomenalism or subjective idealism, as some critics continue to maintain, it isin fact a robust form of realism that is quite adequate to our cognitive needs and thatavoids the excesses of a transcendental realism that Kant exposes in the TranscendentalDialectic. Third, I argue that the familiar objections to transcendental idealism rest uponan implicit commitment to transcendental realism and therefore beg the questionagainst this idealism. Finally, I attempt to turn the tables on those interpreters whopurport to defend significant portions of Kant’s account of experience, while disparag-ing his transcendental idealism, by calling upon them to explain why they think that anempirical realism is not sufficient, but a transcendental realism is required. Underlying thischallenge is an analysis of the latter according to which it holds that space, time, and thecategories apply to objects in general or as such rather than merely to objects of anyexperience that is possible for beings with our forms of cognition. I maintain that this isthe true form inwhich the issue should be posed and that when it is posed in this way thecase for transcendental idealism becomes compelling.

IIThe first essay in the part of this volume devoted to Kant’s practical philosophy, “WeCan Act Only Under the Idea of Freedom,” was initially delivered as a presidential

4 ESSAYS ON KANT

address to the Pacific Division of the American Philosophical Association in BerkeleyCalifornia in 1997.7 Accordingly, it was written with a general philosophical audiencein mind rather than one composed of Kantians or even philosophers with a stronginterest in Kant. I chose as my text this famous passage from the Groundwork becauseDaniel Dennett had helped himself to the phrase “Acting under the Idea of Freedom”

as the title for a central chapter in his influential book Elbow Room, which deals with thefree will problem from an uncompromisingly naturalistic perspective.In addition to making my views accessible to such an audience, the essay has a two-

fold aim. One is to query Dennett’s approach to the issue, which leads him to construethe necessity of acting under the idea of freedom as reducible to a deliberator’sunavoidable ignorance regarding the outcome of a proposed course of action. Againstthis, I argue that Kant’s account of rational agency makes a persuasive case forattributing to an agent, at least from a first-person perspective, a genuine spontaneitywhich eludes the naturalistic framework that is assumed by Dennett and many others tobe all-encompassing.The other is to counter John McDowell’s dismissive treatment of Kant’s idealism in

Man and World. Although such a treatment is quite common in the literature, I focus onMcDowell because he is virtually alone in taking an essentially Kantian view ofspontaneity seriously and rejecting a “bald naturalism,” of the kind proposed byDennett. In light of this, I question McDowell’s understanding of transcendentalidealism, which he seems to have taken over whole cloth from Strawson, and suggestthat his own account rests ultimately on an idealistic commitment of a methodologicalsort.The second essay in this part, “On the Very Idea of a Propensity to Evil,” was

written for a special issue of the Journal of Value Inquiry devoted to Kant’s moralphilosophy.8 Like my previous treatments of this topic, it emphasizes the still notalways appreciated fact that by “radical evil” Kant understood the root or condition ofthe possibility of all moral evil, rather than very great evil, and it seeks to ground thisconception in Kant’s account of rational agency.9 It differs from my other treatments,however, in two respects. First, it contains a response to Allen Wood’s attempt to“anthropologize” radical evil by identifying it with “unsociable sociability.” WhileI acknowledge that the connection between moral evil (in all forms) and unsociablesociability is an important dimension of Kant’s thought, I deny that it is the wholestory. Second, I argue that the part of the story that eludes Wood’s anthropological andtherefore naturalistic account is conceptual rather than metaphysical in nature. Morespecifically, I argue that Kant’s account is grounded in a conceptual analysis of theconditions of imputation, that is to say, what must be presupposed if one takes moralevil seriously. As such, it shares with the previous essay an attempt to find a third way

7 It was first published as Allison (1997a). 8 Allison (2002).9 These include Allison (1990), 146–61, (1991b), and (1996a), 169–82.

INTRODUCTION 5

between an impoverished naturalistic and an objectionable metaphysical approach toKant’s conception of freedom.The third essay in the second part of this volume, “Kant’s Practical Justification of

Freedom,” was written for a collection of essays on practical justification in Kant andhas two aims.10 First, it provides a brief account of what Kant understands by a practicaljustification and the various types of such justification found in his writings. Second, itexamines the different ways in which Kant attempts to provide a practical justificationof freedom in various texts, chiefly the Groundwork and the second Critique. Its mainfocus is on the tension between Kant’s attempt in the former work to ground thenecessity of presupposing freedom in our conception of ourselves as rational agents andtherefore independently of any specifically moral considerations (“We can act onlyunder the idea of freedom”) and his view in the latter that, “[H]ad not the moral lawalready been distinctly thought in our reason, we should never consider ourselvesjustified in assuming such a thing as freedom” (KpV 5: 4n).I resolve the tension by linking the former with freedom as spontaneity and the latter

with freedom as autonomy. In other words, while the conception of ourselves asrational agents, that is, as beings with the capacity to determine ourselves to act on thebasis of reasons (including but not limited to moral reasons), suffices to show thenecessity of presupposing freedom in the sense of spontaneity, it is only the conscious-ness of standing under the moral law (as a “fact of reason”) that brings with it anawareness of our moral autonomy, understood as a capacity to determine oneself to acton the basis of purely moral considerations, independently of any desires and needs wehave as beings with a sensuous as well as a rational nature. And, in this context,I introduce the controversial but to my mind fundamental claim, which I have alsoargued for elsewhere, that the reason why the “critical” Kant eschews a theoreticalproof of freedom is not merely that such a proof is beyond our cognitive capacity, butalso because transcendental idealism, properly understood, precludes the assumptionthat there is some fact of the matter regarding freedom (whether empirical or nou-menal) that could be established theoretically if we had the capacity.The fourth essay in this part, “The Singleness of the Categorical Imperative,” was

initially presented at the Eleventh International Kant Congress at Pisa in 2010. Itsconcern is with the concept of the categorical imperative and its aim is to lay thefoundation for a defense of Kant’s thesis that this imperative is “only a single one”(GMS 4: 421). This thesis needs a defense because in Groundwork 2 Kant presents anumber of formulations of this imperative (the precise number itself being a subject ofdispute), which, to say the least, are not obviously equivalent.My strategy for this defense is to try to show that these formulations are the result of

Kant’s attempt to provide a complete construction of the concept of the categoricalimperative by linking them with various steps in a progressive account of finite rational

10 Allison (forthcoming).

6 ESSAYS ON KANT

agency that constitutes the organizing principle of Groundwork 2. By the completeconstruction of the concept I understand an account of the necessary and sufficientconditions of the possibility of the categorical imperative. I term Kant’s analysis of theconception of rational agency progressive because each step, which is correlated with adistinct formula, adds a fresh dimension to the conception of such agency. My claim isthat this makes it possible to view these formulas as extensionally but not intentionallyequivalent, that is to say, as yielding the same results when applied to the same cases,though not for the same reasons. In so doing, I occupy a middle ground between theviews of Wood, who denies that these formulations are equivalent in any meaningfulway, and Onora O’Neill, who maintains that they are intensionally as well as exten-sionally equivalent.The final essay in this part, “Kant on Freedom of the Will,” provides an overview of

Kant’s accounts of freedom from the 1770s through the “critical” period.11 It discussesthe various conceptions of freedom to be found in the Kantian texts, analyzes theconnection between these conceptions and Kant’s moral theory, explores the contrastbetween the empirical and the intelligible character of the will as contained in the firstCritique, and examines Kant’s controversial views on the relation between freedom andcausal determinism. The distinctive feature of this essay, however, is its attempt toframe Kant’s account of free will historically in relation to the views of both hisimmediate predecessors, who exerted the most influence on him, and his idealisticsuccessors, whose views he greatly influenced. The former group is composed ofLeibniz, Wolff, and Crusius and the latter of Fichte, Hegel, and Schopenhauer.

IIIThe four essays in the third part of this volume are concerned with the third Critique.They discuss the place of this work in Kant’s “critical” philosophy as a whole, theimport of the conception of the logical or formal purposiveness of nature, whichappears in both versions of the Introduction as the a priori principle pertaining to thepower of judgment, and the antinomy of the teleological power of judgment. The firstof these, “Is the Critique of Judgment ‘Post-Critical’?” was initially presented at aconference devoted to “The Idea of a System of Transcendental Idealism in Kant,Fichte, Schelling,” which was held at Dartmouth College in August 1995.12 The essayis largely a response to Burkhard Tuschling’s provocative thesis that the later Kantabandoned his original “critical” view as expressed in the first Critique and relatedwritings in favor of a “speculative” approach and that this is evident not only in thefragmentary Opus postumum, which is the main focus of Tuschling’s Kant-interpreta-tion, but already in the third Critique. My quarrel is not with his account of Kant’sviews in the Opus postumum, about which I remain skeptical but undecided, but rather

11 Allison (2006b). 12 Allison (2000c).

INTRODUCTION 7

with his further claim that an abandonment of “critical” principles is to be found in thethird Critique. Against this I maintain that, while it is certainly true that, by giving acentral place to the reflective power of judgment, the third Critique contains a signifi-cant development of Kant’s previous thought, this involves a deepening rather than anabandonment of “critical” principles. More specifically I argue not only againstTuschling but against Guyer, Horstmann, and others as well, that Kant’s account ofthe principle of “logical” or “formal” purposiveness contains a line of thought that wasalready in place in the Appendix to the Transcendental Dialectic in the first Critique.Finally, in support of the “critical” nature of Kant’s position in the third Critique,I contrast Kant’s conception of and appeal to the concept of an intuitive intellect withthat of the young Hegel to which Tuschling refers.The next two essays in the third part are “Reflective Judgement and the Application

of Logic to Nature: Kant’s Deduction of the Principle of Purposiveness as an Answer toHume” and “The Critique of Judgment as a ‘True Apology’ for Leibniz.” The formerwas initially delivered at a conference on Strawson and Kant at Reading, UK inSeptember of 1999.13 The latter was presented at the Ninth International KantCongress in Berlin in March of 2000.14 Its title expresses my view that Kant’ssomewhat ironical remark at the end of his response to Eberhard that “the Critique ofPure Reason might well be the true apology for Leibniz” (UE 8: 250) is also applicable(without any irony) to the third Critique.Even though one of these essays is concerned with Kant’s relation to Hume and the

other with his relation to Leibniz, they are companion pieces, closely related in bothtime of composition and content. As a result, they contain a partial overlap in theirtreatments of Kant’s transcendental deduction of the principle of logical or formalpurposiveness. Nevertheless, I have included both essays in this volume becauseI believe that they illustrate how a single line of argument can be seen as having twoquite distinct targets or, colloquially expressed, as killing two birds with one stone.Specifically, they jointly show how the view articulated in the deduction of thisprinciple and, more generally, in Kant’s account of the logical or formal purposivenessof nature, provides the basis for both a “critical” reformulation of the argument forpresupposing the existence of natural kinds mounted by Leibniz against Locke in hisNew Essays on Human Understanding and a powerful response to Hume’s skepticalchallenge to the rational grounding of induction. And for this reason I argue that thisaccount must be placed at the center rather than the periphery of Kantian epistemolo-gy, which reinforces the point made in response to Tuschling regarding the “critical”status of the third Critique.The final essay in this part, “Kant’s Antinomy of Teleological Judgment,” was

initially presented at the tenth Spindel Conference at the University of Memphisin October 1991 on the topic of “System and Teleology in Kant’s Critique of

13 Allison (2003). 14 Allison (2001b).

8 ESSAYS ON KANT

Judgment.”15 It attempts to unravel Kant’s perplexing claims that there is the appear-ance of an antinomy between the mechanistic and teleological principles and that thisappearance stems from the confusion of a principle of reflective with one of thedeterminative power of judgment (KU 5: 389).These claims are perplexing because they raise at least three fundamental questions

with which the essay endeavors to deal. (1) What does Kant understand by the“principle of mechanism” and how does it differ from the causal principle of the firstCritique? (2) Given Kant’s views on the requirements for an antinomy, how could hehave thought that the power of judgment is the source of the appearance of one? This isa problem because this power in its determinative capacity does not have any principlesof its own capable of generating even the appearance of an antinomy, while the conflictbetween regulative principles or maxims of the reflective power of judgment that Kantdescribes, seems to yield an actual antinomy rather than merely the appearance of one.(3) What is the relation between Kant’s claim that, as pertaining to the reflective ratherthan the determinative power of judgment, the opposing principles of mechanism andteleology are merely regulative rather than constitutive and the resolution of theantinomy? This is perhaps the most fundamental question because if, as the initialformulation suggests and many interpretations assume to be the case, the resolutionconsists merely in showing that these principles are regulative, then, in addition to theseldom noticed fact that this generates rather than resolves an antinomy, the vast bulkof Kant’s account, which includes a discussion of some of the most suggestive yetobscure conceptions in the third Critique, including the concept of an intuitiveintellect, is irrelevant to the task. By focusing on these aporia of Kant’s account, theessay attempts to provide a plausible reading of the antinomy, which I believe to beessential to an understanding of Kant’s complex views regarding the reflective power ofjudgment as a whole.

IVAlthough the four essays constituting the fourth part of this collection deal with avariety of topics, each of them involves in some way a consideration of Kant’sphilosophy of history and therefore a relation to the third Critique, where we findthe “critical” foundations of this aspect of Kant’s thought. The first of these, “The Gulfbetween Nature and Freedom and Nature’s Guarantee of Perpetual Peace,” waspresented in a symposium on Kant’s conception of perpetual peace at the EighthInternational Kant Congress in Memphis in March 1995.16 It focuses on the relationbetween Kant’s famous essay on the topic and the nature–freedom problem as posed inthe published Introduction to the third Critique. The problem is central to Kant’sphilosophy of history and as formulated in the third Critique it concerns the “immense

15 See note 1. 16 Allison (1996b).

INTRODUCTION 9

gulf” [grosse Kluft], which separates the supersensible from appearances and whichseems to stand in the way of their influencing one another (KU 5: 175). This gulf,which concerns our ways of conceptualizing the domains of nature and freedom, arisesbecause the ends set by the “laws of freedom,” that is to say, the various aspects of thehighest good, ought to be realized in the sensible world, while this world is supposedlygoverned by mechanistic laws of nature that are indifferent to any such ends. Afterspelling out Kant’s general approach to the problem and its connection with hisconception of the reflective power of judgment, I apply this analysis to his accountin Toward Perpetual Peace of how human nature may, apart from any moral considera-tions, be thought to lead humankind to a form of political organization that willinvolve the institution of republican governments and peaceful relations betweenstates.The second essay in this part, “Kant’s Conception of Aufklärung,” was presented at a

session of the Twentieth World Congress of Philosophy in Boston in August of1998.17 It analyses Kant’s conception of enlightenment as it is contained not only inhis essay: “An answer to the question: What is enlightenment?” (1784), but also in hiscomparatively neglected discussions of the topic in “What does it mean to orientoneself in thinking?” (1786) and the third Critique, as well as a related discussion in theAnthropology. It argues that, contrary to the view of Mendelssohn and many of hiscontemporaries for whom enlightenment consists essentially in the acquisition of basictruths regarding the human condition, Kant viewed it in more negative and practicalterms as the escape from a condition of “Unmündigkeit,” understood as an incapacity touse one’s understanding without the direction of others. As such, enlightenment forKant is best expressed in the motto “Sapere aude” (Dare to think for yourself or use yourown understanding) and involves the will as well as the intellect. This connection ofenlightenment with the will is affirmed in Kant’s controversial claim that this conditionof Unmündigkeit is self-incurred and therefore something for which the individual isresponsible. This has been criticized in both Kant’s time and our own and I defendKant against this criticism, suggesting a comparison with Dostoevsky’s “Grand Inquisi-tor.” In addition, I argue that, properly understood, Kant’s conception of enlighten-ment provides the basis for a powerful response to Gadamer’s well-known critique ofenlightenment as a “prejudice against prejudice,” which denies the importance oftradition.The third essay in this part, “Teleology and History in Kant: The Critical Founda-

tions of Kant’s Philosophy of History,” was written for a co-operative commentary onKant’s essay: “The Idea for a Universal History with a Cosmopolitan Aim.”18 Myassigned text for this commentary was the first proposition: “All natural predispositions ofa creature are determined sometime to develop themselves completely and purposively” (IAG 8:18). By exploring the connection between this principle, which Kant applies in the

17 Allison (2000a). 18 Allison (2009a).

10 ESSAYS ON KANT

remaining eight propositions to human nature and the history of humankind, and thecentral ideas of the Critique of the Teleological Power of Judgment, I argue that Kant’sphilosophy of history is fully “critical” rather than being, as is sometimes maintained,merely an exercise in a somewhat whimsical and dogmatic speculation, which isreserved for “occasional pieces.”Since the view of history that Kant sketches in this essay is for the most part the same

as the one provided eleven years later in Toward Perpetual Peace, regarding republicanforms of government and an enduring peace between states as the outcome of anhistorical process to which humankind is led against its collective will by what has aptlybeen referred to as the “cunning of nature,” my discussion covers some of the sameground as the essay devoted to that work. It differs from it, however, in two respects.First, it devotes much more attention to the details of Kant’s account of teleology in thethird Critique, including his conception of an organism, which is the central conceptionin his philosophy of biology, and the distinction between an intrinsic and an extrinsicpurposiveness. Second, it considers the trans-political aim of history to which Kantalludes in the earlier essay but which lies outside the scope of the later one, namely, thesuggestion that history has as its ultimate goal not simply the above-mentioned politicalends, but also the collective realization (by humankind as a whole) of the highest good.It is this goal, which in the third Critique Kant refers to as the “final end of creation,”that constitutes the link between Kant’s philosophy of history and philosophy ofreligion, a topic which I explore in the final essay.This essay, “Reason, Revelation, and History in Lessing and Kant,”was written for a

volume devoted to the topic of faith and reason.19 I added history to the assigned topicand used the occasion to contrast the views of Lessing and Kant on these issues. Thisreflects my long-held belief that what made Lessing and Kant the two most importantreligious-philosophical thinkers of the Aufklärung is that they both not only sharplydistinguished between the question of the truth of the Christian religion and thefacticity of the historical events in which it originated, arguing that Christianitypossesses an “inner truth” that is logically independent of its historical foundations,but in the process proposed an historical understanding of revealed religion in generaland Christianity in particular, which effectively merges the philosophy of religion intothe philosophy of history.Although this historicization is most evident in Lessing’s “The Education of the

Human Race,” in which he sketches in broad strokes a view of religious history ascomposed of three stages: the religion of the ancient Hebrews, Christianity, and a stillawaited “eternal Gospel,” I argue that it also plays an important, albeit more subterra-nean role in Kant’s Religion within the Boundaries of Mere Reason. I further argue,however, that, despite this commonality, which includes a relatively conservativestance toward aspects of traditional Christianity, such as original sin, the doctrine of

19 Allison (2009b).

INTRODUCTION 11

the Trinity, and the vicarious satisfaction, which were anathema to many of itsenlightenment critics, there are important differences between their views. Moreover,I maintain that these differences are due largely to the fact that, whereas for LessingChristianity is merely a stage, albeit an important one, in the religious development ofhumankind in which it is destined to be eventually replaced by a quasi-Spinozistic formof rational religion, Kant, notwithstanding the heterodox nature of many of his views,was committed to a form of Christian theism in which a purified, that is, moralizedform of Christianity emerges as the true moral and therefore rational religion.

12 ESSAYS ON KANT

PART I

Essay One

The Antinomy of Pure ReasonSection Nine (A515–67/B543–95)

(1) Section nine of “The Antinomy of Pure Reason,” which is entitled “On theempirical use of the regulative principle of reason, in regard to all cosmological ideas”(KrV A515/B543) is the last and lengthiest of the divisions of the chapter on theantinomy.1 Its function is to spell out the positive implications of the resolution of theantinomy that arises from the application to the sensible world of the illusory principleof reason: “If the conditioned is given, then the whole sum of conditions, andhence the absolutely unconditioned is also given” (KrV A409/B436). In previoussections, Kant had argued that, in its application to the sensible world, this principleyields a four-fold antinomy, based on the four cosmological ideas, each of which is anidea of the totality of appearances considered with respect to certain categories: thetotality of events in time and objects in space; of the parts of any given compositesubstance; of causes for any given event; and of contingent beings. Kant also claimed tohave shown that in each case it is possible to provide equally compelling proofs forantithetical propositions regarding this totality, namely, that it requires some uncondi-tioned first member (the thesis) and that it must be thought as an infinite series, only thetotality of which is unconditioned (the antithesis).Kant’s generic solution to this antinomial conflict turns on an appeal to transcen-

dental idealism. He argues that, with this doctrine in place, it is possible to resolve theantinomy by showing that it rests on the misunderstanding (attributed to the opposingtranscendental realism) that the sensible world exists as a thing in itself.2 Operating onthe basis of this misunderstanding and proceeding according to the above-mentioned

1 Although parts of this essay have been substantially rewritten in an endeavor to attain greater clarity and tocorrect a number of errors and infelicitous formulations in the original version, I have largely retained the formatof the original, which was itself dictated by the need to provide uniformity in what amounted to a cooperativecommentary on the entireCritique of Pure Reason. Following this format, the first part of the essay is devoted to abrief statement of the position and function of the portion of text for which I was responsible (section nine of“The Antinomy of Pure Reason”); the second provides an overview of the structure and content of this section;the third and longest, which is itself subdivided into four parts, corresponding to the four antinomies, contains acommentary on the text. The original version also contained a fourth section dealing with “questions ofinterpretation,” which I have omitted because I have found it dated and substituting a more up-to-date andsubstantial discussion would have required a major expansion of length. Since the formatting of the originalversion did not allow for endnotes, the few that are contained here have all been added.

2 For my views on the relation between transcendental idealism and transcendental realism see Allison(2004), 20–49 and (2006b), which is included in this volume.

principle of reason, the transcendental realist inevitably falls prey to the illusion that thetotality of these conditions is already given, which entails that for each of the four ideasthere must be either some first, unconditioned condition or an infinite series ofconditions. By contrast, the transcendental idealist, having rejected this conception ofthe sensible world, is able both to reject this conclusion and to reinterpret reason’sdemand to seek the conditions for any given conditioned as a regulative rather than aconstitutive principle. So construed, the requirement to seek conditions is still in place,but without the transcendentally realistic assumption that the totality of conditions isalready given prior to the search and therefore needs merely to be made manifest. AsKant succinctly puts it, this thought that the conditions must be regarded as given[gegeben] prior to the search for them is replaced by the critical principle that this searchis “given to us as a problem” [aufgegeben] (KrV A498/B526).

(2) Given this result, the official concern of section nine is to determine the significancefor the investigation of nature of this principle of reason, construed as a demand to seekthe totality of conditions for any given conditioned. The claim is that the investigationof nature in light of this requirement, which amounts to a quest for explanatorycompleteness or closure, enables one to attain the legitimate end of reason in thetheoretical domain, namely, the securing of the widest possible empirical employmentof the understanding without succumbing to dialectical illusion (KrV A517/B545).Kant’s discussion of the first two cosmological ideas closely follows this line of

thought and consists essentially in the further development of the conclusions thatwere already affirmed in section eight. The first of these ideas has to do with what Kantterms the “cosmological idea of the totality of the composition of the appearances of aworld-whole” (KrV A517/B545), that is, the age and size of the world. Appealing tothe results of his initial analysis of the conflict (the first antinomy), Kant reaffirms thatthe world can be assigned neither a first beginning in time nor an outermost limit inspace, but that this does not entail that it is infinite in these respects. Instead, he claimsthat the empirical regress through which the magnitude of the world is determinedmust be thought as continuable indefinitely (ad indefinitum).The second idea concerns the composition of substances in the world, that is, the

“totality of division of a whole given in intuition” (KrVA523/B551). Since the analysis ofthe second antinomy has shown that the parts of which such substances are thought to becomposed do not exist in themselves prior to and independently of the empirical regressthrough which they are determined (the division), they cannot be said to be either finiteor infinite in number. Although this is in substantial agreement with the treatment ofthe first cosmological idea, for reasons to be considered later, Kant also insists that here theregress is to be thought as continuing to infinity (ad infinitum) rather than merelyindefinitely, even though one cannot posit an (actual) infinite number of parts.Despite this difference, both analyses presuppose and build upon what, up until this

point, would appear to be the basis of Kant’s analysis of the antinomy in all of its forms,namely, that in each case the conflict is resolved by showing that, rather than being

16 PART I

contradictory opposites, the thesis and antithesis are simply contraries and are bothfalse. But immediately upon completing this analysis, Kant greatly complicates mattersby adding a lengthy concluding remark in which he introduces the distinction betweenthe “mathematical” and the “dynamical,” which he had previously applied to both thecategories and principles (KrV A528/B556–A532/B560). In the case of the principles,which are defined as rules for the objective employment of the categories (KrV A161/B200), those designated mathematical are concerned with appearances qua intuited,that is, with the manner in which appearances must be given in space and time, if theyare to conform to the conditions of apperception (as extensive and intensive magni-tudes). By contrast, the dynamical principles concern the existence of appearances inspace and time and are conditions of experience as opposed to mere intuition (KrVA160/B199–A162/B202). For present purposes, however, what is important aboutthis latter set of principles (which includes that of causality and modality) is that, eventhough their scope is limited to objects of possible experience, they refer to what isthought rather than intuited. In short, they are concerned with the rules for the empiricalthought of an object or an objective state of affairs.Since the cosmological ideas (like the principles) are based on the categories, it is not

surprising that Kant thought that a similar distinction would also apply to them. What isperhaps surprising, as well as central to his entire account, is that Kant took this distinctionto require a radically different treatment of the two species of antinomy. Thus, we nowlearn that the solution sketched above, which asserts that the antinomy is resolved becausethesis and antithesis are both seen to be false, holds only for the antinomies arising fromthe first two or mathematical ideas and that in the case of the last two, or dynamical ideas,the conflict is resolved in a radically different manner, indeed, one which establishes thecompatibility rather than the falsity of thesis and antithesis.This difference stems from a difference in the nature of the items thought together

or “synthesized” in the respective ideas of totality. In the case of the mathematicalideas, all of the items are necessarily homogeneous in the sense of being intuitablespatiotemporal entities and occurrences. Consequently, the rejection of the transcen-dentally realistic assumption that the world or the elements of which it is composedexist in themselves leads naturally in each case to the rejection of both the thesis andantithesis. By contrast, in the case of the dynamical ideas, which concern the kind ofcausality operative in the world and the question of whether the contingency in theworld requires grounding in a being that exists necessarily, the conditions combinedtogether in the idea of totality may be heterogeneous in the sense that they need not allbe regarded as sensibly given, that is, they include those that are merely thought as wellas those that are intuited. Accordingly, the theses, which affirm an unconditioned,“free” causality and the existence of a necessary being, and the antitheses, which denyany such causality or a necessarily existing being, may both be true. At least there is noconflict between them, since the former refer to an “intelligible” (merely thinkable)ground that cannot be met with in experience, while the latter limit themselves towhat can be an object of possible experience.

SECTION NINE OF THE ANTINOMY OF PURE REASON 17

(3.1) Kant begins his solution to the first cosmological idea by noting that here, as in theother cosmological questions, the regulative principle for the empirical employment ofreason (always seek conditions) is grounded in the proposition that “in the empiricalregress there can be encountered no experience of an absolute boundary, and henceno experience of a condition as one that is absolutely unconditioned empirically”(KrV A517/B545). This is because any such experience would, by definition, have tocontain a bounding of appearance by nothing or the void, which is empiricallyimpossible. And since the present concern is solely with reason in its empirical employ-ment, empirical impossibility is sufficient to exclude such a bounding altogether.Moreover, Kant continues, the fact that any condition reachable in an empirical regress

must itself be regarded as empirically conditioned of itself yields the rule that, however farwemay advance in the series of conditions, we must always search for still higher members ofthe series, which may or may not be cognizable empirically (KrV A518/B546). In otherwords, while there is no point at which the search for conditions must in principle beterminated, there is also noguarantee that further conditionswill be empirically accessible. Inlight of this,Kant suggests that the operative question iswhether the regressmust continue toinfinity (ad infinitum) or merely that no limit can be specified, that it continues indefinitely(ad indefinitum). The bulk of the discussion is devoted to an argument for the latteralternative,which is already implicit in the denial of any guarantee that the further conditionsfor which we are required to search are even in principle accessible to empirical inquiry.Kant’s analysis turns largely on his idealistic or “constructivist” principle that the

magnitude of the world is to be determined by the magnitude of the empirical regressrather than vice versa (KrVA519/B547). Since theworld as a whole is never given to us inintuition to be inspected as a discrete object, we cannot determine its magnitude (age orsize) independently of the regress from conditioned to condition (from present to moreremote past states or from nearby to distant objects in space). In fact, the only claimwe areentitled to make about this regress is that for every given member of the series ofconditions we are required to seek out higher and more remote conditions (KrVA519/B547). This follows directly from the rule stated above, which is itself an expressionof the regulative demand of reason to seek the condition for every conditioned. As Kantnotes, however, this entails that, “[W]e can say nothing at all about the magnitude of theworld, not even that in it there is a regressus in infinitum (KrV A519/B547).” The latter ispresumably ruled out on the grounds that to affirm an infinite regress is to assume that theworld has an infinite magnitude (KrV A5420/B548), which has already been precludedby the rejection of any determinate claim regarding the magnitude of the world. Byelimination, then, we are left with the regressus in indefinitum, which is distinct from theregressus in infinitum, since it determines no magnitude in the object (KrV A520/B548).In light of these considerations, Kant offers two answers, one negative and the other

positive, to the cosmological question regarding the magnitude of the world. The firstand negative answer is that “The world has no first beginning in time and no outermostboundary in space” (KrV A520/B548). This is a straightforward denial of the thesisposition and it is based on the premise that such a boundary could never become an

18 PART I

object of possible experience because it would require the perception of something asbounded by an empty time or space, which Kant had already claimed to be impossible.The obvious question raised by this answer is how it differs from the claim of the

antithesis. Moreover, the issue is crucial; for unless Kant’s claim can be shown to differsignificantly from that of the antithesis, we would be forced to conclude that hissolution to the antinomy consists essentially in the endorsement of the latter, inwhich case the claim that there is a genuine antinomial conflict resolvable only by atranscendental critique would have to be rejected. Kant deals with this issue in animportant footnote in which he distinguishes both his mode of proof and conclusionfrom the dogmatic position of the antithesis. According to Kant, the latter is rooted inthe assumption that the world of sense is “a thing whose totality is given in itself priorto any regress,” from which it is concluded that it must be infinite (occupy all times andplaces), if it is to have any determinate location in space and time (KrV A521/B549n).This line of reasoning is dogmatic because, by assuming that the magnitude of the

world is given prior to the regress from conditioned to condition, it effectively treats itas a thing in itself, from which it validly (but not soundly) concludes that, since themagnitude of the world cannot be finite, it must be infinite. Otherwise expressed, thedogmatic version of the antithesis assumes that the world, considered as a totality, isactually gegeben, not merely aufgegeben. By contrast, Kant’s “critical” view, as reflectedin his first and negative answer to the question of the magnitude of the world, is basedon the premise that this magnitude is given only in the regress, that is, aufgegeben, fromwhich it follows merely that it is not finite rather than that it is infinite.This conclusion is reaffirmed in Kant’s second and affirmative answer to the question

of themagnitude of the world, which is that the regress in the series of appearances in thedetermination of the magnitude of the world proceeds in indefinitum (KrV A521/B549).But rather than reiterating the argument already advanced in support of this claim, Kantsimply notes its equivalence to the complex proposition that, even though no absolutemagnitude may be assigned to the sensible world, the empirical regress (through whichalone its magnitude can be determined) is subject to the rule that it must always advancefrom a conditioned member of the series to one more remote. And, as Kant points out,all of this is for the sake of the possible empirical employment of the understanding,which is the proper concern of reason in its theoretical use (KrV A522/B550).Kant takes pains, however, to obviate a possible misunderstanding of this rule of

empirical progression. To this end, he remarks that the rule does not prescribe a regresswithout end with respect to a determinate kind of appearance. For example, it does notentail that in tracing the generations of living beings we shall never arrive at a first pair or inthe series of cosmic entities an outermost sun. All that it requires is that the advance be“from appearances to appearances,” that is, from and to objects of possible experience (KrVA522/B550). Thus, it is only this respect that the advance must proceed in indefinitum.

(3.2) In sharp contrast to the preceding, Kant begins his discussion of the solution of thesecond cosmological idea by declaring that here the regress from conditioned to


condition does proceed in infinitum and not merely in indefinitum. The difference is thatin this case, unlike the previous one, we are presented with a given whole, so that wecan think the parts as already given together with the whole. Consequently, theprinciple that the parts are given only through the regress no longer applies. Instead,what is not given independently of the regress is the totality of the division of the parts,that is, its decomposition. Moreover, Kant maintains that this precludes us fromclaiming either that the composite is composed ultimately of simple parts or that, ifinfinitely divisible, it must consist of infinitely many parts.Since in the resolution of the first antinomy Kant explicitly rejects the option of an

infinite regress on the grounds that it entails an actual infinite magnitude, it might seemsurprising to find him now claiming that infinite divisibility does not entail an infinitenumber of parts in the composition. But this is because we are now dealing with asuccessive infinite, which, as such, can never be viewed as a whole (KrV A524/B552).Thus, the question of whether the regress from conditioned to condition is infinite ormerely indefinite turns on the issue of what may be assumed to be “given” as a whole.If, as in the case of the first regress, nothing is so given (since the magnitude of theworld is determined only in and through the regress), the regress can only be said tocontinue indefinitely; but if, as in the case of the second regress, a whole is given inintuition and the regress is to the parts of which this whole is composed, then itproceeds ad infinitum. As before, however, both the finitistic claim of the thesis (thatthe composite is composed ultimately of simples) and the infinitistic claim of theantithesis (that it is composed of an infinite number of parts) are ruled out on thegrounds that they are incompatible with the conditions of an empirical regress.In the remainder of this section Kant discusses the scope of this general principle

concerning the divisibility of the composite. Moreover, in so doing, he both indicates itsclose connection with the doctrine of the infinite divisibility of space and expresses hisopposition to various aspects of the Leibnizian position. To begin with, hemaintains that itapplies to delimited portions of space (Kant does notmention time), which are all infinitelydivisible “wholes”without being composed of an infinite number of parts. This reflects theview, affirmed in the observation on the thesis of the second antinomy, that space shouldbe viewed as a totum, rather than a compositum, since the parts are possible only in the wholeand not the whole in the parts (KrV A438/B466). And since it applies to space, Kantconcludes that it applies also to that which occupies space, namely, body, which is likewiseheld to be infinitely divisible without being composed of an infinite number of parts.It is the move from the infinite divisibility of space to that of things in space that

brings Kant into direct conflict with the Leibnizian position, as well as with his ownearlier view.3 As Kant here reconstructs this position, it attempts to block this move bymeans of an appeal to the concept of substance. Since matter or the composite is

3 In his 1756 essay “Physical Monadology” [Monadologiam Physicam] Kant maintained a view in which theinfinite divisibility of space, which is established by geometry, is compatible with the existence of physicalmonads or indivisible particles of matter. Kant retained this view in his writings of the sixties.

20 PART I

composed of substances, which, as the ultimate subjects of composition, must persistthroughout all decomposition, it follows (according to the Leibnizians) that matter orbody, in contrast to the space which it occupies, cannot be thought to be infinitelydivisible. If it were, it would mean that nothing would remain if all composition wereremoved (in thought), which conflicts with the conception of substance as persisting.As he does elsewhere in similar contexts, Kant admits that this objection would hold

if bodies were things in themselves, that is, objects thought merely through pureconcepts of the understanding, without any reference to sensibility and its conditions.But he denies that it applies to substance in appearance or matter, because the latter “isnot an absolute subject, but only an abiding image [beharrliches Bild] of sensibility; it isnothing at all save as an intuition in which nothing unconditioned is to encounteredanywhere” (KrV A525–6/B553–4). Despite its subjectivistic tone, Kant is here merelyexpressing the “critical” view that, insofar as we are dealing with appearances (in thetranscendental sense), that is, with objects considered as they appear, we cannot leaveout the formal conditions of their intuition (space and time), from which it follows thatwe cannot posit any “absolute subject” or simple substance, since such a subject,although in a sense required by thought, is incompatible with these conditions.4

In the final paragraph of this section, Kant denies that the infinite divisibility thesis isapplicable to organic matter, which again brings him into direct conflict with Leibniz,who held that the organization of matter and therefore “life” proceeds to infinity. Kantdenies that such a pan-vitalism is a “thinkable hypothesis” because it presupposes thatmatter is divided, prior to the regress, into a determinate, yet infinite, number of parts(discrete quantities). In other words, since, ex hypothesi, we are here dealing withdiscrete quantities (determinate organic beings), the thesis of infinite divisibility entailsan actual infinite number of such beings or parts. This leads, however, to the self-contradictory idea of a series which, as infinite, must be viewed as not completable, butwhich, as composed of discrete quantities or determinate things, must nevertheless beassigned a determinate number (KrV A527/B555). Much as in the critique of theinfinitistic position in the thesis of the first antinomy, Kant rejects this doctrine on thegrounds that it rests on the incoherent idea of a completed infinite series.

(3.3) Kant’s treatment of the latter two cosmological ideas is separated from hisdiscussion of the first two by the previously mentioned remark in which he distin-guishes between the mathematical and the dynamical transcendental ideas and theircorresponding antinomies. The analysis of the first of these dynamical ideas, which isthe main focus of our concern, provides the occasion for Kant to apply the generalresults of the third antinomy to the free will problem. Since the antinomy itself wasconcerned with the properly cosmological question of whether reason must admit anunconditioned first cause or posits an infinite chain of conditioned causes, its relevance

4 For Kant’s further development of this line of thought, see MAN, 4: 505–8.


to the latter problem is far from obvious and Kant takes great pains to spell out theconnection. He first calls attention to it in the observation on the thesis of theantinomy, where he notes that the transcendental idea of freedom, equated withabsolute spontaneity, is both an essential ingredient in and the problematic aspect ofthe ordinary “psychological concept” of freedom (KrV A448/B476). This conceptionof freedom of the will as involving absolute spontaneity (now characterized as “practi-cal freedom”) and the problem of its compatibility with the causality of nature thenbecome the central issue in the discussion of the idea in section nine.The importance that Kant himself places on the discussion is reflected in the compara-

tive length and complexity of his analysis. In contrast to the fairly cursory treatment givento the other three ideas, Kant provides an initial six-paragraph overview of the problem,arguing that it is transcendental rather than empirical and, as such, cannot be solvedwithout appealing to the transcendental distinction between appearances and things inthemselves (KrV A532/B560–A537/B565). This is followed by a schematic statement ofthe solution provided by transcendental idealism in terms of the distinction between anempirical and an intelligible character under the heading: “Possibility of causality throughfreedom unified with the universal law of natural necessity” (KrV A538/B566–A541/B569). And, in a lengthy final section, entitled “Clarification of the cosmological idea offreedom in its combination with universal natural necessity,” Kant applies the generalprinciples laid out in the resolution of the antinomy to the particular case of the humanwill and its agency (KrV A542/B570–A558/B586).Kant introduces the problem by noting that there are only two conceivable kinds of

efficient causality: according to nature and from freedom. The former is the conceptionaffirmed in the second analogy, according to which a state follows from a precedingstate in accordance with a universal rule. For present purposes, the relevant feature ofsuch causality is that, as subject to the conditions of time, the “causality of the cause”[Kausalität der Ursache], that is, its actual productive power or efficacy, must have itselfcome into being in time (otherwise its effect would have always existed), which entailsthat this causality must have had an antecedent cause, etc. By contrast, freedom, in thecosmological sense affirmed in the third antinomy, is defined as the power of beginninga state spontaneously [von selbst] (KrV A533/B561). Kant points out that freedom, soconstrued, is a transcendental idea, the object of which cannot be given in any possibleexperience. Its cosmological function, as affirmed in the thesis of the antinomy, is toprovide the closure or thought of absolute totality, which is demanded by reason andwhich cannot be provided by the former kind of causality.Returning to the point already made at KrV A448/B476, Kant next claims that the

practical concept of freedom, that is, the ordinary idea of free will, is based on thistranscendental idea (KrV A533/B561). We are further told that this is both the sourceof the difficulty in the traditional problem of freedom and the reason why it is agenuinely transcendental rather than a merely empirical problem. This means that it isessentially a conceptual issue, which cannot be resolved by appealing to psychologicalor physiological considerations.

22 PART I

The key to the transcendental nature of the problem lies in Kant’s underlying concep-tion of rational agency. He maintains that, while the human will shares with the animalwill (arbitrium brutum) the property of being sensuously affected, that is, moved by sensuousimpulses, it is also the case that these impulses are not sufficient to determine the humanwill or, more generally, the will of a finite rational agent. Such agents have (or at least takethemselves to have) a power of self-determination, which, Kant suggests, can only beconceived in terms of the idea of spontaneity or transcendental freedom. Moreover, thisself-determination is specificallymanifest in action based on the recognition of an ought. Ashe puts it in the decisive passage defining the problem:

[Practical freedom] presupposes that although something has not happened, nevertheless it oughtto have happened, and its cause in appearance was thus not so determining that there is not acausality in our power of choice such that independently of those natural causes, and evenopposed to their power and influence, it might produce something determined in the temporalorder in accord with empirical laws, and hence begin a series of occurrences entirely from itself.(KrV A534/B562)

Kant’s point is that to regard oneself as a rational agent is to take oneself as having thecapacity to act on the basis of the recognition of an ought and that this capacitypresupposes an independence from complete determination by natural causality.Thus, insofar as we take ourselves as rational agents, we also take ourselves as genuineself-determiners, whose choices and actions are not simply the causal consequences ofantecedent conditions. And this means that we attribute to ourselves a capacity for“absolute beginnings,”which is modeled on the transcendental idea that was initially athome in the cosmological domain.Given this analysis of rational agency and its conditions, the problem is to conceive

how freedom, so construed, could be compatible with the natural necessity thatgoverns human behavior, considered as a series of occurrences in the phenomenalworld. As is already clear from the previously discussed remark preceding the analysis ofthe dynamical ideas, the key to the answer lies in transcendental idealism. An appeal tosuch idealism is necessary in order to solve the problem because it provides the onlyvehicle for the creation of the conceptual space required for the thought of an“intelligible” (non-empirical) ground of appearance that could serve as a conditionof appearance, without itself being an empirically conditioned member of the series ofappearances. Thus, Kant points out that if appearances were things in themselves, therewould be no room for freedom, since nature would provide the complete andsufficient determining ground of every event (KrV A536/B564). This is no longerthe case, however, if we take appearances for no more than they really are, namely,“mere representations, connected in accordance with empirical laws,” for then, “theymust have grounds that are not appearances” (KrV A537/B565). Under this scenario,the effects of such a putative “intelligible causality” would be appearances and, as such,subject to the causality of nature, whereas this intelligible causality itself would not beso subject.


Although the outcome of this initial formulation of the problem is largely negative,consisting in the demonstration of the incapacity of the common standpoint oftranscendental realism, which regards appearances as things in themselves, to dealwith the problem of freedom, it is also positive to the extent that it shows that whatis required is an account of how the intelligible causality of freedom can be thought asthe ground of the empirical causality of nature. The underlying assumption is that onlyif the relation between the two species of causality can be conceived in this way canthey be conjoined in the determination of a given action. On any other view of theirrelation, either the merely intelligible, i.e., non-empirical, causality of freedom wouldhave to be regarded as redundant, since the sufficient conditions of the explanation ofthe action would be provided by the series of natural causes, or it would be necessary toassume a break in the causal order of nature, which would conflict with the conditionsof experience as set forth in the Analytic.Kant attempts to deal with this problem by introducing the conception of character

[Charakter] and the distinction between an empirical and an intelligible character.5 Inthe firstCritique, Kant uses the term to designate both a kind of cause and a causal agent.Every efficient cause, he tells us, “must have a character, that is, a law of its causalitywithout which it would not be a cause at all” (KrV A539/B567). The empiricalcharacter of a cause or causal agent is just its modus operandi, that is, its disposition ortendency to behave in certain ways under given conditions. As such, an empiricalcharacter is something that is inferred rather than directly observed. What makes itempirical is the fact that it is inferred from the observable behavior of the agent inquestion and functions as the ground of the explanation of this behavior.Appealing to a kind of via negativa, Kant introduces the problematic conception of an

intelligible character by contrasting it with the empirical variety. As purely intelligible,it cannot be inferred from experience and, by definition, it does not conform to theconditions of possible experience. And since time is the universal condition of possibleexperience, such a character “would not stand under any conditions of time” (KrVA539/B567), which means that we could not speak of something happening in or tothis character, or of its being determined by antecedent conditions. Indeed, its soleconceivable function is to serve as the intelligible ground of the empirical character,and, as such, it is only accessible (if it can regarded as accessible at all) in and throughthat character. Accordingly, it stands in a complex relationship with the empiricalcharacter: serving as its transcendental ground or cause (its ratio essendi), while the latteris its sensible sign or indicator (its ratio cognoscendi).6

Although this conception of intelligible character remains completely opaque andproblematic, since there is no way to determine whether an entity possessing it is evenpossible, its mere conceivability allows Kant to suggest that if, in fact, there were some

5 For my more detailed treatment of Kant’s distinction between empirical and intelligible character seeAllison (1990), 29–53. It is also briefly discussed in Allison (2002), which is contained in the present volume.

6 For a useful development of this point, see Willaschek (1992), 118–42.

24 PART I

being in the sensible world to which such a character could be ascribed, then itscausality (and the being itself) would have to be viewed from two points of view. AsKant puts it, “Of the faculty of such a subject we would accordingly form an empiricaland at the same time an intellectual concept its causality, both of which apply to oneand the same effect” (KRV A538/B566). And, of course, Kant’s aim is to suggest thathumans, as rational agents, are precisely such beings.In order to understand this suggestion, we must first note that Kant regarded it as

unproblematic that the human will or power of choice [Willkür] has, like everything elsein nature, an empirical character (KrV A546/B574), which he describes as “nothing otherthan a certain causality of his reason, insofar as in its effects in appearance this reasonexhibits a rule, in accordance with which one could derive the rational grounds and theactions themselves according to their kind and degree, and estimate the subjectiveprinciples of his power of choice” (KrV A549/B577). Although it may appear strangeto find the will in its empirical character described as a causality of reason, Kant’s point isthat, even at the empirical or phenomenal level, the intentional actions of human beingsexhibit a character that is distinct from that of merely physical or mechanical causes,namely they reflect a set of underlying intentions. These intentions constitute the“subjective principles of his power of choice”; they are empirical insofar as they can beinferred from overt behavior and used to explain past actions and to predict future ones.The notion of an empirical character therefore involves an explicitly deterministic

(although not reductionistic) picture of human agency. Moreover, Kant notes that it isthis conception of agency to which one appeals when, in the manner of anthropology,one is observing an agent’s behavior with the aim of “investigating the motive causes ofhis actions physiologically” (KrV A550/B578). In other words, the conception of anempirical character functions as a presupposition of the causal explanation of humanactions. To explain such an action is just to show that it follows from the character ofthe agent, together with the relevant set of background conditions.What is crucial for Kant, however, is that this is not the only point of view from

which we can consider human actions. As he proceeds to point out, they can also beconsidered normatively in relation to practical reason, that is, “insofar as reason is thecause of producing them by themselves” (KrV A550/B578). So considered, actions areevaluated in terms of rational norms which proclaim what ought or ought not to haveoccurred, quite independently of what did in fact occur. Although the issue iscontroversial, it seems reasonable to take Kant to be making a general claim aboutrational norms, which include but are not limited to moral norms. Thus, Kant statesthat reason imposes imperatives in “everything practical” [in allem Praktischen] (A547/B575), which presumably includes prudential as well as moral matters, and that “theought that reason pronounces” is operative “whether [what is willed] is an object ofmere sensibility (the agreeable) or even of pure reason (the good).” In either case, Kantsuggests, “[R]eason does not give in to those grounds which are empirically given, andit does not follow the order of things as they are presented in intuition, but withcomplete spontaneity it makes its own order according to ideas” (KrV A548/B576).


So construed, the spontaneity of reason provides the key to both the Kantianconception of intelligible character and its relation to the empirical character of thewill of rational agents. Kant’s view is that both understanding and reason are spontane-ous in the sense that they express the epistemic activity of the subject in organizing andunifying the data passively received through sensibility. But considered merely ascognitive faculties they do not exercise a causal power in the proper sense and thereforecannot constitute a character. Such power can, however, be attributed to reasonconsidered as practical, that is, insofar as it prescribes oughts regarding action. As alreadyindicated, Kant’s position is that to take oneself as acting on the basis of norms (oughts) isto attribute to oneself a power to act that is not determined by antecedent conditions(including one’s own antecedent state), which means that insofar as we deliberate aboutwhat we ought to do, we, implicitly at least, attribute such a power to ourselves. To besure, we may be deceived in so doing; but this would mean that we are deceived withregard to our rational agency, since the attribution of such a power is inseparable fromthe conception of ourselves as agents governed by rational norms.Accordingly, to consider the intelligible character of reason as the ground of the

empirical character of the will is to regard it as the source of its subjective principles, themaxims on which the agent acts; and this is also the basis for the imputation to rationalagents of actions based on these principles. The underlying idea is that the principles onwhich an agent acts are themselves freely chosen. In other words, rational agents do notsimply havemaxims, like animals (or chemical substances) have dispositions to behave incertain ways under given conditions. Rather, they choose maxims for themselves, withthis choice being grounded in rational norms and attributed to the spontaneity of thesubject.7 Correlatively, to say that the empirical character serves as a sign or indication ofthe intelligible character is just to say that the subjective principles on which a rationalagent acts reflect (albeit ambiguously) that agent’s fundamental choices and values.Although for the most part Kant’s account is first-person based, since it concerns

how we are rationally constrained to conceive ourselves, insofar as we take ourselves asagents subject to rational norms, its connection with imputation makes it applicable tothird-person accounts as well. Indeed, Kant attempts to clarify the third person,imputational side of the story by means of the notorious example of a malicious lie,which is intended “clarify the regulative principle of reason [an unconditioned,spontaneous causality] through an example of its empirical use” (KrV A554/B582).Faced with such an action, Kant suggests that we first inquire into its motive causes

[Bewegursachen] and then seek to determine the degree to which the action and itsconsequences may be imputed to the agent. In considering the former question, whichconcerns the empirical character of the action and the agent, we appeal to explanatoryfactors such as “a bad upbringing, bad company . . . the wickedness of a natural temperinsensitive to shame . . . carelessness and thoughtlessness,” as well as other “occasioning

7 In Kant’s later writings, the rational norms are attributed toWille and the spontaneity toWillkür. I discussthis important distinction in Allison (1990) 129–36 and (2006b), which is included in this volume.

26 PART I

causes” (KrV A554/B582). In short, it is assumed that the action can be fully explainedin terms of a combination of environmental factors and character traits; but, despitethis, we still blame the agent. Moreover, we do so not because of his bad disposition oreven his previous way of life, but rather because “one presupposes that it can beentirely set aside how that life was constituted, and that the series of conditions thattranspired might not have been, but rather that this deed could be regarded as entirelyunconditioned in regard to the previous state, as though with that act the agent hadstarted a series of consequences entirely from himself” (KrV A555/B583).The clear implication of this extremely controversial passage is that when we impute

an action to an agent we consider it as if it were a new beginning. We need not,however, take Kant to be making the totally implausible claim that in imputing anaction we leave out of consideration the agent’s past history and character, as if thesewere completely irrelevant. The point is rather that we do not view these factors as inthemselves causally sufficient to account for the action, since that would rule out thepossibility that the agent could have acted otherwise. In the terms that Kant uses in hispreliminary discussion, we assume that the set of empirical causes appealed to in theexplanation of the action, “was . . . not so determining that there is not a causality in ourpower of choice such that, independently of those natural causes and even opposed totheir power and influence, it might produce something determined in the temporalorder in accord with empirical laws, and hence begin a series of appearances entirelyfrom itself” (KrV A534/B562).In a set of important Reflexionen that are closely related to this discussion of freedom

Kant makes essentially the same point by stating that actions are to a large extent“induced [veranlasst] but not entirely determined by sensibility; for reason must providea complement of sufficiency [ein complement der zulänglichkeit]” (R 5611 18: 252); and inthe same context he speaks of reason as using but not being determined by the naturalcondition of the subject (R 5612 18: 253). In these, and similar passages Kant isclaiming that we necessarily appeal to the idea of a spontaneous, unconditionedcausality as a model or regulative principle in the contexts of both rational deliberationand imputation; for unless we do so, we cannot subject these actions to rational norms(whether moral or prudential).8

Kant does not infer from this analysis that we can know that we are spontaneousagents, possessed of an intelligible character and capable of initiating new series. It israther that we are required to attribute such a character to our agency insofar as we takeourselves as rational agents. Accordingly, Kant casts the entire analysis in an explicitlyhypothetical form, indicating that the claims about intelligible causality apply only ifreason has “causality in regard to appearances” (KrV A557/B582).9 As already noted,to deny that reason has such causality is to deny that we are genuine rational agents orself-determiners, even though we take ourselves to be such, and nothing that Kant says

8 For a discussion of these Reflexionen see Heimsoeth (1967).9 On this point see also KrV A545/B573, A547/B575, and A548/B576.


here undermines that possibility. This is precisely why he remarks at the end of analysisthat his intention has not been to establish the reality or even the (real) possibility offreedom, but merely to show that the alleged antinomy between freedom and causaldeterminism is illusory, since “nature at least does not conflict with causality throughfreedom” (KrV 558/B586).10

(3.4) Whereas the third cosmological conflict concerns the regress from conditioned tounconditioned causality, the fourth deals with the move from the conditioned orcontingent to the unconditioned or necessary existence of substance. The dispute isthus conducted in terms of the modal contrast: necessity–contingency, rather than thecausal relation, and it concerns the ground of the existence of substance rather than ofits states. At issue is whether it is legitimate to posit a necessarily existing being(substance) as the ground of the contingent beings (substances) in the world in orderto provide the unconditioned that reason seeks or, alternatively, whether the conceptof such a being is to be rejected on the grounds that everything encounterable inexperience has merely a contingent existence.As was the case in the third conflict, Kant maintains that if appearances were things

in themselves, or, alternatively, if everything were appearance, the thesis position couldnot be maintained, that is, there could be no necessarily existing being, just as therecould be no free causality, since everything encounterable in space and time isconditioned with respect to its existence as well as its causality (KrV A559–60/B587–8). And he likewise again insists that, given the transcendental distinctionbetween things as they appear and as they are in themselves, and the dynamical natureof the series of existing substances, it is possible to maintain that both thesis andantithesis may be true. Moreover, this makes it possible to grant a regulative functionwithin experience for the principle (affirmed in the antithesis) that everything in thesensible world has a merely conditioned existence, without foreclosing the possibilityof positing (in accordance with the thesis) a necessarily existing being that serves ascondition of the series as a whole (KrV A560/B588).Kant is once again careful to point out that this does not suffice to establish the

actuality or even the real possibility of such a being, but merely the compatibility of itsassumption with the conditioned nature of everything in the sensible world. Thus, theregulative principle that everything in the sensible world has an empiricallyconditioned existence is preserved (KrV A561/B589), while a conceptual space is

10 For Kant the reality of freedom can be established only from the practical point of view as a necessarypresupposition of our agency. Moreover, this is precisely what he attempted to do in his writings on moralphilosophy, beginning with the well-known claim that “to every rational being possessed of a will we mustalso lend the idea of freedom as the only one under which he can act” (GMS 4: 448). This makes theassumption of freedom practically necessary, but leaves in place the epistemic possibility that we are not reallyagents. For my discussion of this issue, see Allison (1990), 217–24, 245–9, and (1997a), (2000b), and(forthcoming), which are included in this volume.

28 PART I

opened for the “optional [willkürlichen] presupposition of a necessary, even thoughpurely intelligible condition” (KrV A562/590).Up to this point, Kant’s discussion of the fourth conflict closely parallels that of the

third. But there are at least two significant respects in which his treatment of the latterdiffers from that of the former. The first is the decidedly more critical stance that Kanttakes toward the idea of a necessarily existing being than toward that of a causality offreedom. In the latter case, as we have seen, he provided an elaborate analysis of theineliminable regulative function of the idea of freedom in the conception of ourselvesas rational agents. By contrast, he now remarks that an absolutely necessary being, asproduct of the understanding [Verstandeswesen], may in itself be impossible, eventhough this impossibility cannot be established in the manner of the antithesis, byreasoning from the contingency and dependence of everything in the sensible world(KrV A562–3/B590–1). Moreover, in the “Ideal of Pure Reason,” Kant does reject asincoherent the idea of a necessarily existing being on the grounds that any particularbeing, including the ens realissimum, can be thought without contradiction as non-existent (KrV A613–16/B641–4).A second major difference between the treatment of the third and fourth conflicts,

which Kant himself emphasizes, concerns the ontological location of the entities towhich these ideas supposedly refer. In the case of unconditioned causality, the causalagent was itself a member of the sensible world and only its causality was thought as“intelligible.” In the case of necessary existence, however, Kant suggests that, “thenecessary being must be thought as entirely outside the series of the world of sense (asan ens extramundanum), and merely intelligible” (KrV A561/B589). In other words, inthe one case an intelligible character is assigned to something with a phenomenal side (ahuman being), while in the other a distinct intelligible existence is assigned to atranscendent being.Although this formulation of the difference reflects the account of freedom just

considered, where Kant is concerned with the relevance of the transcendental idea forthe thought of human agency, it misrepresents the discussion in the actual antinomy.For there, it will be recalled, Kant was initially concerned with transcendental freedomas a cosmological idea, which is thought necessary to ground the totality of appear-ances. And in this context the idea of an innerworldly or immanent free causality wasintroduced largely as an afterthought. Nevertheless, there remains some justification forKant’s distinction between the two dynamical ideas, since the idea of a necessarilyexisting being is certainly distinct from that of an unconditioned causality and theformer (but not the latter) can be thought only as a transcendent being.11

11 As Kant points out, removing this being from the sensible world is necessary to prevent it “from beingsubjected to the law of the contingency and dependence of all appearances” (KrV A561/B589). The thesis ofthe fourth antinomy argues, however, that “To the world there belongs something that, either as a part of itor as its cause is an absolutely necessary being” (A452/BB480). Accordingly, Kant apparently accepted theclaim of the antithesis that the necessary being cannot pertain to the sensible world. For an insightfuldiscussion of some of the complexities and peculiarities of the fourth antinomy see Grier (2001), 218–29.


Kant further develops this latter point in a final paragraph characterized as a “Conclud-ing remark to the entire antinomy of pure reason,” which serves as a transition from thecosmological concerns of the latter to the properly theological concerns of the Ideal.Recalling the important distinction between the transcendental and the transcendentdrawn early in the Dialectic (KrV A296/B352), Kant remarks that the cosmologicaltranscendental ideas, which concern the unconditioned in the sensible world, becometranscendent when this unconditioned is posited outside the sensible world. At that pointreason ceases to be concerned with the unconditioned or closure with respect to thephenomenal as such, and is instead concerned with the intelligible ground of merelyintelligible objects, that is, objects thought through pure concepts in complete indepen-dence of the conditions of sensibility. Moreover, Kant insists that the cosmological ideathat gives rise to the fourth antinomy compels us to take this step because of thetranscendent nature of the intelligible object in which contingency terminates (KrVA566/B594). Accordingly, reason is led to the thought of an absolutely necessarybeing, fromwhich it then attempts to derive the concept of contingent things consideredas merely intelligible objects. And with this thought Kant effects the transition to the nextchapter of the Dialectic: “The Ideal of Pure Reason.”

30 PART I

Essay Two

Where Have all the CategoriesGone? Reflections onLonguenesse’s Reading of Kant’sTranscendental Deduction

With Kant and the Capacity to Judge, Beatrice Longuenesse has written a bold, impor-tant, and exciting book concerning the major arguments of the TranscendentalAnalytic. Moreover, the entire work is organized around a central thesis that runsdirectly counter to most contemporary readings of the Critique, namely that,

[N]either the argument of the Transcendental Deduction of the Categories . . . nor the System ofPrinciples of the Pure Understanding, can be understood unless they are related, down to theminutest details of their proofs, to the role that Kant assigns to the logical forms of our judgments,and to the manner in which he establishes the table of categories or pure concepts of theunderstanding according to the ‘guiding thread’ of these logical forms.1

In almost 400 pages of careful analysis, which reveal a mastery of the Kantian texts, theirimmediate background in the rationalist Schulphilosophie, and the vast secondary litera-ture, Longuenesse proceeds to demonstrate this claim; indeed, “down to the minutestdetails.” Seldom has a promissory note in an introduction been so fully repaid in thebody of the work. Thus, despite the inevitable disagreement about some of these details,not all of which are minute, I think it fair to say that from now on no serious Kantinterpreter will be able to ignore either the “guiding thread” itself or her analysis of it.What makes the “guiding thread” so central to Kant’s critical project on Long-

uenesse’s reading is the foundational role given to the forms of judgment. Indeed, forLonguenesse, these forms usurp the objectifying function usually assigned to thecategories. As she expresses the matter in her introduction, the basic claim is thatKant’s thesis concerning the necessity of conformity of appearances to the categories“can be understood correctly only in light of the objectifying role that Kant grants tothe logical forms of judgment as forms of reflection.”2 Or, as she later puts it, “If we disregard

1 Longuenesse (1998), 5. 2 Ibid., 12.

the priority of the logical forms over the categories, Kant’s whole argument is renderedincomprehensible.”3

As Longuenesse herself points out, this emphasis on the forms of judgment, particu-larly when combined with a focus on the distinct synthesis speciosa, leads to a privilegingof the B-Deduction. To be sure, she does not neglect the A-Deduction completely,since she provides a rich and subtle analysis of the three-fold synthesis. Nevertheless,because of its virtually complete neglect of the logical forms, she limits its significance tothat of a via negativa, which by ruling out anything like a Humean “empiricopsycho-logical genesis” of perceptions, prepares the way for the positive and definitive accountprovided in the B-Deduction.4 In fact, she goes so far as to suggest that the second andthird parts of her study can be read as an extended commentary on essential sections ofthis Deduction (}19 and }20 corresponding to the former and }24 and }26 to the latter).5

Moreover, in surveying the recent literature on the B-Deduction, she does me thesignal honor of suggesting that my interpretation is the closest to hers.6 Naturally, theagreement is not complete. In particular, she suggests that we differ in our views ofhow Kant relates the categories to the forms of judgment and how each of these relateto the sensible synthesis. And on this basis she proceeds to criticize some of my claimsregarding the structure of the argument. Accordingly, it seems appropriate for me toreturn the honor by focusing on her views on these two issues and their bearing on thestructure of the B-Deduction. Unfortunately, this makes it necessary to neglect some ofthe more interesting and important features of her book, in particular her remarksabout the connections between the first and third Critiques. But inasmuch as anyattempt to do justice to that issue would require a full-length treatment in its own right,it must be reserved for a future occasion.

IEven though she emphasizes the importance of Kant’s characterization of the cate-gories as “concepts of an object in general, by means of which the intuition of an objectis regarded as determined in respect of one of the logical functions of judgment” (KrVB128), Longuenesse denies that this commits Kant to the view that the empirical use ofthe logical functions requires the subsumption of appearances under the categories. Infact, this is the basic lesson she derives from the distinction between judgments ofperception and judgments of experience, which, in contrast to most interpreters, sheregards as an essential ingredient of Kant’s position rather than as a merely preparatorydevice adopted mainly for expository purposes in the Prolegomena.7 Thus, judgments ofperception, on her view, are genuine judgments involving the logical functions ratherthan mere associations (though they are based on merely associative connections) and,as such, they conform to the conditions of the objective unity of apperception as

3 Longuenesse (1998), 78. 4 Ibid., 9.5 Ibid., 9, 72, 80. 6 Ibid., 60 note 3. 7 Ibid., 194.

32 PART I

characterized in the B-Deduction. Nevertheless, they do not involve an application ofthe categories, at least not the “full fledged” categories or, as she also puts it, a full-fledged application of them. The latter occurs only in judgments of experience, whichalone are fully objective.8

As the above indicates, Longuenesse’s account of the role of the categories is highlycomplex and nuanced. Basically, her claim is that they operate at both ends of thecognitive process. As fully reflected concepts, they come into play only in judgments ofexperience, which explicitly apply them to a sensible content that has already beenreflected upon according to the logical forms.9 But they also function at the initial stage(and therefore presumably in judgments of perception) under the guise of the logicalforms. At this stage they guide reflection by generating the schemata that will later bereflected as concepts in the fully objective judgments of experience. And it is only inthis somewhat Pickwickean sense that Longuenesse is willing to grant the thesis that“every empirical judgment—and, similarly every synthesis of the sensible given inorder to form judgments—is an ‘application’ of the categories.”10 This generation doesnot, however, occur through judgment per se, but rather through the transcendentalsynthesis of the imagination, otherwise known as the synthesis speciosa. Accordingly, sheinterprets this synthesis in strongly teleological terms, as being guided by our capacityto judge [Vermogen zu urteilen], to the end of realizing our power of judgment[Urteilskraft].11

Given the impossibility of discussing here all of the facets and ramifications of thishighly original and suggestive account of the nature and functions of the categories,I shall focus on its implications for the reading of the B-Deduction. The first, withwhich I shall be concerned in the remainder of this section, involves the connection ofthe categories, so construed, with the intellectual synthesis, which is the form ofsynthesis under consideration in the first part of the Deduction. And here I see aproblem. For by denying that an empirical judgment involving the logical functions,that is to say, any empirical judgment, necessarily involves a use of the categories,Longuenesse seems to undermine the conclusion of the first part of the Deduction. Orso I shall argue.In sharp contrast to the A-Deduction, which begins with an explicit focus on the

temporal nature of human experience, the first part of the B-Deduction abstracts fromthe particular forms of human sensibility. It is thus concerned with the categories (notmerely the logical forms of judgment) as rules for a discursive understanding in general,that is, one which is dependent on sensibility for the data to be unified, but is notfurther constrained by any particular forms of this sensibility. For this reason Kant

8 Ibid., 170–80.9 Longuenesse seems to regard this as a consequence of Kant’s anti-innatism and the doctrine of the

priority of judgments over concepts, which is closely connected therewith. See Longuenesse (1998),199–200, 252–3.

10 Ibid., 199.11 Ibid., 8.

WHERE HAVE ALL THE CATEGORIES GONE? 33

characterizes the synthesis performed by an understanding, so conceived, as purelyintellectual. Correlatively, the categories are claimed to relate through this synthesis to“objects of intuition in general” (KrV B150).The argument of the first part is intended to establish this result, which I take to be

equivalent to demonstrating the objective validity of the categories, that is, theirvalidity with respect to objects considered as correlates of judgmental acts or, equiva-lently, “objects of intuition in general.” Moreover, I agree with Longuenesse that thepivot of this argument is }19, where Kant appeals to his conception of judgment as thesynthetic activity (intellectual synthesis) through which given representations arebrought to the objective unity of apperception, and therefore (by }17) also broughtinto relation to an object. We likewise agree that the objective validity, which Kanthere attributes to judgment as such, is not to be equated with truth (since that wouldcommit Kant to the absurdity that every judgment as such is true). It is rather that everyjudgment makes a claim to truth or has a truth-value, which is enough to distinguishjudgment from a merely imaginative association of the same representations.As already indicated, however, Longuenesse denies that judgment, so conceived,

involves any use of the categories. In fact, this is precisely the basis on which she arguesfor the compatibility of the account in }19 with the distinction drawn in the Prolego-mena between judgments of perception and judgments of experience. Once again,categories (in the full-fledged sense) come into the picture only with the latter and allthat necessarily pertains to judgment as such are the logical forms. But if this is Kant’sposition, it is difficult to understand the conclusion drawn in }20. For after noting in acapsule summary of the argument up to this point that the manifold, “so far as it is givenin a single [Einer] empirical intuition, is determined in respect of one of the logicalfunctions of judgment . . . ,” he proceeds to claim:

Now the categories just are these logical functions of judgment, insofar as they are employed indetermination of the manifold of a given intuition. Consequently, the manifold in a givenintuition is necessarily subject to the categories. (KrV B143)

Longuenesse’s reading apparently has no place for this last move, and therefore no basisfor understanding Kant’s subsequent claim that with it a “beginning is made of adeduction of the pure concepts of understanding” [Verstandesbegriffe] (KrV B144).Given her reading, it seems that Kant should have contented himself with pointingout that the unity of apperception is brought about in accordance with the logicalforms of judgment, and not gone on to add that “the unity of consciousness of themanifold in a single intuition is effected by means of the category” (KrV B144). Indeed,Kant’s preliminary conclusions in }20 and }21 appear to run directly counter to herpreviously cited denial that an empirical use of these forms, which presumably wouldbe one with respect to intuition, necessarily involves the categories.In response to this line of objection, Longuenesse might reply that she has denied

only the application of the full-fledged categories as reflected concepts and that she hasexplicitly left room for that other, Pickwickean “application,” in which, in the guise of

34 PART I

the logical forms, they guide the formation of schemata. Such a response, however,involves two difficulties. First, it leaves no room for a distinction between the logicalforms and the categories at the purely conceptual or judgmental level.12 Consequently,it cannot explain the inference that Kant makes from the former to the latter at the endof }20, an inference that marks the culmination of the first part of the Deduction.Second, and more important, appeal cannot be made to their schematizing function atthis point because the focus of discussion in the first part of the Deduction is on a purelyintellectual synthesis, which involves no reference to the forms of human sensibility. Inother words, what seems to be required by the argument is something that Long-uenesse (at least as I understand her) cannot allow, namely, a role for the categories in apurely intellectual synthesis such as Kant describes in the first part of the B-Deduction.For their function as rules for this synthesis does not appear to correspond to either ofthe two roles that she assigns to them.It was largely in order to account for such a role that I introduced the distinctions,

which Longuenesse finds objectionable, between objective validity and objectivereality, on the one hand, and Objekt and Gegenstand, on the other.13 These wereintended as expository devices to be used for distinguishing between the kinds of claimsmade for the categories in the two parts of the Deduction and therefore for clarifyingthe progressive nature of Kant’s argument as a two-step proof. Although I still findthese distinctions useful for that purpose, I shall not insist upon them here.14 What I dowish to insist upon, however, is the necessity of finding a place for the categories (asdistinct from the logical forms) as rules for a purely intellectual synthesis or, equiva-lently, for the unification of the manifold of an intuition in general. The point is notthat there is such a manifold, since our sensibility provides us with a spatiotemporalmanifold, not a manifold of intuition in general. Nevertheless, it seems to me that it isessential for the comprehension of the categories as rules for a discursive understanding,which expresses itself through the various logical forms of judgment, that their functionbe characterizable at this level of abstraction, that is, as pure, rather than as schematized,concepts. And it also seems to me that this is precisely what Kant was trying to do in thefirst part of the B-Deduction by connecting them with a purely intellectual synthesis.It appears from some remarks in her book that the reason Longuenesse rejects this

line of analysis is that she does not really view the two parts of the B-Deduction as

12 Elsewhere, I have tried to account for such a role by suggesting that the pure concepts be understood asconcepts that are presupposed by the activity of judgment. Very roughly, the idea was not that judgingcategorically and hypothetically requires an application of what Longuenesse terms the “full-fledged”categories of substance and causality (here we are in complete agreement), but rather that judging in thismanner does involve certain forms of conceptualization and these are what is thought in the correspondingpure concepts. In short, the forms of judging are at the same time forms of conceptualization, which explainswhy these logical functions, considered as forms of the unification or determination of a manifold of sensibleintuition, are equated with the categories. For my most recent treatment of this issue, see Allison (2004), esp.146–56.

13 See Allison (1983), 133–72.14 For a partial recantation, see my addendum to this essay.


constituting a progressive two-step proof. Instead, she regards the second part, dealingwith the synthesis speciosa, as accounting for the judgmental activity sketched in the first.For example, in turning from the first to the second synthesis, she remarks, that, “It isnow time to consider what makes those intellectual syntheses, or discursive judgments,possible in the first place. And for this we must retreat [my emphasis] from the synthesisintellectualis to the synthesis speciosa, or the transcendental synthesis of the imagina-tion.”15 Moreover, shortly thereafter she characterizes the intellectual synthesis as thegoal to be reached by the imaginative one.16 In fact, on occasion, she even seems toidentify the intellectual synthesis of the first part with analysis, which would of courseexplain why this “synthesis” could be viewed as both presupposing the imaginativesynthesis and as being the goal at which it is directed.17

Despite the many interesting and important things that Longuenesse has to sayabout the relation between analysis and synthesis, I have great difficulty acceptingher apparent attempt to use this relation to explicate the connection between thetwo parts of the B-Deduction. To begin with the obvious, I fail to see how a synthesis,even a purely intellectual one, can be equated with analysis; or if it is, how Kant couldclaim in }21 that even a beginning has been made in the deduction of thepure concepts.18 To my mind, Kant’s language, as well as the entire discussion in}21, strongly suggest that the move from the intellectual to the imaginative synthesisshould be taken as an advance from a general consideration of the judgmental functionof the categories (as rules for the synthesis of a sensory manifold in general) to theirspecification under the peculiar conditions of human sensibility rather than as a“retreat” or regress from something conditioned (analysis) to its condition (imaginativesynthesis).For these reasons, then, I remain puzzled about what role, if any, Longuenesse is

willing to assign to the categories (as distinct from the logical functions) in the first partof the B-Deduction. Clearly, it cannot be their “silent” role as generators of theschemata that are intended eventually to be reflected in judgments of experienceunder the categories in the full-fledged sense; for this is the concern of the secondpart. Equally clearly, however, it cannot be as full-fledged categories either, since theypresuppose these very schemata. But according to her account, these exhaust thepossible roles for the categories. Consequently, the move that Kant makes in }20from the logical forms to the categories and which he characterizes in }21 as a“beginning” in their deduction remains unexplained.

15 Longuenesse (1998), 197.16 Ibid., 203.17 Ibid., 111 note 14.18 I have argued elsewhere that the first part of the B-Deduction is to be seen as an attempt to derive the

analytic implications of the analytic principle of the synthetic unity of apperception. See Allison (1996a)41–52 and (2004), 163–78. But this is quite different from the claim that the intellectual synthesis is itself to beunderstood as an act of analysis.

36 PART I

IILonguenesse and I agree in attributing decisive significance to the second part of the B-Deduction. Accordingly, we likewise agree that the argument is not to be viewed asconsisting in a relatively trivial inference from what holds of the genus (sensibleintuition in general) to the species (human sensibility and its contents).19 As alreadynoted, however, she herself points out that we differ in our understanding of how thecategories supposedly relate to human sensibility and its a priori forms.I take it as non-controversial that the second part consists of essentially two steps: the

first relates the categories to the forms of human sensibility or pure intuition byconnecting both with the transcendental synthesis of the imagination (}24); the secondrelates them to empirical intuition (}26). I likewise take it as non-controversial that theformer move is the decisive one, since the latter follows from it straightforwardly.Accordingly, the key issues are how the former move is to be interpreted andevaluated, that is, how we are to understand the connection between the forms ofsensibility, the synthesis speciosa, and the categories.I have argued that in }24 Kant is to be read as trying to establish that the latter

synthesis must, like its intellectual counterpart, be governed by the categories on thegrounds that it is necessary for the representation of the unity of time, and that therepresentation of this unity must conform to the categories as conditions of the unity ofapperception.20 Thus, rather than being an analytic consequence, which would be thecase if the unity of consciousness itself entailed the unity of time, the latter is aconstraint on human apperception that stems from the forms of sensibility.Although Longuenesse likewise emphasizes the centrality of the synthesis speciosa, she

contends that her position is more radical than mine, and, indeed, virtually all inter-preters. This radicality claim is two-pronged. First, instead of seeing the second part ofthe Deduction as concerned with relating the categories (via the synthesis speciosa) tospace and time as given intuitions, she takes Kant’s position to be that space and timeare themselves products of the imaginative synthesis. Thus, on her reading, the majormove in the second part of the Deduction is a revision of the results of the Transcen-dental Aesthetic, by means of which what was formerly seen as a contribution ofsensibility (the pure intuitions of space and time) is now regarded as the outcome of thesynthesis that Kant describes as an “effect [Wirkung] of the understanding on sensibility”(KrV B152).21

Longuenesse’s second radical thesis is that this effect does not involve the categories,at least not directly. Rather than identifying the “understanding” with the pureconcepts of the understanding, she equates it with the capacity to judge. The claim isthat it is this “capacity” that somehow affects sensibility, thereby generating “the pureintuitions of space and time as the necessary intuitive counterpart to our discursive

19 See Longuenesse (1998), 213; Allison (1996a), 35.20 See Allison (2004), 186–93. 21 Longuenesse (1998), 13.


capacity to reflect universal concepts, concepts whose extension (the multiplicities ofsingular objects thought under them) is potentially unlimited.”22 I shall examine eachof these claims in turn.The first is based on her analysis of a key portion of }26 and the much discussed

footnote attached thereto. In the main text Kant refers back to the Aesthetic, remark-ing that space and time are “represented a priori not merely as forms of sensibleintuition, but as themselves intuitions which contain a manifold of their own” (KrVB160). Kant’s contention is that the representation of these manifolds, like that of anyother, rests upon a unifying synthesis, which ultimately must accord with the cate-gories. He expands upon this in the note by distinguishing between a form of intuitionand a formal intuition, with the latter presupposing a “synthesis which does not belong tothe senses but through which all concepts of space and time first become possible”(KrV B161n).Although this note has been interpreted in a wide variety of ways, the generally

accepted reading, with which I concur, is that the contrast between a form of intuitionand a formal intuition is to be understood as between an indeterminate and adeterminate sense of “pure intuition,” with “determination” being understood interms of the above-mentioned synthesis. By contrast, Longuenesse insists that therepresentations of space and time are not simply determined by the imaginativesynthesis, but are actually generated or produced by it.23 A corollary of this contention,which she is particularly concerned to defend, is that the “formal intuition” of the notecorresponds to both the “form of intuition” and the “pure intuition” of the Aesthetic,and therefore that the latter two are themselves essentially equivalent. Since Kantclearly states in the note that this formal intuition is the product of a synthetic activity, itfollows that the representations of space and time as pure intuitions and, more specifi-cally, as infinite magnitudes, which in the Aesthetic were characterized as simply“given,” reemerge as products of a transcendental synthesis. In short, the space andtime of the Aesthetic are now regarded as entia imaginaria.24

Longuenesse also criticizes the standard reading in light of her understanding ofKant’s use of “form.” She maintains that the term has meaning for Kant only in relationto “matter,” which refers to the determinable and “form” to its determination.25

Consequently, the notion of an “indeterminate form,” to which the reading sheopposes seems committed, becomes something of an oxymoron.

22 Longuenesse (1998), 224.23 Ibid., 216–24.24 In his table of the division of the concept of nothing appended to the Amphiboly chapter, Kant

characterizes “pure space” and “pure time” in this manner on the grounds that they are not themselvesobjects intuited, though they are forms of intuition (KrV A291/B347). Longuenesse appeals to this passage[(1998), 305] where she identifies being entia imaginaria with being products of the synthesis speciosa, butI think that this clearly misrepresents the sense in which Kant there intends the expression.

25 Longuenesse (1998), 223.

38 PART I

I find this questionable, however, since Kant clearly does contrast form withdeterminateness. Thus, in contrasting inner sense and apperception, he states that theformer contains the “mere form of intuition,” which he explicitly distinguishes from a“determinate intuition” (KrV B154). But setting that aside, Longuenesse’s alternativereading faces at least two significant exegetical problems of its own, though to hercredit she is well aware of them. First, since Kant explicitly contrasts “form of intui-tion” and “formal intuition”within the note itself, she must explain why Kant used theformer expression in conflicting ways in different parts of the Critique. Second, giventhe above-mentioned understanding of “form,” she must explain the presence of“form of intuition” within the note, since, as contrasted with “formal intuition,” itclearly cannot there mean something determinate.In responding to the first of these problems, Longuenesse accepts my analysis of

“form of intuition” in the note as meaning a form of intuiting, while insisting thatwithin the Aesthetic it is both a form of intuiting (though in a different sense than in thenote) and a form of the intuited. In an endeavor to illustrate this seemingly paradoxicalthesis, she appeals to the analogy of following with a stick the outline of a drawing onthe ground; the idea being that the form of the drawing and of the figure drawn are oneand the same.26 Finally, she concludes that these forms are equivalent in the Aestheticbecause “as ‘forms of sensibility’ (forms of intuiting), they are more than the ‘merecapacity to intuit’ [my expression]. They are the forms of a sensibility affected both by thething in itself and by spontaneity.”27

Although I find nothing inherently objectionable in the latter claims, I must admit tosome difficulty in understanding what they are supposed to show and how they relateto her analogy.28 Granted, forms of sensibility are forms of a sensible intuiting that isaffected by the thing in itself. Granted also that for Kant the same sensibility is affectedby the activity of the understanding (self-affection). But does it follow from this thatthe form of what is sensibly intuited is itself a product of the latter mode of affection?Why could not one just as easily claim that it was a product of the former, which wouldmake Kant into a transcendental realist with regard to space and time?Longuenesse’s analogy seems compatible with such a view, since she speaks of

following with a stick the outline of a drawing on the ground, which suggests thatthe form of the activity (tracing with the stick) is conditioned by the already producedsketch. But her claim about the generative function of the synthesis speciosa points to amore Fichtean picture, according to which our form or manner of sensibly intuiting is

26 Ibid., 218 note 12.27 Ibid.28 Since the above seems intended, at least in part, as a criticism of my views, I wish to note that I did not

suggest that forms of sensibility (qua forms of intuiting) are nothing but “the mere capacity to intuit.” WhatI actually said was that they may be characterized as innate capacities or dispositions to intuit things in a certainway, that is, spatiotemporally; and I referred in this context to the same passage from Kant’s response toEberhard to which Longuenesse appeals in her account of the distinction between a form of intuition and aformal intuition. See Allison (1983), 97 and 345 note 32.


conditioned by the form of what is intuited, while the latter is itself a product of thetranscendental synthesis and therefore an ens imaginarium. In that event, however, it isdifficult to understand what remains of Kant’s conception of sensibility as a distinctfaculty with an a priori form.Nevertheless, I do not think that Longuenesse intends her revisionary account of

space and time as a complete rejection of Kant’s theory of sensibility in the manner ofpost-Kantian idealism. In fact, the gist of her reply to this line of objection can begleaned from her response to the second of the above-mentioned exegetical difficultiesgenerated by her reading, namely, how to interpret “form” in the “form of sensibility”referred to in the note. In discussing this issue, Longuenesse appeals to what she terms“Kant’s evolutionary, ‘epigenetic’ conception of the conditions of representation.”29

The idea here is that, though the actual representations of space and time are productsof the synthesis speciosa, the conditions of the possibility of such a production must lie inthe nature of receptivity. In other words, the latter “must be constituted in such a way”that the sensory manifold “potentially ‘allows of being ordered’ in space and time.”30

And, in light of this, she suggests that “form of intuition” in the note is to beunderstood as referring to a mere “potential form” or potentiality for such form.Accordingly, it is this merely potential form that is actualized through the synthesis.By regarding sensibility as providing a “potential form,” which is then actualized by

the synthesis speciosa, Longuenesse’s reading preserves an essential role for sensibility andits a priori contribution to human cognition. The reason why the manifold of ourintuition has a spatiotemporal form is due to the nature of our sensibility after all, that is,to our “form” or manner of intuiting, rather than to the synthesis speciosa. AlthoughI welcome this result, I am also somewhat puzzled by it, since it seems to conflict withboth Longuenesse’s main critical point about “form” and her emphatic claims regard-ing the radicality and novelty of her interpretation. For it remains unclear to me why a“potential form” should be any less problematic than an “indeterminate” one, or evenwhere the difference between them is supposed to lie. Moreover, if, as I suspect, thereis no significant difference, then I question the radicality and novelty of her interpreta-tion of the connection between the synthesis speciosa and the forms of sensibility.There can, however, be little doubt regarding the radicality of the other main prong of

her reading of the second part of the B-Deduction, namely, the connection between thesynthesis speciosa and the categories. In essence, she argues that just as it is a mistake to supposethat “formof intuition” is opposed to “formal intuition” as indeterminate to determinate, so“it is also a mistake to suppose that ‘formal intuition’ is determined by concepts.”31

Although this supposedly mistaken reading is suggested by Kant’s characterization ofthe synthesis speciosa as an “effect of the understanding on sensibility,” Longuenesse has twoobjections against the inference from this to the conclusion that this synthesis is governedby concepts, and especially by the categories. The first consists in an appeal to Kant’s claim

29 Longuenesse (1998), 221. 30 Ibid. 31 Ibid., 224.

40 PART I

in the note that the formal intuition “precedes any concept.”32 The second is a reiterationof her central thesis that the categories, as full-fledged (reflected) concepts, only emerge atthe end of the intellectual process, and therefore can hardly be assumed to be operative atits initial stage. How, then, are we to construe the claim that the understanding affectssensibility? Her answer, which epitomizes her whole interpretation of the first Critique,turns on the previously noted identification of the understanding with das Vermogen zuurteilen. Accordingly, for Longuenesse it is this capacity that affects sensibility, therebygenerating the pure intuitions of space and time.33

I believe that the first objection can be dealt with fairly easily. What Kant actuallyclaims in the note is that “In the Aesthetic [my emphasis]” he treated the unity (of aformal intuition) “as belonging merely to sensibility, simply in order to emphasize thatit precedes any concept . . . ” (KrV B162). Since in the Aesthetic it could only be theconcepts of space and time and not the categories that are at issue, the natural reading isto take Kant as referring in the note merely to the former. Moreover, this is fully inaccord with the teaching of the Aesthetic itself, where the claim is that concepts ofspace and time presuppose the intuitions.34

The second issue is far more complex and I can here do little more than give voice tosome of my perplexity and reservations. To begin with, I very much like the idea thatthe action of the understanding in affecting sensibility is to be viewed as orientedtoward the goal of conceptualization and judgment. Where I begin to part companywith Longuenesse is her insistence that the understanding at this juncture is bestconstrued as the capacity to judge. Admittedly, Kant does characterize it in theseterms within the context of the “guiding thread.”35 But he also characterizes it inmany other ways as well in the Critique, including as a faculty of concepts, of rules, asapperception, and as spontaneity. Moreover, since it is the latter characterization of theunderstanding that Kant features in his discussion of its connection with inner sense(KrV B150, 158), it seems natural to focus on it in an account of this connection. Nowspontaneity is the defining property of the act of unification or synthesis, and we caneasily think of this act as both rule-governed and as determining the content of innersense. Accordingly, it is presumably such an act that Kant had in mind when hecharacterized the understanding as spontaneity. By contrast, I do not see how a mere“capacity to judge,” which Longuenesse also describes as a “possibility or potentialityfor forming judgments,” can do anything, including affect or determine inner sense.36

If this is to be made intelligible, it must be taken to mean that an act of understandingconsists in an exercise of this capacity, which I consider to be Kant’s actual position andto be fully compatible with a characterization of this act as one of spontaneity.

32 Ibid. 33 See note 24.34 See Allison (1983), 91–2 and (2004), 109–10.35 See KrV A69/B94 and A81/B106, and Longuenesse (1998), 7.36 Longuenesse (1998), 7.


The question thus becomes how we are to understand this originary act or exerciseof the capacity to judge, which Longuenesse repeatedly describes as the “first applica-tion of the categories.” Clearly, it cannot be an act of judgment proper, since thatpresupposes already reflected concepts and only occurs at the other end of thecognitive process. Her answer, as we have already seen, is that this act is “guided” bythe capacity to judge in the teleological sense that it is directed toward the eventual goalof fully objective judgments under the full-fledged categories. Once again, however,I wonder whether this Pickwickean sense in which this act supposedly involves thecategories is sufficient to capture Kant’s claim to have established their objective reality.For I think that any reasonable reading of the latter claim requires that the act not onlybe oriented toward the eventual formation of the categories, but in some sense alreadycategoreally determined. Longuenesse denies this for the reasons noted above, whichleaves her with the logical functions as the determining forms of unification. She is notbothered by this, however, since on her reading the categories are “as it were,engrained in the mind as logical functions.”37

But this again will not do, albeit for somewhat different reasons than was the case in thefirst part of theDeduction. There, it will be recalled, the problemwas that reducingKant’sclaim to one concerning the role of the logical functions could not account for theinference drawn from these functions to the categories. Now the problem is that it seemsas inappropriate to refer to the logical functions in connection with the synthesis speciosa asit does to the categories as full-fledged concepts. For as Longuenesse herself emphasizes,these functions are discursive forms of the combination of concepts for the purpose ofjudging; and at this stage there are as yet no concepts to be combined in judgment.Once again, however, I think that the situation can be clarified by returning to

Kant’s definition of the categories as “concepts of an object in general by means ofwhich the intuition of an object is regarded as determined in respect of one of thelogical functions of judgment” (KrV B128). As before, I think that the crucial point isthat this definition allows for a functional distinction between the categories and thelogical forms, while preserving their essential identity as forms of understanding. Giventhis functional distinction, we can say that the categories operate in the second part ofthe Deduction in the same way as they do in the first, namely, as forms of the synthesisof the sensible aiming at the unity of apperception. The only difference is that theynow receive a “transcendental content” or filling through their relation to the forms ofhuman sensibility. Accordingly, the synthesis speciosa does involve the categories in anon-Pickwickean sense as its governing rules. At least this is what the Deduction mustshow if it is to claim success. But for all her subtlety and originality, I do not think thatLonguenesse shows how this is possible.38

37 Longuenesse (1998), 144.38 I am here questioning Longuenesse’s claim that “if we accept Kant’s argument concerning the role

played by the logical functions of judgment and the synthesis speciosa, the question of right (quid juris) has beenresolved . . . ” (1998), 244.

42 PART I

A Response to a Response: AnAddendum to “Where Haveall the Categories Gone?”

Although the main purpose of this addendum is to discuss Beatrice Longuenesse’sresponse to my essay, I also wish to use this occasion to acknowledge the importance ofher work for both the revised version of my Kant’s Transcendental Idealism (2004) and,before that, Kant’s Theory of Taste (2001a). Beginning with the latter, at the time of thecomposition of this essay, I was hard at work on Kant’s Theory of Taste and, whilecontemplating a revised version of Kant’s Transcendental Idealism, I had not yet begunwork on that project.1 As a result, the initial impact on me of Longuenesse’s bookmainly concerned a topic that is somewhat peripheral to her central focus in this work,namely, Kant’s account of reflective judgment as found in the two Introductions to thethird Critique. In particular, I found her remark that what was new in the third Critiquewas not, as is usually thought to be the case, the notion of reflective judgment as such,but rather that of a “merely reflective judgment” to be deeply suggestive, because itplaced the conception of reflective judgment or, more precisely, the reflective powerof judgment, at the center rather than the periphery of Kant’s thought.2 In addition toKant’s Theory of Taste, the results of my efforts to work out this line of thought arecontained in several of the essays contained in this volume.3

Kant and the Capacity to Judgewas also an important influence on my efforts to rethinksome of my views for the revised version of Kant’s Transcendental Idealism (2004).Specifically, it helped me gain a greater appreciation of the importance of the Meta-physical Deduction and pointed the way to what I believe to be improvements in myaccount of the Transcendental Deduction. With regard to the latter, Longuenesse’scriticisms led me to abandon my thesis in the original version of Kant’s Transcendental

1 “Where Have all the Categories Gone?” was written in 1999 for delivery that year at an author meetscritic session of the American Philosophical Association meeting on “Kant and the Capacity to Judge” inNew Orleans.

2 Longuenesse (1998), 164.3 This applies particularly to Allison (2001b), (2003), and (2009a). Other essays in this collection that deal

with various aspects of purposiveness, namely, (1991a), (1996b), and (2000c) were written prior to myacquaintance with Longuenesse’s book.

Idealism (1983) that the distinction between Object and Gegenstand had systematicimport.4 According to this reading, the first part the B-Deduction is devoted to thedemonstration of the necessity of the categories forObjecte, understood as objects in thebroadest sense (“in sensu logico”), and the second to the demonstration of their necessityfor Gegenständen, understood as full-fledged phenomenal objects. Despite what seemslike textual evidence in support of it, I was led to reject this by Longuenesse, whopointed out that the object at issue in the first part of the Deduction is defined as theobject of intuition as such and is therefore an intuited object rather than merely an objectin the most general or logical sense.5 Thus following (at least to some extent) Long-uenesse’s suggestion, I now take the tasks of the two parts of theDeduction as concernedrespectively with the demonstration of the necessity of the categories for objects ofsensible intuition in general, that is, for the thought of such objects, considered indepen-dently of the manner of their intuition, and the demonstration of their necessity for theperception and experience of phenomena.6 While I do not consider this a fundamentalrevision of my original view, I do take it as an improvement, inasmuch as abandons theextremely vague and potentially misleading notion of an object in sensu logico.Having underscored the importance of Longuenesse’s book for my own work,

I shall devote the remainder of this addendum to her published response to my critiqueof her reading of the B-Deduction.7 At the heart of her response, both to myself and tomy co-commentator (Sally Sedgwick), is a reiteration of her core thesis that thecapacity to judge [Vermogen zu urteilen] is definitional of the understanding for Kant,since it encompasses the functions of conceptualization, judgment as subsumption, andeven syllogistic inference.8 Apart from my continued puzzlement about how a capacity(as contrasted with its exercise) can do anything, including determine sensibility,I now find myself in basic agreement with Longuenesse, when she suggests that thischaracterization is privileged over the other depictions of the understanding that Kantprovides, because refers to its most fundamental activity.9 Of itself, however, thisrelocates rather than resolves the problem of the relation between the categories andthe understanding; for now the question becomes their relation to this capacity tojudge. As in her book, Longuenesse maintains that they are operative at both ends ofthe epistemic process, namely, the synthesis that prepares the sensible manifold forconceptualization and the subsumption of the conceptualized data under a category.10

Again I agree, but I question how these two functions of the categories relate to thetwo parts of the B-Deduction.

4 As is clear frommy initial response to Longuenesse, I still held such a view in 1999. Although I continueto believe that the complementary distinction between objective validity and objective reality is applicable tothe two parts of the Deduction, I did not insist upon it in the revised version of Kant’s Transcendental Idealism.

5 Longuenesse (1998), 70 note 17 and 110–11 note 14.6 See Allison (2004), 162 and 476 note 11.7 Longuenesse (2000), esp. 100–6.8 Ibid., 92–5.9 Ibid., 92–3.10 Ibid., 96.

44 PART I

Although it is tempting to suggest that these two functions are neatly correlated withthe two parts of the Deduction, this turns out not to be the case; and, if I understandher correctly, Longuenesse does not claim that it is. Rather, she holds that in the firstpart the categories are connected with the intellectual synthesis of a manifold ofintuition as such and in the second with the synthesis speciosa or figurative synthesis ofthe imagination; so that the subsumptive use of the categories is not properly at issue inthe Deduction at all, but is instead the concern of the Analytic of Principles. At thislevel of description we are still in basic agreement; but significant differences emergewhen we come to the question of the role of the categories in these two kinds ofsynthesis. In response to my comments, Longuenesse maintains that “[I]n part one Kantargues that the categories, albeit originating in the understanding alone, are conceptsunder the guidance of which the synthesis of any sensible intuition achieves that kind ofunity that allows it to be related to an object represented as distinct from our represen-tation of it.”11 Correlatively, she claims that, “In part two, he [Kant] argues that space andtime themselves, the forms of our sensibility, stand under the very same unity of apperceptionwhose discursive forms are the logical forms of judgment, and in which the categoriesthus originate as ‘concepts of an object, by means of which the intuition of the object isconsidered as determined with respect to a logical function of judgment’.”12

I have two points to make regarding Longuenesse’s account of the first part of theDeduction. First, it is unclear why she maintains that the object of an intuition as suchmust be considered as distinct from our representation of it; for I would have thoughtthat “inner” as well as “outer” objects fall under this description. Second and mostimportant, it seems evident that everything turns on what Longuenesse understands by“guidance.” Based upon her account in Kant and the Capacity to Judge, I take it that thisis to be understood in the teleological (or, as I have also called it, “Pickwickean”) sense,such as is allegedly at work in judgments of perception, according to which thecategories guide the understanding (the capacity to judge) in its endemic quest toform schemata, which through reflection can then be converted into concepts.As noted above, I agree with Longuenesse that some such process is involved in

concept-formation for Kant; in fact, I view this as one of the important insights containedin her book. I do not, however, see how this can be the sense operative in the first part ofthe B-Deduction. For one thing, as I argued in my essay, any appeal to schematism is outof place in the first part of the Deduction, where Kant is concerned with an intellectualsynthesis that abstracts from the question of the nature of sensibility. For another, I thinkthat any such thesis is too weak to establish the conclusion of the first part of theDeduction, which moves directly from the logical forms of judgment to the categories.As I argue in Kant’s Transcendental Idealism, this (together with the Metaphysical Deduc-tion towhich the first part of the Transcendental Deduction is closely connected) requiresthat we see the categories as embedded in the activity of judgment in the sense that

11 Longuenesse (2000), esp. 102. 12 Ibid., 102–3.

ADDENDUM TO “WHERE HAVE ALL THE CATEGORIES GONE?” 45

judging under a certain form necessarily involves a certain mode of unification of themanifold, which corresponds to (without being identical with) the unification thought inthe category.13 And I further argue that it is precisely this view of the categories thatunderlies Kant’s conclusion to the first part of the Deduction, when he claims to haveshown that “themanifold in a given intuition is necessarily subject to the categories” (KrVB143).14Unless Longuenesse is prepared to accept this view or something like it, I believethat she still owes us an account of how the subjection of themanifold of sensible intuitionin general to the categories is supposed to emerge from the appeal to the logical functionsof judgment in the first part of the Deduction.As Longuenesse herself acknowledges, however, our main area of disagreement

concerns the second part of the Deduction, where Kant relates the categories to humansensibility.15 Although the issues are complex, I believe that the essential point can beformulated succinctly. It concerns Longuenesse’s claim that, according to Kant, “spaceand time themselves, the forms of our sensibility, stand under the very same unity ofapperception whose discursive forms are the logical forms of judgment.” In myview, the truth is closer to the reverse of this; that is to say, it is not that space andtime stand under the unity of apperception, as if the unity of apperception weresomehow the source of their unity, but rather that the unity of apperception is itselfsubject to conditions imposed by space and time as forms of sensibility. In other words,despite the impression left by the first part of the Deduction, the necessity that allappearances are perceived as unified in a single time is not a requirement of the unity ofapperception, which would entail that there is a logical contradiction in the thoughtthat appearances might be perceived as located in two (or more) times (or spaces), but aconsequence of the nature of time (and space) as articulated in the Aesthetic.I further believe that this disagreement is reflected in our readings of Kant’s deeply

ambiguous and much discussed note at KrV B160–1, in which he distinguishesbetween space and time as formal intuitions and as forms of intuition. Perhaps themost contentious aspect of this note is its last sentence, where Kant writes: “For sincethrough it [the synthesis speciosa] (as the understanding determines the sensibility) spaceand time are first given as intuitions, the unity of this a priori intuition belongs to spaceand time and not to the concept of the understanding (}24).” At the risk of gross over-simplification, I think it fair to say that Longuenesse puts primary weight on the firstclause of this sentence, whereas I put it on the second. As I noted in my paper,Longuenesse takes Kant to be here claiming that space and time as products of thesynthesis speciosa are entia imaginaria and that this makes explicit something that wasmerely implicit in the Aesthetic.16 Although she denies that, qua such products, spaceand time are subject to the categories, Longuenesse also suggests that this is the keymove in the Deduction, since it entails that whatever is given in space and time, that is,

13 See Allison (2004), 147–56. 14 Ibid., 177. 15 Longuenesse (2000), 103.16 Although Longuenesse insists that her account does not involve a revision of the Aesthetic [(2000),

104–5], I find this questionable.

46 PART I

in sensible intuition, is subject to the unity of apperception, which presumably bringsthem under the categories in the Pickwickean sense described above, though not in thefull-fledged sense.17 By contrast, on my reading, the emphasis is to be placed on thesecond clause, which, with its claim that the unity of a priori intuition belongs to spaceand time rather than the concept of the understanding (the categories), affirms theindependence of sensibility and its forms vis-a-vis the understanding and its conditions.18

Indeed, this independence underlies what both Longuenesse and I view as the source ofthe problematic with which the Deduction is concerned, namely, the possibility that“appearances could . . . be so constituted that the understanding would not find them inaccord with the conditions of its unity . . . ” (KrV A90/B123). Assuming that the task ofthe Deduction is to exorcize this specter, there would seem to be two possible strategiesfor accomplishing this. One, which I believe to be that adopted by Longuenesse, is toargue that space and timemust be shown to be themselves subject to the conditions of theunity of apperception. Admittedly, this is the most natural way of understanding theargument of the Deduction and, eschewing details, it was the route taken by Kant’sidealist successors. Its consequence, however, is that the distinction between sensibilityand understanding is merely relative or provisional, which I believe to be incompatiblewith two fundamental principles of Kantian thought: the irreducible contrast between thespontaneity of the understanding and the passivity of sense and the thesis that cognitionrequires the cooperation of both (what I term the discursivity thesis). Although Long-uenesse acknowledges the importance for Kant of these principles, she also insists that herinterpretation of the B-Deduction preserves them.19 But it is precisely this that I finddifficult to understand; for if space and time are products of the synthesis speciosa, what is thecontribution of sensibility and its a priori forms?20

Nevertheless, it must also be admitted that to insist upon the independence of thesensibly given from the conditions of apperception or, what amounts to the same thing,the requirements of understanding (in Longuenesse’s terms the capacity to judge) is notwithout problems of its own. Specifically, it seems to leave the specter intact or, at best,to suggest that the agreement of the sensibly given with the requirements of theunderstanding is a merely contingent matter, a kind of pre-established harmony,which is likewise a view that Kant emphatically rejects.21 We thus seem to beconfronted with a dilemma that puts in question the coherence of Kant’s project inthe Deduction and which is directly related to the present-day debate regarding

17 Longuenesse (2000), 104–5.18 I interpret the first clause in light of my distinction between a determinate and an indeterminate

intuition. See Allison (2004), 115–16.19 See Longuenesse (2000), 107.20 Longuenesse’s answer seems to be that receptivity provides a potential form or the potentiality for form,

which is then actualized by the synthesis speciosa. I have expressed my doubts about this move in “Where Haveall the Categories Gone?”

21 Kant explicitly rejects such a view, presumably with reference to Crusius, in the final paragraph of theB-Deduction (B167–68). I believe that he makes essentially the same point in the A-Deduction in terms ofhis conception of transcendental affinity A113–14 and A122.

ADDENDUM TO “WHERE HAVE ALL THE CATEGORIES GONE?” 47

conceptual content: we can either insist that the spontaneity of the understanding“goes all the way down,” in which case the contribution of sensibility appears to belost, or we can maintain that the agreement of the sensibly given with our understand-ing is a merely contingent matter. On either view, however, the synthetic a priori statusof Kant’s claims regarding the categories is lost; for the first alternative requiressacrificing their syntheticity and the second their apriority.22

My proposal for avoiding this dilemma, which I believe corresponds to Kant’sconsidered view, is to introduce a third alternative, namely, that the synthetic unity ofapperception and with it the categories are necessary for representing the unity of timeand (space) as presented in the Aesthetic, as opposed to being the source of their unity. Inthe endeavor to clarify this complex and deeply controversial issue, I find it useful to askwhy Kant thought it necessary to introduce a special synthesis of the imagination (thesynthesis speciosa) to complement the intellectual synthesis of the first part. I believe thatthe answer lies in the nature of the connection between the imagination and therepresentation of time (and to a lesser extent space). This representation, like anycomplex thought, requires a synthesis because time (or space) contains a manifold andthe representation of a manifold as a manifold presupposes a synthesis. But this synthesisis the work of the imagination (rather than the understanding) because it is by definitiona capacity to represent in intuition what is not present and the representation of timepresupposes the representation of past and future times as parts of a single time.23

If this is correct, it follows that the unity belongs to space and time rather than to theconcept of the understanding. And it also follows that there is a sense in which theconformity of appearances with the conditions of the unity of the understanding isconditional, which is just to say that the specter of appearances not conforming to theseconditions is not logically impossible. Nevertheless, if Kant’s argument is sound, suchnon-conformity can be claimed to be phenomenologically impossible, since appearancesthat did not conform would be “nothing to us.” Moreover, though I cannot pursuethis matter here, it seems evident that all that Kant need assume to get the argument offthe ground is the non-controversial premise that we have perceptions, that is, con-scious representations (KrV A320/B376). This is because the brunt of the argument ofthe decisive }26 is to demonstrate that perception, through the empirical synthesis ofapprehension, stands under the categories and it is virtually as an afterthought that Kantproceeds to claim that experience, defined as cognition through connected perceptions(B161), is likewise subject to the categories.24

22 One of the central claims of my analysis of the argument of the B-Deduction is that the conclusion ofthe first part, namely, that “the manifold in a given intuition . . . necessarily stands under the categories”(B143) is derived analytically from the (analytic) principle of the synthetic unity of apperception. See Allison(“Where Have all the Categories Gone?” note 18). Accordingly, it is only in the second part, where thecategories are connected first with the forms of sensibility and then with their content (empirical intuition)that we have a synthetic a priori claim, which I take it is why Kant links the Deduction specifically with theargument of }26.

23 Ibid., 191–3.24 For my discussion of this see Allison (2004), 197–201.

48 PART I

Essay Three

Kant and the Two Dogmasof Rationalism

Philip Kitcher once suggested that Kant came close to writing “Two Dogmas ofEmpiricism.”1 Although I do not intend to challenge this intriguing suggestion,I shall attempt to take the matter in the opposite direction and propose that, with atleast equal justification, Kant may be thought to have come close to producing a workentitled “Two Dogmas of Rationalism.” In fact, in many ways the Critique of PureReason is just such a work. Or so I shall argue.The discussion is divided into five parts. In the first, I specify the two dogmas and

attempt both to explain their inherent plausibility and provide a preliminary sketch ofKant’s reaction to them. The remaining parts are devoted to a consideration of fourdomains in which these dogmas are at work and the Kantian critique of them is carriedout: the nature of human cognition; the distinction between analytic and syntheticjudgments; the nature of space and time; and the ideality of appearances. Far frombeing isolated topics, we shall see that these are intimately related to one another.

IThe first and best known of these dogmas is often referred to as the “predicate-in-notion principle” or some close variant thereof. It states that in every true propositionthe predicate is contained in the concept of the subject. The second does not comewith a familiar name and I shall call it the “reducibility principle.” It maintains thatsensible knowledge acquired through experience is reducible [in principle] to theintellectual variety, which is supposedly attained through the pure understandingindependently of any appeal to experience. Although each of these dogmas is asso-ciated directly with Leibniz, who was clearly the most important representative of therationalist tradition for Kant, they actually have a broader scope and may be viewed asunderlying assumptions of classical rationalism.However, before proceeding with the analysis of these dogmas and the Kantian

critique of them, there are two points to be noted. First, they reciprocally entail one

1 Kitcher (1981).

another. If one assumes the predicate-in-notion principle, then one is thereby com-mitted to the reducibility principle and vice versa. Second, they are mirror images ofQuine’s more famous dogmas, namely, the analytic–synthetic distinction and what heterms “reductionism,” which Quine initially characterizes as “the belief that eachmeaningful statement is equivalent to some logical construct upon terms which referto immediate experience.”2

In considering the first of these dogmas, it is important to realize that, howeverstrange the predicate-in-notion principle may seem to us today, it has an inherentplausibility, which was exploited by its rationalist proponents. In fact, for Leibniz, it is asimple consequence of a consideration of the nature of truth and it underlies the well-known rationalist dictum that “whatever I clearly and distinctly perceive about a thingis true or is assertable of the thing in question.”3 As Leibniz famously put it, “It iscommon to every true affirmative proposition, universal and particular, necessary orcontingent, that the predicate is in the subject, that is, that the notion of the predicate issomehow contained in the notion of the subject.”4

Here, as elsewhere when dealing with this topic, Leibniz is effectively asking:inasmuch as a true proposition is, by definition, one which correctly affirms (or denies)a predicate of a subject, what could its truth consist in other than its predicate beingincluded in (or, in the case of negative propositions, excluded from) the notion orconcept of the subject? To Leibniz, at least, this seemed to be virtually axiomatic.Nevertheless this seemingly innocent “axiom” has a number of not so innocentimplications, which Leibniz exploits in constructing his philosophical system. Firstand foremost, if everything truly predicable of x is contained in the concept of x,then for every x there must correspond a complete concept containing everything trulypredicable of it. Moreover, since finite beings like ourselves obviously do not possesssuch concepts, at least not of individual things, it follows that we must regard them aspossessed by God.Leibniz uses this principle to distinguish between necessary and contingent truths.

The former are those in which the containment of the predicate in the concept of thesubject can be ascertained by a finite process of analysis (either immediately or bydeduction from evident axioms), whereas the latter are those in which this is notpossible because of the infinite complexity of the concept. The former pertain to thedomain of “eternal truths,” which features, but is not limited to, those of mathematics;the latter pertain to what Leibniz calls “truths of fact.” It is not that one cannotdetermine a simple truth of fact, say that I (Henry Allison) am currently sitting beforemy computer, without providing an infinite analysis; it is rather that one cannot fullyunderstand the grounds of this truth (its sufficient reason) apart from possessing thecomplete concept of Henry Allison. It is in this sense that our sensible cognition, such

2 Quine (1963), 20. 3 Leibniz (1989), 26. 4 Ibid., 95.

50 PART I

as the example cited above, must be viewed as reducible in principle (though not by us)to the intellectual or purely conceptual variety, which is just the second dogma.Expressed in Kantian terms, this means that all true propositions are analytic, though

only a small subset of these (truths of reason) can be shown to be such because we lackthe requisite capacity for an intellectual intuition, which could comprehend infinitecomplexity at a glance.5 Clearly, if this were true, the analytic–synthetic distinctionwould have either to be rejected outright or viewed as a variable one, which is entirelya function of one’s level of cognition or, what amounts to same thing, of the“adequacy” of one’s ideas. In other words, what is synthetic for some might be analyticfor others, who possess a more adequate concept of the subject, and it necessarily is sofor God, in which case the distinction would lose much of its philosophical bite.Accordingly, it is no coincidence that this objection was raised against Kant by his

Leibnizian opponents.6 Indeed, since Kant uses this distinction as one of his majorweapons in his attack on traditional metaphysics, of which the Leibnizian is a promi-nent example, they had good reasons for doing so. For, seen from their point of view,Kant’s distinction rests upon a problematic assumption of its own, which is theconverse of the Leibnizian, namely, not everything that may be truly predicated of x iscontained in the concept of x. In fact, when looked at from a Leibnizian or, more generally,a rationalist point of view, this Kantian assumption seems highly paradoxical. For whatcould conceivably ground the truth of the predication of some property of x, otherthan the fact that it is contained in the complete concept of x?Kant’s answer to this question is simple and, from his point of view, obvious,

namely, such predication is made possible through an appeal to the intuition of x.Rationalism could hardly accept this response, however, since it rests on the assump-tion of a difference in kind (not merely degree) between intellectual concepts andsensible intuitions. And to accept this would require rationalism to abandon its seconddogma.In order to appreciate what is at issue here, we must revisit briefly the rationalist

position, against the backdrop of which Kant`s critique of the second dogma must beunderstood. As we have seen, essential to this position, and to the conception of truthon which it is based, is the notion of the complete concept of an individual. By its verynature, such a concept must be infinitely complex, since it encompasses everythingtruly predicable of the individual, including its relations to other individuals andultimately to the universe as a whole. Although rationalism denies that we (or anyfinite being) actually have such concepts, it tacitly assumes the legitimacy of what wemight term the “concept of a complete concept” as an idea that is to be approachedasymptotically, if never fully attained. Moreover, since it thereby appeals to a concept

5 For Leibniz, not even God could perform an infinite analysis, since such an analysis is by definition notcompletable. He also maintains, however, that such an analysis is not necessary, since God can grasp theinfinite immediately through an intuition. See, for example, (1989), 28.

6 I discuss this topic in Allison (1973), 36–46.

KANT AND THE TWO DOGMAS OF RATIONALISM 51

accessible only to God in its analysis of human cognition, rationalism may be appro-priately described as committed to a theocentric paradigm.7

By contrast, Kant’s so-called “Copernican revolution”may be viewed as a turn to anexplicitly anthropocentric model of cognition, a “paradigm shift” if you will. It is not,however, merely a shift to the human understanding (or human nature) as the firstobject of philosophical reflection. That is the standpoint of classical empiricism, whichfor Kant is equally “pre-Copernican.” It is rather something more radical and counter-intuitive, namely, the assumption that the human understanding is itself the source ofthe normative principles on which our cognition is based. Indeed, this is the truesignificance of Kant’s famous “experiment”: “Let us once try whether we do not getfarther with the problems of metaphysics by assuming that the objects must conform toour cognition” (KrV Bxvi). To suggest that objects conform to our cognition is not toimply that the human mind somehow creates its world ex nihilo; it is rather to advancethe epistemological thesis that objects must conform to the conditions through whichalone they can become objects for us.For present purposes, however, the essential point is that for Kant God’s knowledge

is not a norm against which the human variety is to be measured and found wanting,but a problematic concept through a contrast with which the latter is to be understood.Expressed in Kantian terms, this conception of cognition serves a “critical” rather thana “dogmatic” function: it frees us from one picture of cognition (the theocentric) inorder to enable us better to appreciate another (the anthropocentric). Moreover, it isonly against this backdrop that we can understand not merely Kant’s critique ofrationalism but the “critical” project as a whole.

IISince the kind of cognition that pertains to the anthropocentric paradigm is discursive,I call Kant’s claim that our cognition is of this sort the “discursivity thesis.” To say thatour cognition is discursive is to say that it is based on concepts; but since conceptsrequire some already given data to be brought under them, discursive cognition is to beunderstood as consisting in the application of concepts to sensible intuition, which isthe sole source of the requisite data. Thus, such cognition, at least as Kant conceives it,assumes an ineliminable duality or distinction in kind between its sources or conditions.As Kant puts it at the very beginning of the Transcendental Analytic:

Our cognition arises from two fundamental sources in the mind, the first of which is thereception of representations (the receptivity of impressions), the second the faculty for cognizingan object by means of these representations (spontaneity of concepts); through the former anobject is given to us, through the latter it is thought in relation to that representation (as a meredetermination of the mind). Intuition and concepts therefore constitute the elements of all our

7 For my discussion of this thesis, see Allison (2004), 27–38.

52 PART I

cognition, so that neither concepts without intuition corresponding to them in some way norintuition without a concept can yield a cognition. (KrV A50/B74)

As this passage indicates, what makes intuitions and concepts irreducible to one anotheris that the former are connected with the receptivity of the senses and the latter withthe spontaneity of the understanding. Accordingly, all our intuitions, including thepure or a priori variety, are sensible in nature. And since without the contribution ofsensibility there would simply be nothing to be thought through concepts, sensibilitymakes an indispensable contribution to human cognition. It is equally important,however, to keep in mind that sensible intuition is merely a necessary and not also asufficient condition of discursive cognition. As its name suggests, the latter also requiresconcepts through which the intuitively given content can be thought.According to Kant’s oft-cited formula:

Thoughts without content are empty, intuitions without concepts are blind. It is thus just asnecessary to make the mind’s concepts sensible (i.e., to add an object to them in intuition) as it isto make its intuitions understandable (i.e., to bring them under concepts). (KrV A51/B74)

As a first step in understanding the import of the discursivity thesis, we must becomeclear about the nature of the two elements (concepts and intuitions) required fordiscursive cognition. But since this is a complex issue, about which there remains agood deal of controversy in the literature, I can here attempt to provide merely agreatly simplified account.8

In his Logic, Kant defines an intuition simply as a “singular representation” [reprae-sentatio singularis] (JL 6: 91). He repeats this in the Critique, adding that it is “immedi-ately related to the object” [bezieht sich unmittelbar auf den Gegenstand] (KrV 320/B 377).Thus, a Kantian intuition has two defining properties: singularity and immediacy,which turn out to be inseparable. What such an intuition does is to present particularsto the mind and, for reasons that should become clear shortly, it can do so only if itpresents these particulars “directly” or “immediately,” that is, without the mediation ofa concept. Moreover, it must be kept in mind that this is intended to apply to the“genus” intuition, which itself contains two radically distinct species: the sensible andthe intellectual. The former is characterized as “sensible” because it requires anaffection by some object; the latter is termed “intellectual” because it consists in thespontaneous excogitation of an object, without any sensory input. Although ourintuition is necessarily of the former sort and is the only kind with which we areacquainted, Kant thought the division important for at least two reasons: first, it helpsto distinguish his view from that of Leibniz and other rationalists; second, it indicatesthat the characterization of our intuition as sensible is informative, which, in turn, isnecessary for appreciating the discursive nature of our cognition.

8 For a fuller treatment, see Allison (2004), 77–88.


Correlatively, in his Logic Kant characterizes a concept (as opposed to an intuition) as“a universal representation, or a representation of what is common to several objects,hence a representation insofar as it can be contained in various ones.” Consequently, itis redundant to speak of universal or common concepts, as if concepts could be dividedinto universal, particular, and singular. “Concepts themselves,” Kant remarks, “cannotbe so divided, but only their use” (JL 9: 91). In the parallel account in the Critique, Kantnotes that a concept, in contrast to an intuition, refers to its object “mediately by meansof a mark [eines Merkmals], which can be common to several things” (KrV A320/B377).In other words, because of its generality, a concept can refer to an object only by meansof features that are also predicable of other objects falling under the same concept.Thus, unlike a Leibnizian complete concept, a Kantian (discursive) concept, no matterhow complex, can never refer uniquely to a particular object. That is why intuition isrequired to acquaint the mind with particulars, thereby determining whether or notthe concept has an application.Since a concept is a collection of marks (or “partial representations”), which are

themselves concepts, every concept may be viewed as a collection of concepts. Forexample, the concept of gold is that of a yellow, malleable metal that is soluble in aquaregia etc., each of which is itself a concept potentially applicable to other things besidethose composed of gold. Kant does not, however, understand such a collection ofmarks in the manner of Berkeley and Hume as merely the product of frequentexperience, joined together in the mind by principles of association and broughtunder a general term. Instead, he considers concepts as rules for the unification inthought (not the imagination) of these marks. Accordingly, to have the concept of goldis to have the thought of these (and other) marks as linked together conceptually. Thisis not to say that being a yellow metal somehow entails being soluble in aqua regia. Thepoint is rather that if something is thought to be gold it must also be thought ascontaining properties corresponding to its marks, that is, as metal, yellow, malleable,and soluble in aqua regia, etc.This presupposes what has been aptly termed the “fixity” of concepts and we shall

see below that it provides the key to understanding the Kantian view of analyticity.9

But we first need to explore somewhat more fully the nature and function of conceptsin general, which requires understanding their connection with judgment. In fact, thefundamental unit of discursive thought for Kant is the judgment, not the concept.Discursive thinking, for Kant, just is judging and the latter consists in the subsumptionof the data provided by intuition under concepts stemming from the understanding.Thus, while there can be no judging without concepts, it also follows that “theunderstanding can make no other use of its concepts than that of judging by meansof them” (KrV A68/B93).

9 On the notion of the fixity of concepts, see Beck (1967), 237.

54 PART I

In the same context, Kant also remarks that “Concepts . . . serve as predicates ofpossible judgments” (KrV A69/B94). In characterizing concepts in this way, Kant isnot limiting their function to that of being logical or grammatical predicates. Forexample, in Kant’s paradigmatic judgment, “All bodies are divisible” (which he regardsas analytic), both concepts function as predicates in the judgment. Since the subjectconcept (“body”) provides the initial content for the judgment, it may be said to standin a more direct, though still not immediate, relation to the object, because no conceptcan do that. Thus, the judgment affirms that the x’s (the data given in the intuition towhich the concept is applied) also fall under the broader concept (“divisibility”). Thisanalysis enables Kant to claim that,

All judgments are . . . functions of unity among our representations, since instead of an immediaterepresentation a higher one, which comprehends this and other representations under itself, isused for cognition of an object, and many possible cognitions are thereby drawn together intoone. (KrV A69/B94)

Setting aside obscurities in Kant’s account of judgment, it might appear that hisdiscursivity thesis, which is completely intertwined with the conception of judgment,is not particularly newsworthy or controversial. Indeed, it has been viewed by present-day philosophers in much the same way as the predicate-in-notion principle wasregarded by Leibniz and his fellow rationalists. Typical of this approach is Strawson,for whom discursivity reduces to the inescapable necessity in any thinking aboutexperience or empirical knowledge to assume a “duality of general concepts, on theone hand, and particular instances of general concepts on the other.”10

Although this may be correct as far as it goes, to leave matters here is to ignore therevolutionary nature of Kant’s thesis. In order to understand this, however, it isnecessary to view this thesis in its historical context as challenging both rationalismand empiricism. The challenge to empiricism is obvious and need not be pursuedfurther here. Since the empiricists (with the partial exception of Locke) tended to viewideas as particular images, they really had no place for anything like a Kantian conceptand the discursive view of cognition it entails. But the challenge to the rationalists callsfor some comment, since they typically did not deny conceptual representation as suchand were as opposed as Kant to the “picture theory” of the empiricists.The issue is a subtle one and once again I cannot here pretend to give it the attention

it deserves. The main point is that, even though the rationalists did not altogether rejectconceptual representation and the discursive model of cognition it entails, they didtend to regard it as inferior to intuitive cognition. This is perhaps clearest in the case ofSpinoza, who distinguished between ratio, which is just discursive cognition, and whathe terms scientia intuitiva, which supposedly acquaints us directly with the essence ofindividual things, rather than indirectly by means of concepts applicable to a number ofthings (common notions).11 In Malebranche, Spinozistic intuition becomes the capac-

10 Strawson (1966), 20. 11 See Spinoza (1982), 90 [Ethics, Part 2, Proposition 40, Scholium 2].


ity to see all things in God; while in Leibniz, Spinoza’s distinction between ratio andscientia intuitiva reappears as the contrast between “symbolic” or “blind” and “intuitive”cognition. Moreover, though both may be adequate, only the latter is “absolutelyperfect,” which might be glossed as “fully adequate.”12 In fact, the point is alreadyevident from rationalism’s characteristic tendency to elevate the idea, with its norma-tive properties of clarity, distinctness, and adequacy, over the judgment as the basicepistemic unit. Once the idea is given such a status, the first dogma of rationalismbecomes unavoidable and the discursivity thesis a non-starter.From the Kantian standpoint, perhaps the most important feature of rationalism’s

approach to conceptual representation and discursivity is its ambiguous view of theepistemic function of sensible intuition. Rationalists tend to deny any essential cogni-tive role for such intuition. Although it is generally acknowledged that some sensoryinput is required to set the mind’s cognitive machinery in motion, it is also claimed thatthis is merely a consequence of our finitude and does not conflict with the coredoctrine that sensory cognition is the lowest and least adequate kind and ought to bereplaced by more adequate modes of cognition to the extent possible. At the sametime, however, by considering cognition in this way, rationalism shows itself com-mitted to the thesis that the senses of themselves yield some cognition, albeit of thelowest grade. Moreover, here again, rationalism comes into conflict with Kant’sdiscursivity thesis, since the latter affirms that without the contribution of the under-standing there could be no cognition at all, not even the highly inadequate kindassumed by the rationalists.Finally, it must be emphasized that rationalism’s ambiguous treatment of the sensory

component of human cognition is no mere accident or failure to recognize theobvious. It is rather a direct consequence of its commitment to a theocentric paradigm.That is why, from Kant’s point of view, nothing less than a Copernican revolution isrequired in order to recognize the discursive nature of human cognition and theindispensable but limited role therein of sensibility.

IIIInasmuch as post-Quinean discussions of the analytic–synthetic distinction tend tofocus on the concept of analyticity and the notorious difficulties involved in giving it anacceptable definition or characterization, it is not surprising that Kant’s critics concen-trate on his manifestly inadequate account of analytic judgments in the Introduction tothe Critique. Kant there describes such judgments in two supposedly equivalent ways:(1) those in which “the predicate B belongs to the subject A as something that is(covertly) contained in this concept A”; and (2) as “those in which the connection ofthe predicate is thought through identity” (KrV A6–7/B10–11). For reasons that I shall

12 Leibniz (1989), 25.

56 PART I

not rehearse here, these formulations have met with almost universal disapproval fromKant’s time to the present. Equally important, though less often noted, they give thefalse impression that synthetic judgments are to be regarded virtually as an afterthought,since they appear to be defined in essentially negative terms as those that are notanalytic. And from this point of view, which is that of those who, like Quine,understand the analytic–synthetic distinction in empiricist terms, this effectivelymeans that synthetic judgments as such are seen as relatively unproblematic, with thewhole problem being with those that are supposedly both synthetic and a priori.Although the centrality of the problem of synthetic a priori judgments to the

“critical” project can hardly be gainsaid and will be touched upon below, this approachto Kant’s analytic–synthetic distinction is seriously misleading for at least two reasons.First, it ignores the fundamentally anti-rationalistic orientation of the distinction itself,that is, it fails to recognize its significance as an essential ingredient in the Kantiancritique of the first dogma of rationalism. As we have seen, this dogma poses a challengeto the possibility of synthetic judgments in general, at least insofar as they are thought tobe more than provisionally or relatively synthetic, not merely to those that are suppo-sedly a priori. Second, it neglects the close connection between the analytic–syntheticdistinction and the discursivity thesis or, what amounts to the same thing, Kant’saccount of judgment. The distinction, after all, is claimed to be between differentspecies of judgment. Accordingly, if we are to understand Kant`s distinction betweenthese two species of judgment, it is necessary to keep in mind their common genus.Obvious as it may seem, the latter point is generally neglected, perhaps because of

the casual way in which Kant introduces the distinction in the Introduction to theCritique and the corresponding portion of the Prolegomena, neither of which makes anyreference to the nature of judgment as such. Nevertheless, this may easily be explainedby their location in the texts prior to any discussion of the topic. Like many philoso-phers, Kant was frequently confronted with the problem of not being able to sayeverything that needed to be said on a topic at once. Thus, his initial presentation ofmany important distinctions tends to have a provisional character and often needs to bereinterpreted in light of his subsequent treatment of the subject. The distinctionbetween analytic and synthetic judgments is no exception.Fortunately, Kant does indicate elsewhere, if not fully explain, the connection between

the analytic–synthetic distinction and his underlying theory of judgment. Perhaps themost helpful of these is his brief discussion of the matter in the published version of hisLogic, where the distinction between the two species of judgment is drawn in terms of acontrast between a formal and a material extension of knowledge (JL 9: 111).Analytic judgments are there said to provide a formal extension of knowledge by

clarifying or explicating what is already contained in a given concept. This involves theuncovering of implications of which one may not have been previously aware, butwhich are derivable by strictly logical means from this concept, which once again isregarded as fully determinate or “fixed.” Kant takes “All bodies are extended” as anexample of an analytic judgment, which he renders schematically as “To everything x


to which the concept of body (a + b) belongs, belongs also extension (b)” (JL 9: 111).This shows that in analytic judgments the predicate (b) is related to the object (x) byvirtue of the fact that it is already contained (as a mark) in its concept (the logicalsubject). Analytic judgments are therefore “about” an object; they have a logicalsubject and, as Kant’s example shows, they may also have a real one. Nevertheless,since the truth or falsity of the judgment is determined merely by analyzing the conceptof the subject, the reference to the object (the real subject) is otiose.13

In his polemic with Eberhard, who pressed him on the matter from a Leibnizianpoint of view, Kant supplements this account by introducing what amounts to adistinction between immediately and mediately analytic judgments. “All bodies areextended” is immediately analytic, because “extension” (together with “figure,”“impenetrability,” etc.) is a mark of the concept “body.” By contrast, “All bodies aredivisible” is only mediately analytic because “divisibility” is not itself part of theconcept (logical essence) of body, but rather of one of its constituent marks (“exten-sion”). In other words, it is a mark of a mark. Despite the fact that this implies that thejudgment rests on an inference and in that sense extends our cognition, Kant insists thatthis does not amount to a difference in kind, since in both cases the predicate is derivedfrom the concept of the subject by a process of analysis.This shows that Kant’s conception of analyticity is of a piece with the discursivity

thesis. As the preceding account indicates, both rest upon the conception of a conceptas a set of marks (themselves concepts), which are thought together in an “analyticunity,” and which can serve as a ground for the cognition of objects. These markscollectively constitute the intension of a concept. One concept is contained in another,just in case it is either a mark of the concept or a mark of one of its marks. In either case,it is subordinated to the concept in which it serves as a mark, which is precisely therelation that is brought out in an (affirmative) analytic judgment. Thus, unlike mostcontemporary conceptions of analyticity, Kant’s is thoroughly intensional.A synthetic judgment, by contrast, extends our knowledge in a “material” sense.

Kant’s example is “All bodies have attraction,” which he renders schematically as “Toeverything, to which the concept of body (a + b) belongs, belongs also attraction (c)”(JL 9: 111). Like its analytic counterpart, this judgment asserts a connection betweenthe predicate (c) and the object (x), which is thought through the concept (a + b). But,unlike the latter, it asserts this independently of any connection between the predicateand the concept of body. To be sure, in the judgment the predicate is connected withthis concept; but the connection is grounded in, and mediated by, the reference ofboth to an intuition of the object. Consequently, the judgment extends our knowledgeof x (in this case, of all x’s) by providing a determination or property of x that is notalready contained in its concept. This is what is meant by a “material extension.”

13 Beck (1967), 230.

58 PART I

Looked at from this point of view, the problem is to explain how a materialextension is possible at all, whether it be a priori or a posteriori. And since this isprecisely what is denied by rationalism’s first dogma, Kant’s account of the possibility ofsuch judgments may be viewed as his response to the challenge posed by that dogma.But, even though this effectively transforms the notorious problem of the synthetic apriori into a special case of a more general problem rather than a freestanding one,synthetic a priori judgments present difficulties of their own, which call for somecomment. Within the Kantian framework, what is distinctive about such judgments isthat they require both pure concepts and pure intuitions. It is obvious that they requirepure concepts as predicates and some corresponding intuition. Without the former theycould make no claim to universality and necessity (the criteria of the a priori), andwithout the latter they would not be synthetic. The question is therefore why anempirical intuition does not suffice and a pure one is required. Indeed, this invites thefurther question of whether something like the latter is even possible. Certainly, Kant’spredecessors, rationalist and empiricist alike, had no place for such a hybrid conception.Moreover, Kant himself poses the problem in a particularly sharp form when he asks,“How is it possible to intuit something a priori?” (Pro 4: 281). This is a problem thatarises only for intuitions (not concepts), since it suggests that something could be givenin intuition prior to and independently of our experiential encounter with it, that is tosay, a priori.I shall return to the latter question in the next section, in connection with a

consideration of Kant’s views on space and time, which, as pure forms of sensibility,are alleged to be the sources of the required pure intuitions. But before doing so, it isnecessary to address the first and lesser of these questions. In brief, the answer is that amerely empirical intuition does not suffice for a synthetic a priori judgment because ofits particularity. As the representation of a particular individual, an empirical intuitioncannot exhibit intuitively the universality thought in the concept. To cite a familiarmathematical example: as synthetic, the judgment that the sum of the interior angles ofa triangle is equal to two right angles must be grounded in an intuition of a triangle,while, as a priori, it cannot be grounded in the intuition of any particular triangle. Itspossibility thus rests upon there being some non-empirical or pure intuition ofsomething like “triangularity as such,” that is, a singular representation that nonethelesscan “attain the generality of the concept, which makes this valid for all triangles, rightor acute etc.” (KrV A141/B180). Moreover, the problem is somewhat more complexin the case of judgments involving pure, as opposed to mathematical, concepts, becausethese cannot be directly exhibited in intuition. For example, we cannot just “see” orperceive a cause, though we can judge that certain occurrences are such. As I haveargued elsewhere, however, Kant’s solution to this problem is to assign such a role tothe transcendental schema.14

14 See Allison (2004), 210–28.


IVKant’s account of space and time is, among other things, a direct assault on rationalism’ssecond dogma. Without explicit reference to Leibniz, Kant initially characterizes thelatter’s position as the view that space and time are “only determinations or relations ofthings, yet ones that would pertain to them even if they were not intuited.” This iscontrasted with Kant’s own view that they are “relations that only attach to the form ofintuition alone, and thus to the subjective constitution of our mind, without whichthese predicates could not be ascribed to anything at all” (KrV A23/B37–8). Kant hereexpresses a partial agreement with Leibniz. They are one in rejecting the Newtonianabsolutistic theory in favor of a relational one, but they differ radically in the under-standing of the nature and function of the relations involved. As Kant makes clear, theissue between them turns on the connection of these relations to the “subjectiveconstitution of our mind,” which he here equates with its form or manner of sensibleintuiting. By contrast, for Leibniz, at least as Kant read him, human sensibility and itssubjective conditions have nothing to do with the true nature of space and time,though, as we shall see below, they do play a major role in explaining why we thinkabout them in the way that we do.In order to understand Kant’s critique of the Leibnizian view (I shall limit the

discussion to space), it is necessary to take a brief look at the latter. Here the chieftext is Leibniz’s correspondence with Clarke. Although Leibniz’s initial concern in thisexchange was to criticize the Newtonian absolutistic theory rather than to spell out orto defend his own view, under pressure from Clarke he was led to do both, particularlyin his fifth and final letter. Space, as Leibniz there construes it, is nothing but the orderof coexisting phenomena and in light of this he attempts to explain why we nonethe-less naturally tend to think of it in the Newtonian (and perhaps commonsensical)manner as something more than such an order.Since Leibniz viewed the latter as a fiction (and a theologically dangerous one at

that), it is not totally surprising that his account seems almost Humean at points.Basically, his claim is that space viewed as something more than an order is animaginative gloss, which results from our inability to perceive distinctly minutedifferences between the relations of the situation of things and which is itself aconsequence of the limitation of our senses. This inability, Leibniz further argues in aproto-Humean fashion, leads to a confounding of resemblance or agreement withnumerical identity. On this basis, he then contends that the notion of an absolute space,which somehow subsists apart from the order of things in it, is a “mere ideal thing,”meaning thereby a fiction.15 Thus, while the “true,” that is, adequate, representation ofspace as an order of coexisting things is purely intellectual, the one that we actuallyhave contains a sensory component, which reflects the limits of our conceptualcapacities.

15 Leibniz (1956), 69–72.

60 PART I

This is diametrically opposed to the Kantian view, according to which space serves asa condition of the representation of such an order rather than as a sensibly distorted orconfused product thereof. Later in the Aesthetic and elsewhere, Kant expresses hisdisagreement by claiming that the Leibnizians falsify the concept of sensibility and ofappearance by “considering the distinction between sensibility and the intellectual asmerely logical” (KrV A44/B61). “Logical,” as Kant here understands it, means adifference in degree (clarity and distinctness) rather than in kind. Thus, the claim thatthe Leibnizians “falsify” the concepts of appearance and sensibility is equivalent to theclaim that they regard what is, in fact, a difference in kind between two radicallydistinct species of cognition (the sensible and the intellectual) as if it were merely amatter of degree. Once again, this is the second dogma of rationalism, which, as alreadynoted, is entailed by the first.Kant’s basic strategy is to turn the tables on Leibniz. Rather than being derived from

some kind of pre-spatial ordering, conceivable by the intellect alone, which, due to theunavoidable limits of our senses, is mistaken for something more, the representation ofspace is now seen as a subjective condition of the cognition of the very order fromwhich Leibniz claims it is acquired. Kant further argues that this representation stands inan ineliminable relation to human sensibility, which blocks the kind of reduction that isaffirmed by the Leibnizians.In the Kantian lexicon, this dual thesis amounts to the claim that the representation

of space is both a priori and intuitive. After establishing this thesis to his own satisfactionin what in the second edition of the Critique is called the “Metaphysical exposition ofthis concept” (KrV B37–40), Kant proceeds to argue in the “Transcendental exposi-tion” (KrV B40–1) that this alone is able to account for the synthetic a priori status ofgeometry, a thesis that a Leibnizian rationalist must deny, since it runs directly counterto the first dogma.This is Kant’s notorious “argument from geometry,” which has come under intense

criticism ever since the discovery of non-Euclidean geometries, if not before. Fortu-nately, for present purposes, this issue can be ignored, since Kant viewed his account ofgeometry as an application or confirmation of the result already attained in the“Metaphysical exposition” rather than as an independent argument. Although Kantis committed by his account of the conditions of the possibility of synthetic a priorijudgments in general to the view that if geometry is to be regarded as both syntheticand a priori it must rest on a pure intuition of space, this is merely a necessary and notalso a sufficient condition of the possibility of geometry so conceived. Accordingly,I shall here consider the argument of the “Metaphysical exposition” in its own terms,bracketing the issue of the adequacy of Kant’s views on geometry.This argument is composed of four sub-arguments. The first two attempt to show

that the representation of space is a priori rather than empirical and the second two thatit is an intuition rather than a concept. In essence, the apriority arguments contend thatthe representation of space must be regarded as a priori, since it serves as a condition ofthe experience of any spatial ordering of things and therefore cannot, on pain of


circularity, be regarded as derived from such an experience. Taken by itself, this isperfectly compatible with the possibility that the representation of space is an a prioriconcept (an innate idea), which, as such, has nothing essentially to do with humansensibility. Consequently, we must look to the intuition arguments to establish theconnection with sensibility that is necessary in order to refute the Leibnizian view.Indeed, if the representation of space were a concept, the second dogma of rationalismwould hold.Assuming his underlying discursivity thesis, according to which concept and intui-

tion constitute the two elements of human cognition, Kant in effect argues byelimination. Since space cannot be the former it must be the latter, that is, therepresentation of an individual that is somehow immediately present to the mind,prior to and independently of any act of conceptualization. It is not that we cannotform concepts of space (clearly, we can and do); it is rather that any such conceptspresuppose a prior intuition. Thus, the original representation of space must be anintuition.Although Kant provides two distinct arguments for this thesis, one from the

singularity of space (we can represent to ourselves only a single space of which allparticular spaces are delimited parts) and the other from its infinity or limitlessness,I shall here consider these together, since they turn on essentially the same point,namely, Kant’s view of a concept as a general representation. We have seen that aconcept, so understood, is composed of marks (partial representations), which arethemselves concepts and which can serve in turn as marks of other concepts. Inother words, every concept, qua concept, has both an intension and an extension.The former is constituted by the marks of which it is composed and the latter by theconcepts falling under it of which it serves as a mark.Given this, Kant has little difficulty in showing that the representation of space

possesses neither of the defining features of a concept, though, as a singular representa-tion, it does fit the definition of an intuition. First, unlike the relation between aconcept and its marks, we cannot think of this single space as a product of its parts(particular spaces). On the contrary, unlike the relation between a concept and themarks of which it is composed, that is, its intension, the parts of space presuppose thewhole and can be thought as parts only in relation to it. Second, these parts are thoughtas in space rather than as falling under it, as they would if space were a concept and theyconstituted its extension.It is in the context of his conclusions from the analysis of the representation of space

as both pure (a priori) and intuitive that Kant explicitly addresses the problem of thepossibility of an a priori intuition posed above. His solution is that such an intuition ispossible only “if it contains nothing else except the form of sensibility, which in me as subjectprecedes all actual impressions through which I am affected by objects” (Pro 4: 282). In otherwords, the only thing that could conceivably be intuited a priori by discursive cognizersis the form or manner in which the sensible data for the thought of objects are given tous in empirical intuition. As devoid of sensation (the “matter” of empirical intuition),

62 PART I

such a form would be a priori, while as pertaining to intuition rather than to thought, itwould belong to sensibility. And since the representation of space has been shown to bea pure intuition, it follows that the content of this intuition (what is actually intuited apriori through it) is merely a subjective form under which objects are given to us inempirical intuition.Kant concludes from this that, while “empirically real” (applicable to everything

encountered in human experience), space is nonetheless “transcendentally ideal,” thatis, a nonentity, if one abstracts from the subjective conditions of human sensibility andconsiders it as either a property or relation of things as they are in themselves. And sinceLeibniz is committed to the latter alternative, it follows that his view is incorrect.It might seem that a Leibnizian could attempt to evade the force of this argument by

challenging the discursivity thesis on which it is based. For if Kant’s sharp concept–intuition distinction be denied, one can hardly argue in the Kantian manner that therepresentation of space must be intuitive simply because it is not conceptual. In fact, thedenial of the discursivity thesis might be thought to open up conceptual space forthe possibility that our representation of space is a kind of intellectual intuition, albeitone that must remain confused for finite beings. It is not so easy to argue in this way,however, since, as we have seen, it would involve a host of questionable epistemologi-cal and metaphysical assumptions of its own, not the least of which is a commitment tothe two dogmas.

VAlthough Kant had already characterized space and time as transcendentally ideal in theAesthetic, he only gets around to defining transcendental idealism itself in the Dialectic.He does so in two places, only one of which is retained in the second edition. In theone that is retained, Kant depicts this idealism as the doctrine that “all objects of anexperience possible for us, are nothing but appearances, i.e., mere representations,which, as they are represented, as extended beings or series of alterations, have outsideour thoughts no existence grounded in itself.” This is contrasted with a “transcendentalrealism,” which is charged with making “these modifications of our sensibility intothings subsisting in themselves, and hence makes mere representations into things inthemselves” (KrV A490–01/B518–19). As Kant here describes these two forms oftranscendentalism, they appear to encompass the entire philosophical landscape.16 Oneeither regards the objects of possible human experience as appearances or one does not,in which case they are viewed as things in themselves. Moreover, if this is true, itfollows that transcendental realism must be understood in a very broad sense asincluding all other philosophical positions, many of which are not generally regarded

16 For further accounts of the relation between transcendental idealism and transcendental realism seeAllison (2004) 20–48 and 384–95, as well as Allison (1998b) and (2006b), both of which are included in thepresent volume.


as forms of realism. In other words, it is to be understood as a kind of metaphilosophicalstance from which the problem of human knowledge is considered, rather than as adistinct metaphysical or epistemological position.If this is correct it follows that transcendental realism’s opposite, transcendental

idealism, is likewise to be considered in this way, that is, as a metaphilosophical stance,rather than, as it usually is taken to be, a distinctive metaphysical doctrine. Moreover,so considered, it may be seen as a correlate of the anthropocentric paradigm, which isthe product of Kant’s Copernican revolution.In order to appreciate what is at stake here, it is essential to recognize that the very

idea of a Copernican revolution, as described above, brings with it an idealisticcommitment of a sort, inasmuch as it regards the human mind (with its sensible andintellectual conditions) as the source of the norms and principles through which whatcounts as “objective” is determined by and for cognizers such as ourselves. Since theseconditions include the sensible conditions through which objects are given in intuition(space and time), it follows that we can cognize objects only as they appear to us underthese conditions, not as they may be in themselves independently of them. And, as wehave already seen, Kant’s position here differs in toto from the rationalist’s with itstheocentric paradigm, which assumes that the cognitive task is to apprehend truths thathold independently of any conditions or principles of the human mind, that is, to comeas close as possible to the God’s-eye view of things from which alone the truth is fullyaccessible. Indeed, we have seen that precisely such a view underlies the Leibnizianreduction of the sensible to the intellectual. Although this is not a reduction that wecan perform, it is one that the theocentric paradigm requires us to assume is performedby God.The methodological nature of Kant’s idealism becomes more apparent when one

examines the key distinction between appearances and things in themselves on which itis based. To begin with, “things in themselves” is to be understood as elliptical for“things as they are in themselves” or, more precisely, “things considered as they are inthemselves.”17 In other words, the Kantian distinction is between two ways ofconsidering things, that is, the very objects of our experience, in a reflection on theconditions of their cognition (transcendental reflection) rather than between twoontologically distinct kinds of thing. And since, lacking a capacity for an intellectualintuition, sensibility and understanding are the only cognitive tools by means of whichthese things can be considered, it follows that to consider them independently of theirrelation to human sensibility and its a priori conditions just is to consider them asputative objects of some “pure understanding,”which is not ours. Once again, this is toadopt the theocentric paradigm. Moreover, this also explains why Kant insists that wecan think things as they are in themselves, even though we cannot cognize them assuch.

17 On this point see Prauss (1974), esp. 13–23.

64 PART I

Before proceeding further, it must also be noted that, despite the theological terms inwhich transcendental realism’s conception of the cognitive task has been characterized,it is the picture implicitly adhered to by common sense, which naturally assumes thatthe proper object of cognition is a reality existing an sich. From this standpoint,anything short of this simply does not count as knowledge, or, if it is regarded assuch, it can be only knowledge of how things seem to us rather than of how they reallyare. At one place in the Critique Kant refers to this assumption as the “common butfallacious presupposition of the absolute reality of appearance” (KrV A535/B564),while at another he characterizes it as the “common prejudice” (KrV A740/B768).Nevertheless, it may still seem odd for Kant to charge rationalism, particularly the

Leibnizian variety, with affirming the absolute reality of appearance, thereby effectivelyregarding the spatiotemporal objects of human experience as if they were things inthemselves. Since Kant’s main objection against Leibnizian rationalism is that itsystematically denies a positive cognitive role for sensibility, and since it is preciselybecause of the ineliminable role of the latter that Kant thinks that we are acquaintedmerely with things as they appear rather than as they are in themselves, it might seemmore appropriate for Kant to charge such rationalism with simply denying the reality ofappearances rather than with conflating them with things in themselves. Consider, forexample, the following passage from Leibniz, which is typical of his thought on thematter:

If Bodies are phenomena and judged in accordance with how they appear to us, they will not bereal since they will appear differently to different people. And so the reality of bodies, of space, ofmotion, and of time seems to consist in the fact that they are phenomena of God, that is, theobject of his knowledge by intuition [scientia visionis]. And the distinction between the appear-ance bodies have with respect to us and with respect to God, is, in a certain way, like thatbetween a drawing in perspective and a ground plan. For there are different drawings inperspective, depending upon the position of the viewer, while a ground plan or geometricalrepresentation is unique. Indeed, God sees things exactly as they are in accordance withgeometrical truth, although he also knows how everything appears to everything else, and sohe eminently contains in himself all other appearances.18

This passage provides a nice illustration of how Kant appropriated Leibnizian termi-nology, while giving it a radically different sense. For both thinkers, bodies in space andtime are “mere phenomena,” and space and time themselves are regarded as “ideal.”From the Kantian point of view, the difference is that bodies are not “phenomena ofGod,” since there is no such thing, and space and time are transcendentally ideal. Inother words, looked at from the framework of the anthropocentric paradigm, the latterare forms of human sensibility, which partially define the conditions and bounds ofpossible experience for beings like ourselves, rather than unavoidable, sources of thelimitations of our cognition. For the same reason, Kant would reject the favorite

18 Leibniz (1989), 199.

KANT AND THE TWO DOGMAS OF RATIONAL ISM 65

Leibnizian metaphor of a perspective. It is not that we perceive the same world thatGod does, albeit from a limited perspective or point of view, but rather that weperceive our own world, that of possible experience, which is governed and limitedby the conditions of our cognition.This does not mean, however, that our world is, ontologically speaking, less than

“really real,” since there is no other. What it means instead is that the traditional idea ofa God’s-eye view of things (a “view from nowhere”) is to be understood as a limitingconcept, helping to define the conditions and limits of our cognition, rather than as agoal to be approached asymptotically. In fact, the reason why such a limitation isnecessary is that the human understanding, as a faculty of thought, has a naturaltendency to extend itself beyond its proper sphere, which is limited by our forms ofsensibility. Thus, like the Cartesian will, the reach of the Kantian understandingexceeds its grasp. Moreover, for Kant, the clearest example of this in modern philoso-phy is the Leibnizian monadology and the two dogmas on which it is based.

66 PART I

Essay Four

Transcendental Realism, EmpiricalRealism, and TranscendentalIdealism

Customarily, the debate regarding the interpretation of transcendental idealism is thoughtto concern the question of whether Kant’s transcendental distinction between appearancesand things in themselves is between two types of thing (the “two-object” or “two-world”view) or two aspects of ordinary empirical objects (the “two-aspect” view). But eventhough I have long been associatedwith the latter camp, I have thought formany years thatthis is not the most helpful way to frame the issue. The problem lies in an ambiguity in thetwo-aspect view. It can be understood eithermetaphysically, as a form of property dualismin which objects are assigned both phenomenal and noumenal properties, or methodolog-ically, as a contrast between twoways inwhich objects can be considered in a philosophicalreflection on the conditions of their cognition. Accordingly, I take the fundamentalquestion to be whether transcendental idealism is to be understood in the latter way oras a form of metaphysical dualism (whether as a thing or a property dualism being a matterof relative indifference). And I have further thought that the best way of addressing thatquestion is through a consideration of the view which Kant opposes to transcendentalidealism, namely, transcendental realism. If this realism is identified with a particularmetaphysical doctrine then transcendental idealism must be as well; but if, as I maintain,transcendental realism cannot be so understood, then neither can Kant’s idealism.1

I shall here argue obliquely for the latter alternative by exploring the nature of thedifference between transcendental and empirical realism. The discussion consists of fourparts and a brief appendix. The first part maintains that, rather than being either a distinctontological doctrine or a mere label for everything to which Kant was opposed, transcen-dental realism should be understood as the view that spatiotemporal predicates are

1 In arguing for a non-metaphysical interpretation of transcendental idealism, I do not intend to deny thatthis idealism has important ontological or, more broadly, metaphysical implications. Clearly, the arguments ofthe Aesthetic, Analytic, and Dialectic, all of which are intimately connected with transcendental idealism,have such implications and were intended by Kant to have them. Thus, if anyone wishes to preserve the term“metaphysical” for Kant’s central claims I have no objection. As will become clear in due course, what I wishto insist upon here is simply that transcendental idealism is not itself to be understood as a metaphysical theorythat affirms that the phenomenal has a lesser degree or kind of reality than the noumenal.

applicable to things in general. Since this view is shared by all ontologies (at least all thosewith which Kant was familiar), transcendental realism is not committed to a particularontology; but since it is contrasted with an empirical realism, which limits the scope ofthese predicates to objects of possible experience, it is also not a vacuous label. And fromthis I conclude that transcendental idealism likewise does not constitute a distinct onto-logical position, but is rather to be equated with an empirical realism that insists upon thisscope limitation.2 The second and third parts deal with two possible objections to thisreading. The first is that it conflicts with Kant’s official account, which charges transcen-dental realism with conflating appearances with things in themselves rather than withinflating claims about objects of possible experience into claims about things in general. Byexamining the relationship between the concepts involved, I argue that there is noincompatibility between the two characterizations. The second line of objection is thatsuch an interpretation of transcendental idealism trivializes it by reducing it to a recom-mendation of epistemological modesty. I respond by acknowledging that transcendentalidealism, so construed, does consist essentially in such a recommendation, but deny thatthis trivializes it. The point is further discussed in the fourth part, which analyzes Kant’sindirect argument for transcendental idealism drawn from his resolution of the anti-nomies. In the appendix, I surmise that one reason why many contemporary Kantiansare dismissive of transcendental idealism is that they mistakenly assume that a centralconcern of the Critique is to establish a robust realism in the face of a radical skepticalchallenge. I suggest instead that Kant argues from rather than to such realism and that hiscentral concern is to limit its scope to objects of possible experience.

IKant defines transcendental realism in two places in the Critique and in each of them hecontrasts it with transcendental idealism. In the first of these, he characterizes it as theview “which regards space and time as something given in themselves (independent ofour sensibility)” and suggests that such a realist “interprets outer appearances . . . asthings in themselves, which would exist independently of us and our sensibility andthus would also be outside us according to pure concepts of the understanding.” Bycontrast, transcendental idealism is defined as the doctrine that all appearances are “tobe regarded as mere representations and not things in themselves, and accordingly thattime and space are only sensible forms of our intuition, but not determinations givenfor themselves or conditions of objects as things in themselves” (KrV A369). In thesecond, transcendental realism is accused of making “modifications of our sensibility

2 The view I am here attributing to Kant has obvious affinities with the position which Hilary Putnamterms “internal realism,” and which he regards as Kantian. I am not sure, however, to what extent Putnamwould be willing to accept my reading of Kant as an account of what Kant actually held as opposed to whathe should have held. For a useful discussion of Putnam’s internal realism and its relation to Kant see Moran(2000).

68 PART I

into things subsisting in themselves, and hence makes mere representations into thingsin themselves;” whereas transcendental idealism affirms that “all objects of an experi-ence possible for us, are nothing but appearances, i.e., mere representations, which, asthey are represented, as extended beings or series of alterations, have outside ourthoughts no existence grounded in itself” (KrV A490–91/B518–19). Although it ismore explicit than the first, both accounts indicate that the dispute between the twoforms of transcendentalism concerns primarily the nature of space and time and, as aconsequence of this, the nature of things encountered in them. In essence, Kant’scharge is that by viewing space and time as “given in themselves,” independently of theconditions of sensibility, transcendental realism conflates spatiotemporal entities, whichfor the transcendental idealist are “mere appearances,” with things in themselves.Since any number of views could be characterized as regarding space and time as

“given in themselves” and since Kant himself accuses philosophies of many differentstripes of conflating appearances with things in themselves, these characterizations oftranscendental realism seem much too vague to define a metaphysical position withwhich transcendental idealism might meaningfully be contrasted.3 Alternatively, if, as issometimes done, transcendental realism is identified with the scientific realism of theCartesians and Newtonians (roughly what Berkeley understood by “materialism”),then the situation seems even worse. First, it implies that Kant’s idealism is akin toBerkeley’s, something which Kant vehemently denied.4 Second, it invites the familiarcharge of neglected alternatives. For, clearly, Kant did not intend to suggest that suchrealism and his idealism exhaust the philosophical universe. Moreover, this issuebecomes particularly pressing in connection with Kant’s indirect argument for tran-scendental idealism through the negation of transcendental realism; for unless it isassumed that they are contradictory opposites, this argument cannot get off the ground.5

We thus appear to be confronted with a dilemma. If transcendental realism is seen asan amorphous, ill-defined metaphysical theory, it becomes difficult to see how it couldcontribute to an understanding of transcendental idealism; whereas if we identify itwith a particular metaphysical theory it leads to the outright dismissal of an argumentwhich Kant thought to be central to his project and, depending on the nature of this

3 This seems to be denied by Ameriks, who at least at one point characterizes transcendental realism as a“particular metaphysical position,” albeit without further identifying the position in question. See Ameriks(2000), 290.

4 Interpreters who take this view include Turbayne (1955), 228, and Al-Azm (1972), 148.5 Allen Wood [ (2007) 9] has stated that he finds “both puzzling and problematic” the way in which

I juxtapose the two forms of transcendentalism, “not as fairly definite positions in metaphysics or epistemol-ogy, but as something like two encompassing world views.” But while I view these as “encompassing” in thesense of exhausting the alternatives, I regard them as contrasting philosophical methodologies rather thanworld views. Moreover, I submit that Kant’s indirect argument for transcendental idealism requires that theybe understood in this way. Although this argument itself may be problematic, it does not seem to meproblematic that it presupposes that the relation between transcendental idealism and transcendental realismbe understood in this way. As far as it being puzzling is concerned, I shall simply point out that Kant affirms asimilar dichotomy in moral theories between those based on the principle of the autonomy of the will andthose based on the principle of heteronomy.

TRANSCENDENTAL REALISM, EMPIRICAL REALISM, TRANSCENDENTAL IDEALISM 69

theory, perhaps to a highly unattractive view of transcendental idealism as well. Myproposed way out of this dilemma is through a consideration of the difference betweena transcendental and an empirical realism. In so doing, I hope to show that the former isnot to be understood as a distinct ontological theory, while at the same time giving itenough shape to preserve the significance of its contrast with transcendental idealism.The operative question is what makes transcendental realism transcendental, which, in

turn, calls for an investigation of Kant’s multiple uses of this term. But since anythingapproaching an adequate treatment of the topic is well beyond the scope of the presentessay, it must suffice to note that Kant works with at least two competing conceptionsof the transcendental. The first and featured conception is the one that he introduces inhis stipulative definitions of “transcendental cognition” (KrV A12/B25). According tothese definitions, such cognition may (very roughly) be characterized as a second-orderactivity concerned with an investigation of the a priori elements of our cognitiveapparatus and the conditions and limits of our a priori cognition of objects.6 Thesecond conception is the traditional one, according to which transcendental cognitionis concerned with the nature of things in general, that is, with the subject matter ofontology or metaphysica generalis, particularly as it was understood by the Wolffianschool. And just as Kant’s contemporaries were quite familiar with the latter but hadgreat difficulty understanding the former, so for present-day readers the situation isvirtually the reverse. Nevertheless, we shall see that the recognition of the continuedpresence in the Critique of vestiges of this earlier conception is crucial for the under-standing of both transcendental realism and transcendental idealism.A good illustration of the juxtaposition of these two conceptions is to be found in

Kant’s remark concerning the meaning of “transcendental cognition,” a remark whichKant advises the reader to keep well in mind, since its import extends to everything thatfollows (KrV A56/B80). Although the bulk of this remark is devoted to underscoringthe distinction between a priori and transcendental cognition, understood as a contrastbetween a first-order cognition, such as is found in mathematics, and a second-order

6 I say stipulative definitions, since Kant offers a significantly different one in each edition. In the firstedition, Kant writes: “I call all cognition transcendental that is occupied not so much with objects but ratherwith our a priori concepts of objects in general” (KrV A12). In the second, transcendental cognition isdefined as that which “is occupied not so much with objects but rather with our mode of cognition of objectsinsofar as this is to be possible a priori” (KrV B25). The most thorough treatment of the subject is by Pinder(1986). According to Pinder, in the A version Kant is trying to indicate that the central focus of transcendentalcognition and therefore of theCritique itself will be on our a priori concepts of objects rather than on objects (orthings) themselves, which would characterize the ontological approach. Since a concern with such conceptsinvolves also one with the objects supposedly falling under them, it will be concerned (albeit indirectly) withthe latter as well.Hence, Kant’s use of the “not so much . . . but rather” [“nicht sowohl . . . sondern”] locution. Bycontrast, in the B version, Pinder thinks that Kant’s focus has shifted to a more narrow concern with thepossibility of synthetic a priori judgments, which reflects the central concern of the Prolegomena. Since thedetails of this shift, as important as they may be for an understanding of the development of Kant’s thought,are not directly relevant to the concern of this essay, I have attempted to provide a characterization of Kant’sdefinitions that covers both versions.

70 PART I

cognition that is concerned with the possibility of the former, in the concludingportion of this remark, Kant adds parenthetically:

Likewise the use [Gebrauch] of space about all objects in general would also be transcendental; butif it is restricted solely to objects of the senses, then it is called empirical. The difference betweenthe transcendental and the empirical therefore belongs only to the critique of cognitions and doesnot concern their relation to their object. (KrV A56–7/B81)

Whereas the contrast between transcendental and a priori cognition is a matter of levelthat between a transcendental and an empirical use is a matter of scope, which is why itpertains to a critique of cognitions. Accordingly, it is here that Kant’s employment of“transcendental” overlaps with the traditional understanding of the term. Moreover,this is neither an isolated occurrence nor a regression on Kant’s part to a pre-criticalstandpoint, since, as we shall soon see, he uses the term in precisely this way incontrasting a transcendental with an empirical use of the categories and principles ofpure understanding. In both cases, it amounts to a contrast between a generic use withrespect to all objects and one restricted to a particular domain of objects, namely,phenomena or objects of possible experience.This has two implications, which are essential to the proper understanding of both

transcendental realism and transcendental idealism. The first is that the differencebetween transcendental and empirical realism consists in the scope assigned to spatio-temporal predicates rather than the degree or kind of reality attributed to them. Inother words, it is not that empirical realism assigns a lesser degree of reality to suchpredicates (as if objects only seem to be spatiotemporal), but merely that it restricts theirapplicability to objects of possible experience. The second is that, though transcenden-tal realism is committed to the proposition that spatiotemporal predicates are ontologi-cal in the traditional sense of applying with strict universality, it is not, as such,committed to any particular ontology of space and time.The latter point is reflected in the Transcendental Aesthetic, where Kant begins by

raising the purportedly ontological question: What are space and time? Four possibi-lities are introduced. They might be: (a) actual entities (substances), (b) determinationsof things (accidents), (c) relations of things that “would pertain to them even if theywere not intuited,” or (d) “relations that only attach to the form of intuition alone, andthus to the subjective constitution of our mind, without which these predicates couldnot be attached to anything at all” (KrV A23/B37–8).7 The first three represent thetraditional ontological options and, as such, apply to things in general. Of these, thesecond and third correspond to the Newtonian and Leibnizian views. The fourth is

7 That Kant lists four possibilities, rather than merely the three that I suggested in Allison (1983), has beennoted by Falkenstein (1995), 147. As he correctly notes, this was already pointed out by both Vaihinger(1892), 131–4, and Martin (1955), 11–12. Moreover, there are several other texts in which Kant clearlydistinguishes between these possibilities, including ID 2: 400 and 403, R 5298: 18: 146–7, and R 5404 18:174. Nevertheless, at least from the time of the Dissertation, Kant effectively assumed that the only twoalternatives worthy of serious consideration were the Newtonian and the Leibnizian positions.


Kant’s “critical” view and reflects his rejection of the whole ontological framework inwhich the question had traditionally been posed, one in which it is assumed thatwhatever status is given to spatiotemporal predicates they apply with strict universality.Thus, from Kant’s more comprehensive point of view, the dispute between theNewtonians and the Leibnizians is reduced to a family quarrel.Seen in this light, Kant’s basic charge against transcendental realism is that it

erroneously assumes that spatiotemporal predicates have an unrestricted scope. More-over, since Kant’s doctrine of the ideality of space and time provides the warrant for hisscope restriction, this further suggests that transcendental idealism, even as it appears inthe Aesthetic, is best seen as a deflationary proposal rather than as an ontological thesis indirect competition with the various forms of transcendental realism.8 Otherwiseexpressed, inasmuch as the first three of the four possible accounts of space and timethat Kant introduces exhaust the recognized ontological alternatives, the fourth, whichrepresents Kant’s “critical” position, might be regarded as proposing a radical alterna-tive to ontology rather than, as it usually taken to be, a novel move within ontology.Specifically, the alternative is to consider space and time as “two sources of cognition”(KrV A38/B55), that is, as conditions of our cognition of things, rather than as them-selves either things (substances), properties, or relations of things as such.9

Nevertheless, this does not suffice to establish Kant’s restriction thesis, since we havenot ruled out the possibility that space and time are conditions of the cognition ofthings in general. To this end, it is essential for Kant to demonstrate their connectionwith human sensibility. What must be shown is that the universality of space and timewithin human experience, a point on which the empirical and the transcendental realistagree, can be understood only on the assumption that their representations are a prioricontributions of human sensibility, from which the scope limitation to objects ofpossible human experience (or at least to the experience of cognizers with our formsof sensibility) follows immediately. Whether Kant actually succeeds in showing this isanother and more contentious question, which I cannot consider here.10

8 I emphasize the Aesthetic because it is here that the ontological reading of Kant’s ideality thesis seemsmost compelling. Thus, if this reading can be challenged here, the stage is set for a comprehensive non-metaphysical interpretation of transcendental idealism.

9 Once again, if anyone wishes to insist that this remains a move within ontology because it involves aglobal rejection of the generally accepted ontological alternatives, I have no objection. I would point out,however, that Kant’s account fundamentally changes the nature of the game by transforming what wereformerly regarded as ontological into epistemic conditions.

10 I analyze and attempt to defend Kant’s argument in Allison (2004), esp. 122–32. Here, I wish merely topoint out that a direct and important corollary of this reading is the assignment of a central place to theintuition arguments of the Aesthetic, since it is through these alone that Kant attempts to link therepresentations of space and time with human sensibility. Assuming their apriority, if space and time wereassigned to the understanding rather than to sensibility, Kant would have had to conclude that they arepredicable of things in general. Accordingly, it is of no little significance for Kant to be able to show that,“Space is not a discursive or, as is said, general concept of relations of things in general, but a pure intuition”(KrV A24–5/B39).

72 PART I

IIEven setting aside the question of the cogency of Kant’s arguments, it might beobjected that this analysis misrepresents his actual accounts of transcendental realism,since the latter make no explicit reference to things in general, but refer instead tothings in themselves. Moreover, rather than accusing transcendental realism of a simplescope confusion, we have seen that Kant charges it with the seemingly more heinouscrime of making “modifications of our sensibility” or “mere representations” intothings in themselves. How, then, it may be asked, are these accounts to be reconciled?The short answer is that, even though the concepts of a thing (or something) in

general and of a thing in itself occupy distinct spheres of philosophical reflection, theyare related in such a way that if spatiotemporal (or, indeed, any) predicates wereapplicable to the former, they would ipso facto be applicable to the latter as well.Textual support for this reading is provided by Kant’s account of the putative tran-scendental use of the categories and their associated principles, by which he under-stands one with regard to the cognition of things in general.11 As was the case withspatiotemporal predicates, this is contrasted with an empirical use, which is restricted tothe cognition of objects of possible experience or appearances.Although Kant defines the categories (nominally) as concepts of an object in general,

he denies that they could have the former employment, since they would then lack theschemata necessary to link them up to the world.12 But he also suggests that if (perimpossibile) they had such an employment, “as conditions of the possibility of things ingeneral, they [could] be extended to objects in themselves (without any restriction toour sensibility)” (KrV A129/B178). Similarly, Kant states that, “The transcendental useof a concept in any sort of principle consists in its being related to things in general andin themselves” (KrV A238/B298), which suggests that he considered the relation of aconcept to things in themselves to be a direct consequence of its relation to things ingeneral and therefore as not requiring any further explanation.13

11 In addition to the passages cited below, Kant refers to a putative transcendental use of the pure conceptsand/or their associated principles at KrV A139/B178, A19/B266, A242, A246/B303, A247/B304, A296/B352–3, A402–3, A515/B544. In most of these places it is contrasted with a legitimate empirical use. In theTranscendental Deduction, however, Kant views this distinction in a quite different way with respect to thefaculties of sense, imagination, and apperception, each of which is claimed to have a legitimate transcendentaluse as well as an empirical one (KrV A94/B127). But here “transcendental” refers to their function asconditions of the possibility of experience, which obviously does not involve any reference to things ingeneral.

12 Although Kant explicitly denies that the pure categories can be defined at KrV A245, he there also statesthat they are “nothing other than the representations of things in general, insofar as the manifold of theirintuition must be thought through one or another of these logical functions.” For Kant’s definitions, see KrVA93/B126, A248/B305, A253, and A290/B346. Thus, following Beck, I think it best to regard Kant asproviding a nominal definition of the categories. See Beck (1965), 61–73.

13 See, for example, KrV Bxxvii, where Kant claims that if the distinction between things as objects ofexperience and the very same things as things in themselves were not drawn, then the principle of causalitywould be valid of things in general as efficient causes; and KrV B410, where Kant suggests that if the rationalpsychologists were right, synthetic propositions “could reach as far as things in general and in themselves.”


Nevertheless, since this connection may not be as obvious to contemporary readersas it was to Kant, it will be useful to take a closer look at the concepts involved. Tobegin with, we have seen that Kant takes the concept of a thing in general over wholecloth from the ontological tradition and that it encompasses whatever pertains to thethought of a thing as such. Accordingly, the project of ontology, traditionally under-stood, is to provide cognition of things by means of an analysis of this concept. Andsince Kant saw this as involving an illicit transcendental employment of the under-standing, he proclaimed that, “[T]he proud name of ontology, which presumes to offersynthetic a priori cognitions of things in general in a systematic doctrine . . . must giveway to the modest one of a mere analytic of the pure understanding” (KrV A247/B304). As this passage indicates, the crux of the problem lies in the pretension ofontology to provide synthetic a priori cognitions of things in general, when, in fact, allthat it can deliver are analytic judgments, which simply unpack what is already thoughtor presupposed in the concept of an object as such. By contrast, a “mere analytic,” thatis to say, Kant’s Transcendental Analytic, supposedly accounts for the possibility ofsynthetic a priori cognitions involving the categories by restricting their scope toobjects of possible experience, just as the Transcendental Aesthetic does for spaceand time.Whereas Kant took over the concept of a thing in general from the ontological

tradition and criticized the way in which it was used, the concept of a thing in itself is acreation of his own, forming one part of the contrast pair: appearance–thing in itself.Moreover, although this concept shares with that of a thing in general the thought ofan independence from any putative conditions of sensible intuition, it understands thisin a significantly different way. In the latter case, it is simply a matter of disregarding themanner in which objects are given in such intuition. Since this manner is assumed notto pertain to the thought of a thing qua thing, it is not considered relevant to atraditional ontological reflection. But inasmuch as the concept of a thing in itselfcontains the thought of something as it is in itself, independently of any sensibleintuition, it requires an active factoring out or exclusion of any contribution ofsensibility rather than merely a refusal or failure to factor it in.It should be clear from this how Kant can move seamlessly from the thought of

things in general to that of things in themselves. Since whatever is predicated of theformer is predicated of things absolutely or in every relation, it must also be predicatedof them as they are in themselves.14 Moreover, it follows from this that there is noincompatibility between an account of transcendental realism as erroneously attribut-ing spatiotemporal predicates to things in general and Kant’s charge that such realism isguilty of attributing them to things in themselves. On the contrary, I believe that thediagnosis offered here enables us to understand why, when seen from Kant’s point ofview, the transcendental realist would invariably make the latter mistake. It is precisely

14 See Kant, KrV A324–25/B380–82.

74 PART I

because he inflates spatiotemporal predicates into predicates of things in general thatsuch a realist unavoidably attributes them to things in themselves as well.

IIIInsofar as this reading emphasizes the importance for Kant’s project of denying thespatiotemporality of things in themselves rather than merely maintaining a cautiouscritical agnosticism regarding the matter, it is in agreement with Paul Guyer’s. We arediametrically opposed, however, in our views regarding the nature, implications, andsuccess of this denial. For Guyer, it is a straightforward thesis in dogmatic metaphysics.As he puts it, “Transcendental idealism is not a skeptical reminder that we cannot besure that things as they are in themselves are also as we represent them to be; it is aharshly dogmatic insistence that we can be quite sure that things as they are inthemselves cannot be as we represent them to be.”15 And, in light of this, he dismissesmy attempted rehabilitation of transcendental idealism as an “anodyne recommenda-tion of epistemological modesty.”16

That transcendental idealism on my reading involves a recommendation of epistemo-logical modesty rather than an illicit venture into dogmatic metaphysics cannot begainsaid.17 The question is whether it is also “anodyne” or otherwise trivial. This line ofobjection seems to have been fueled, at least in part, bymy previous attempts to argue thatthe denial of the applicability of spatiotemporal predicates to things as they are inthemselves follows directly from the connection of these predicates with the forms ofhuman sensibility, together with the Kantian conception of things as they are in them-selves or, more precisely, things considered as they are in themselves. Since by the latter ismeant things considered independently of their epistemic relation to human sensibilityand its conditions or, as Kant sometimes puts it, things as objects of a “pure understand-ing,” it follows that if space and time are, as Kant argues in the Aesthetic, forms orconditions of human sensibility, then they cannot also pertain to things so considered.18

This has generated the suspicion that I have endeavored to resolve a substantivemetaphysical dispute by a semantic sleight of hand, making the non-spatiotemporalityof things as they are in themselves virtually into a matter of definition.19 In addition, it

15 Guyer (1987), 333.16 Ibid., 336.17 Kant himself explicitly says as much when he remarks in response to the Garve-Feder Review: “The

principle that governs and determines my idealism throughout is . . . All cognition of things out of mere pureunderstanding or pure reason is nothing but sheer illusion, and there is truth only in experience” (Pro 4: 374).

18 By a “pure understanding” Kant here means one which, unlike ours, operates independently ofthe conditions of sensibility, that is, one which purports to cognize objects through the pure or unschema-tized categories. This locution is especially prominent in the Phenomena and Noumena chapter.

19 A somewhat different version of the triviality objection has been voiced by Langton (1998), 8–12.Focusing on Kant’s denial that we can have knowledge of things in themselves (which is what she under-stands by “Kantian humility”) rather than on their non-spatiotemporality, Langton argues that, on myreading, this supposedly momentous discovery on Kant’s part reduces to the trivial analytic claim that wecannot know things in themselves because doing so would (by definition) mean knowing them in abstraction


is sometimes argued that simply because, given the nature of our sensibility, thingshappen to appear to us as spatiotemporal, it does not follow that they are not also inthemselves “really” such. And, similarly, simply because in considering things as theyare in themselves one is considering them apart from their epistemic relation to thehuman mind, it does not follow that they do not really (in themselves) have the featuresthat they are taken to have in virtue of this relation.20

Since this line of objection is composed of two distinct parts, I shall respond brieflyto each. My response to the first part is simply that it ignores the context of theproblem, which is that of the classical neglected alternative objection, that is, the chargethat Kant neglected the possibility that space and time might be both forms of humansensibility and pertain in some unspecified way to things as they are in themselves.21 Itdoes not seem to me to be inappropriate to address that objection by pointing out that itrests on a misunderstanding of the terms. Nor does it trivialize Kant’s position. Rather,it indicates that the real work is done in the arguments for the claim that space and timeare forms of sensibility, by which is meant that they structure the way in which thehuman mind receives its sensory data in virtue of its peculiar manner of intuiting.22

However problematic these arguments may be, it should at least be clear that neitherthey nor the conclusions that Kant draws from them are trivial.With regard to the second part of the objection, my position is that its central claim is

correct but irrelevant. Although it is certainly true that the fact that one can consider xwithout considering its y-ness does not entail that it does not possess this property, this isnot what is being claimedwhen spatiotemporal predicates are denied of things consideredas they are in themselves. The point is rather that in considering things in this manner oneis, ex hypothesi, considering them apart from the condition under which alone suchpredicates are applicable to them, namely, in their relation to human sensibility. Accord-ingly, it is not simply the case that spatiotemporal predicates are ignored or set aside whenthings are considered in this manner, but that they are explicitly denied of the object quaconsidered apart from that relation.23 The situation is analogous to the treatment of

from the conditions of our knowledge. My response is that Kant’s revolutionary and certainly non-trivialclaim is that our cognition is governed by sensible conditions. Given this, together with Kant’s account ofwhat cognition of things in themselves (or as they are in themselves) would require, the unknowability thesisfollows. But this hardly makes the latter claim trivial, particularly since the transcendental realists whom Kantwas attacking did not acknowledge that human cognition is subject to sensible conditions in anything like thesense insisted upon by Kant and, as a result, they assumed that we could cognize things as they are inthemselves.

20 For a recent statement of this line of objection, see Howell (2004), 120.21 I have treated this topic in some detail in Allison (1976), (1983), (1996a), 8–11, and (2004), 128–32.22 My analysis of this issue has been challenged by Falkenstein (1995), 301–5. According to Falkenstein,

I ignore the possibility that a transcendentally real space might be just like the space of human sensibilityexcept for its dependence on the latter. For my response to this criticism, see Allison (2004), 130–2.

23 The point here is the same as in the distinction drawn above between the ways in which the concepts ofthings in general and of things as they are in themselves involve an independence from the conditions ofsensibility. The critic is, in effect, treating Kant’s claim about the latter (the non-spatiotemporality of things asthey are in themselves) as if it were about the former.

76 PART I

weight in Newtonian physics. Just as for Newton bodies may be said to have weight onlyinsofar as they are considered as standing in a relation of attraction and repulsion to otherbodies, so for Kant things may be said to have spatiotemporal properties only insofar asthey are considered in their epistemic relation to the human mind and its forms ofsensibility.24 In both cases it is a matter of a thing considered in a certain relation in virtueof which it falls under a particular description, and the same thing considered apart fromthis relation, as a result of which the description no longer applies.Moreover, I believe that this whole line of objection reflects an erroneous, albeit

widely held, transcendentally realistic picture of Kant’s idealism. According to this familiarpicture, things as they are in themselves are equated with things as they “really are,”whereas things as they appear are things as they are for us, subject to the limits imposed bythe nature of our sensibility.25 And given this picture, Kant’s claim that space and time aremerely empirically rather than transcendentally real is taken as implying that they are not“fully” or “really real,” which leads to another uncomfortable dilemma. Depending onone’s view of the “metaphysics of transcendental idealism,” we can either attribute toKant the view that things only seem to us to be spatiotemporal, though in truth they arenot, or we can take him to be positing a distinct set of entities (appearances), which reallyare spatiotemporal, whereas things in themselves (the “real things”) are not.26 In otherwords, according to this picture, transcendental idealism seems to require us to sacrificethe reality of either our cognition or its object.Admittedly, much of Kant’s terminology strongly suggests this unappetizing picture;

but inasmuch as it measures human knowledge in light of the ideal of a “God’s-eye”view, I believe that it distorts the thrust of his view in the Critique. And though thispicture might be challenged in a number of ways, I have here attempted to do so bysuggesting that the difference between a transcendental and a merely empirical realismconcerns the scope of the spatiotemporal predicates they both affirm rather than thekind or degree of reality assigned to them. If this is correct, it follows that transcenden-tal idealism does not require either of the above-mentioned forms of sacrifice. What itrequires is rather the abandonment of the unwarranted presumption that the spatio-temporal structure of our experience is projectable onto things in general.In order to appreciate the force of Kant’s claim, it is essential to keep in mind that the

spatiotemporal structure of human experience does seem to be projectable in this way,which is why, without the salutary lessons of a critique of pure reason, we are allpredisposed to be transcendental realists. Accordingly, Kant found it necessary to curbthis tendency, which he accomplishes in two steps. The first occurs in the intuitionarguments in the Aesthetic, where he argues that our representations of space and time are

24 I initially appealed to this analogy in Allison (1983), 241–2. It has been criticized by Cleve (1999), 147–9.I defend my use of this analogy against Van Cleve’s criticisms in Allison (2004), 42–5.

25 Strawson expresses this view with admirable succinctness when he defines transcendental idealism as thedoctrine that “reality is supersensible and that we can have no knowledge of it” (1966), 38.

26 The classical formulation of this dilemma is by Prichard (1909), esp. 71–100.


intuitions rather than concepts. This provides the basis for the reassignment of theserepresentations to sensibility rather than to the understanding. As already indicated, thisreassignment is crucial because the understanding, by its very nature, does project its“pure” (a priori) concepts onto things in general.27 The second step consists in theintroduction of the concept of the noumenon in the negative sense, which is just theconcept of an object that is not the object of a sensible intuition.28 The function of thisconcept is to “limit the pretension of sensibility” (KrV A255/B311); and since this“pretension” is that sensible, i.e., spatiotemporal, predicates apply to things in general,this limitation is central to Kant’s “critical” project.29 Moreover, it brings with it thereplacement of a transcendental by an empirical realism and therewith a commitment totranscendental idealism.

IVWe learn in the Transcendental Aesthetic and Analytic that the projection of thespatiotemporal structure of our experience onto things in general is unwarranted, but itis only in the Dialectic that we come to appreciate that it is also disastrous. This is the clearmessage of the most important part of the Dialectic: The Antinomy of Pure Reason,where Kantwarns ominously that “If wewould give in to the deception of transcendentalrealism, then neither nature nor freedom would be left” (KrV A543/B571).The loss of nature on the assumption of transcendental realism is the main lesson to

be learned from the “mathematical antinomies.” Confining ourselves to the first ofthese, the problem is that the endeavor to think the spatiotemporal world as a wholeleads to two equally warranted but seemingly contradictory conclusions, namely, thatthis world is infinite in duration and extent and that it is finite in both respects. Andthis, in turn, suggests that the very concept of a spatiotemporal world, like that of asquare circle, must be self-contradictory, since it generates contradictory entailments.30

27 Kant underscores this point in the introductory portion of the Transcendental Deduction common toboth editions, when he notes that the seemingly unrestricted scope of the categories, “not only arousessuspicion about the objective validity and limits of their use but also makes the concept of space ambiguousby inclining us to use it beyond the conditions of sensible intuition, on which account a transcendentaldeduction of it was also needed above” (KrV A88/B120–1).

28 By contrast, a noumenon in the positive sense would be an actual object of a non-sensible intuition.The latter is a problematic concept for Kant because we cannot determine whether such an entity (or modeof intuition) is really possible. Kant indicates the connection between what I have termed the two steps of hisanalysis when he remarks that “the doctrine of sensibility is at the same time the doctrine of the noumenon inthe negative sense” (B307). I discuss the different senses of the noumenon and their relation to thetranscendental object in Allison (2004), 57–64.

29 The concept of the noumenon serves to limit the pretension of the understanding as well, albeit in anindirect manner, by way of the dependence of the understanding on sensibility. Since cognition through theunderstanding (discursive cognition or judgment) requires that its object be given in sensible intuition, andlimitation on the scope of the latter will limit that of the former as well.

30 See Pro 4: 341, where Kant poses the issue in this logical form.

78 PART I

Kant’s resolution consists in claiming that these alternatives are merely contrariesrather than genuine contradictories and that both are false. This is because theseconclusions share the unwarranted assumption that the spatiotemporal world is a“whole existing in itself” (KrV A506/B534).31 Given this assumption, it follows thatthe world must have a determinate duration and extent, which presumably must beeither finite or infinite; that is, to say, there must be some “fact of the matter” regardingthe age and size of the world. After rejecting this assumption on the grounds that itgenerates a contradiction, Kant argues indirectly for transcendental idealism by imply-ing that transcendental realism (in all its forms) is irrevocably committed to it, whereastranscendental idealism is not. Thus, the truth of this idealism supposedly follows fromthe negation of transcendental realism.Although virtually every step in this argument is deeply controversial, I shall here

limit myself to two questions:32 (1) Why is transcendental realism necessarily com-mitted to the proposition that the world is a “whole existing in itself,” such that it mustbe either finite or infinite in the relevant respects? (2) Assuming that transcendentalidealism follows from the denial of transcendental realism, what does this tell us aboutthe nature of this idealism?As elsewhere in the Dialectic, Kant’s diagnosis of the pathology of transcendental

realism in the cosmological domain turns on linking it with his doctrine of transcen-dental illusion.33 The locus of this illusion is the seemingly innocent principle ofreason: “If the conditioned is given, then the whole series of all conditions for it isalso given” (KrV A497/B525). Since the conditions at issue here are just the compo-nents of the sensible world, to assume that the whole series of such conditions is given isto assume that this world constitutes a “whole existing in itself.” What makes thisprinciple seem innocent is that it merely expresses the explanatory requirement to seekthe conditions for every conditioned and not to stop until the absolute totality of theseconditions, which by definition is itself “unconditioned,” is attained. What makes itillusory is that such an absolute totality can never be given (whether as finite or infinite)in a possible experience. But since the cosmological dispute concerns the duration andextent of the spatiotemporal world, it purports to be about something that is, at least inprinciple, an object of possible experience. Thus, the transcendental realist is led by thisillusory principle to “hypostatize” the spatiotemporal world, that is, to treat it as a

31 Although this assumption does not enter as a premise into either the thesis or antithesis argument of anyof the antinomies, it underlies the cosmological debate as a whole. In particular, it makes it possible for eachparty to argue apagogically from the falsity of the alternative to the truth of its own claim. This also enablesKant to vouch for the soundness of each of the proofs, while at the same time claiming that the whole disputeis based on a deep misunderstanding. Admittedly, these proofs remain highly controversial, but I haveendeavored to defend those of the first and third antinomies against the standard objections in Allison (2004),366–84.

32 For my analysis of this argument, see Allison (2004), 388–95.33 For the definitive account of transcendental illusion, see Grier (2001). My own systematic discussion of

the topic, which is greatly indebted to Grier’s but differs on some points, is to be found in Allison (2004),322–32.


higher order empirical object, even though such an “object” can never be given inaccordance with the conditions of possible experience.In view of our previous analysis, this hypostatization can be seen as the direct result

of the inflation of an empirical realism regarding objects of possible experience into atranscendental one encompassing things in general. Although I believe that this con-curs with Kant’s own account, he poses the issue in somewhat different terms.According to Kant, the key to the problem lies in the assumed “givenness” of theconditions. Whereas the participants in the antinomial dispute (who are all transcen-dental realists) assume that the totality of conditions is actually “given” [gegeben]together with any conditioned (in the timeless manner in which the totality of itspremises is given together with the conclusion of a bit of syllogistic reasoning), Kantpoints out that, as spatiotemporal entities or occurrences, these conditions are merely“given as a problem” [aufgegeben] (KrV A498/B526), by which he means that they areaccessible only through the regress or “synthesis” connecting something conditionedwith its conditions. Thus, within this context, it no longer makes sense to speak of anabsolute totality of such conditions, save perhaps as a regulative idea to be approachedasymptotically but never realized.According to Kant, the transcendental realist is blocked from appealing to the gegeben–

aufgegeben distinction because he regards the synthesis connecting something conditionedwith its conditions as one “of themere understanding, which represents things as they arewithout paying attention to whether and how we might achieve acquaintance withthem” (KrV A498/B527).34 In other words, it is precisely because the transcendentalrealist ignores the manner in which these conditions are cognitively accessible that hetacitly assumes that they must be given in their totality in the way in which a complete setof premises is given together with a conclusion. And it is because he assumes this to be thecase that such a realist is unable to resist the pull of the illusory principle that the absolutetotality of conditions is given, even though it is as such not accessible to the human mind.For the transcendental idealist, however, the situation looks rather different. Al-

though it remains natural for such an idealist to think that there must be a totality ofconditions, that is, some ultimate fact of the matter about the duration and extentof the spatiotemporal world, which would be accessible through a “God’s-eye” viewof things, she is also aware that this thought is illusory and is therefore able to avoidsuccumbing to it. Moreover, what makes the latter possible is the realization that thescope of the principles underlying the synthesis connecting a conditioned with its

34 Since by such a synthesis Kant understands one that makes use merely of the pure or unschematizedcategories, in stating that it represents things as they are he is clearly not suggesting that it provides cognitionof things as they are in themselves. Kant’s point is rather that it regards the items synthesized (the conditionedand its conditions) as a collection of objects whose nature is fixed apart from any sensible conditions that maybe necessary for us to access them, that is, as a collection of things considered as existing in themselves. Itshould also be kept in mind that one of the ways in which Kant characterizes the thought of things as they arein themselves is as objects of a “pure understanding.”

80 PART I

conditions is limited to possible experience, which curbs the natural propensity ofreason to extend its reach to the unconditioned.As already noted, unless we view the relation between the two forms of transcen-

dentalism in this way Kant’s indirect argument for transcendental idealism through thenegation of transcendental realism cannot even get off the ground. Nevertheless, it isclear that the negation of transcendental realism does not entail transcendental idealismin any of the metaphysical forms in which it is usually understood, which is why thisargument is so frequently dismissed out of hand, particularly by those who are hostile totranscendental idealism on other grounds.Surely, however, the principle of charity requires us to ask if this dismissal is too

hasty, perhaps based on a failure to appreciate what it is intended to show. Adoptingthis principle, then, let us consider what transcendental idealism must be like, if it wereto be established through the negation of transcendental realism. And when the issue isposed in this way several points become evident. To begin with, it follows that the twoforms of transcendentalism must occupy the same logical space. In other words, iftranscendental realism is an ontological doctrine, transcendental idealism must be onealso. But we have seen that transcendental realism cannot be understood in this way,since it encompasses ontologies of widely different stripes—indeed, if my reading iscorrect, all ontologies. Consequently, transcendental idealism must likewise be under-stood in non-ontological terms.This further suggests that the force of transcendental idealism is deflationary and its

function, particularly in the Dialectic, largely therapeutic. As I have argued above, itprovides the requisite means for resolving the contradiction of reason with itself, which,if unresolved, would mean “the euthanasia of pure reason” (A407/B434). The thera-peutic fruitfulness of transcendental idealism is also illustrated by Kant’s treatment of theproblem of freedom. In its canonical form, the problem is to reconcile the assumptionthat we are transcendentally free agents with the conception of ourselves as causallyconditioned parts of nature. On the traditional readings, this is supposedly accomplishedby a division of metaphysical labor licensed by the resolution of the third antinomy.Depending on one’s version of this idealism, Kant’s position is taken to be either that thephenomenal self is causally determined and the noumenal self transcendentally free (thetwo-object view) or, alternatively, that there is a single self, which is determined quaphenomenon and free qua noumenon (the two-aspect view).It seems clear, however, that neither of these proposed solutions is satisfactory. In

addition to inheriting the notorious difficulties associated with the two-object view,the first appears to commit Kant to an incoherent doctrine of two selves and to yieldcounter-intuitive results regarding the assignment of moral responsibility. As LewisWhite Beck pithily put it, “We assume the freedom of the noumenal man, butwe hang the phenomenal man.”35 And while the second has textual support, it is

35 Beck (1987), 42–3.


commonly thought to be problematic because it apparently requires the ascription ofincompatible properties to the self. As Terence Irwin points out, if an action is causallydetermined qua phenomenal occurrence, then it is causally determined tout court, eventhough considering it from another point of view (noumenally) may involve abstract-ing from or bracketing the causal conditions of the action.36

In order to see how the present interpretation of transcendental idealism provides apossible way out of this morass, we need only revisit the considerations operative in theresolution of the mathematical antinomies. In particular, we must keep in mind thatthe whole dispute rested on the shared metaphysical assumption that there is some factof the matter at issue. Even though Kant characterizes the third and fourth antinomiesas dynamical rather than mathematical and insists that for this reason the claims of thethesis and antithesis may both be true (KrV A529–32/B557–60), I believe thatessentially the same analysis is applicable here.37 In other words, just as the seeminglynatural assumption that there must be some fact of the matter regarding the age and sizeof the world turned out to rest on an illusion, so the equally natural assumption that thequestion of whether we are transcendentally free is one of deep metaphysical fact islikewise illusory. Moreover, if my analysis is correct, falling victim to this illusion is theunavoidable consequence of viewing both the problem itself and Kant’s proposedsolution to it through transcendentally realistic spectacles, which is precisely whatKant’s critics (and many of his defenders) generally do.My claim, then, is not only that transcendental idealism is not committed to the

assumption that there must be some noumenal fact of the matter regarding freedom,but that it functions therapeutically to disabuse us of any such assumption. But whereasin the case of the mathematical antinomies this therapy served merely to block thethreat of a radical skepticism, here it serves also to preserve a place for a conception offreedom that is adequate to the conception of ourselves as rational agents. Basically,what Kant finds necessary is a warrant to assert our freedom from “the practical point ofview,” which is quite distinct from a justification of the metaphysical thesis that we are“noumenally free.” Accordingly, the problem is to explain how we can be warrantedto assert something from the practical point of view that we are explicitly prohibitedfrom asserting theoretically. And, assuming that this requires preserving both some-thing like the Kantian conception of freedom and the normativity of the principles ofunderstanding, the only way in which this could be accomplished is by limiting thescope of the latter. In other words, it is done by deflating a transcendental to a merely

36 See Irwin (1984), 38.37 A prima facie objection to this suggestion is posed by Kant’s contention that in the case of the dynamical

antinomies both thesis and antithesis may be true; for while it is evident that if both thesis and antithesis arefalse there can be no fact of the matter, it might seem that there must be one if both can be true. My responseis that this depends on one’s understanding of “true.” For if, as I suggest below, one understands this in thelanguage of Dummett, Putnam, and others as a “warranted assertibility from a point of view,” then the claimthat both thesis and antithesis can be true need not entail the assumption that there is some metaphysical factof the matter. I discuss the applicability of such a view to Kant in Allison (2004), 48.

82 PART I

empirical realism, which, once again, is just what transcendental idealism does. Withthis deflation in place, it becomes possible to view both the theoretical and the practicalpoints of view as having their own set of norms, while avoiding the assumption thatthere must be some context-independent truth or fact of the matter.

. . .

I conclude with an observation regarding the endeavor of some contemporary Kantiananti-idealists to divorce Kant’s empirical realism from the allegedly disreputable andunnecessary baggage of transcendental idealism. What these philosophers fail to realizeis that this leaves them with a transcendental realism, which should give them reason topause before attempting to separate the substantive results of the Critique from Kant’sidealism. Moreover, I believe that, apart from being held captive by a misguidedpicture of the latter, the reason why this point is so often overlooked is that it isassumed that the real burden of the Critique is to establish a robust empirical realism inthe face of a radical skeptical challenge.38

In calling attention to this prevalent tendency, I am not suggesting that Kant wasunconcerned with external world skepticism—indeed, he claimed it to be a “scandal ofphilosophy and universal human reason that the existence of outer things must betaken merely on faith” (KrV Bxxxix). My point is rather that this was far from the mainconcern of the Analytic, not to mention the Critique as a whole. On the contrary,I think it reasonably clear that the central line of argument of the Analytic proceeds fromrather than to an empirical realism and that it has two primary goals: to determine the apriori conditions of the possibility of experience, which are also conditions of thepossibility of such a realism, and to show that this realism is merely empirical, which isthe fundamental tenet of transcendental idealism.39 In short, to be a transcendentalidealist is to be an empirical realist and vice versa, which suggests that the real questionis why should one want to be anything more?

38 This is especially true of Guyer, who views the Refutation or, more precisely, a version of it containedin Kant’s late Reflexionen, as the culmination of Kant’s transcendental theory of experience. See Guyer (1983)and (1987), 279–329.

39 At least with regard to the direction of Kant’s argument in the Analytic I am in agreement withAmeriks. See his (1978), 273–87 and (2000) 55–63.


PART II

Essay Five

We Can Act Only Under theIdea of Freedom

In his address to the Pacific Division last year, Barry Stroud called attention both to theubiquity of appeals to naturalism in the contemporary philosophical arena and the lackof anything approaching a consensus concerning the meaning of the term.1 As henoted, the varieties of naturalism run the gamut from highly restrictive forms, such asphysicalism and naturalized epistemology, which are exceedingly controversial, tomore inclusive forms, which preserve their popularity by avoiding a commitment toanything recognizably “naturalistic” beyond the rejection of appeals to the super-natural. In the case of the latter forms, Stroud remarks sensibly, the term “naturalism” isoften reduced to an empty slogan, which could more appropriately be rendered by“open-mindedness.”2

There is, however, at least one central area of philosophical inquiry where natural-ism is not only alive and well, but has a fairly determinate sense, namely, the question offree will or agency. Despite an ongoing debate around the edges of this topic, for sometime the ruling orthodoxy has been a form of compatibilism that is naturalistic in thesense that it dismisses any account of agency that is not positively related to theframework of nomological explanation operative in the natural sciences. For upholdersof this point of view, action descriptions may have their own language (that of reasons)and need not be reducible to physicalistic or neurophysiological accounts, but at theend of the day these descriptions must be mappable onto the causal order of nature,which, in contemporary terms, is usually thought to involve either token-tokenidentity or supervenience.3 Moreover, there appears to be much in favor of thisapproach, since human beings are parts of nature and their intentional actions can beregarded as events in the natural order, even if, considered as actions, they are takenunder different descriptions.

1 Stroud (1996), 43–55.2 Ibid., 54.3 I have characterized the naturalistic approach to free agency in deliberately broad terms in order to

encompass the array of non-reductive views, including Davidson’s anomalous monism, which is sometimesappealed to in order to model what is termed “Kant’s compatibilism.” For a recent statement of that view,based largely on the prior work of Ralf Meerbote, see Hudson (1994). I respond to this reading of Kant inAllison (1990), 76–82, and (1996c).

Nevertheless, not all philosophers have been willing to accept this naturalizedconception of agency, which, in one form or another, has been with us at least sincethe seventeenth century. One who did not is Immanuel Kant, and it is his views thatI shall discuss tonight. My main text will be Kant’s famous remark in the Groundworkthat “to every rational being having a will we must necessarily lend the idea of freedomalso, under which alone he acts” (GMS 4: 448). In his metaphysical lectures Kantmakes essentially the same point by claiming that “Freedom is a mere idea and to actaccording to this idea is what it means to be free in the practical sense,” to which headds that “Freedom . . . is practically necessary—man must therefore act according to anidea of freedom, and he cannot act otherwise” (Met M 29: 898).

IIt has seemed to some that this claim, or something like it, can be given a respectablenaturalistic sense. In fact, no less a naturalist than Daniel Dennett has helped himself tothe phrase “Acting Under the Idea of Freedom” as the title for a central chapter of hisinfluential book on the free will problem.4 Dennett’s main concern in this chapter iswith the activity of deliberation and with the threat to the reasonableness of engagingin this activity supposedly posed by determinism. He has little difficulty disposing ofthis threat, showing that it derives its apparent plausibility from the conflation ofdeterminism with fatalism. And given this, he argues that all that is required fordeliberation is a certain “epistemic openness,” meaning thereby the deliberator’s igno-rance about the outcome of a proposed action, which is quite compatible with theassumption that this outcome is causally determined. Consequently, the dictum thatwe can act only under the idea of freedom becomes the proposition that we mustdeliberate as if our futures were open, something which, from the agent’s perspective, isnecessarily the case, even in a deterministic universe.In order to appreciate fully Dennett’s naturalistic gloss on the Kantian dictum, it is

useful to consider it in light of his well-known discussion of the “intentional stance,”which he contrasts with the “physical,” “design,” and “personal” stances.5 A stance forDennett is essentially a predictive, explanatory strategy, and the intentional stance isone that treats the behavior of the “system” being investigated as rational (though notperfectly rational), that is, as operating on the basis of beliefs and desires. But since thechoice of a stance is a purely pragmatic matter, governed largely by considerations ofexplanatory expediency, the adoption of the intentional stance need not ascribe actualbeliefs, desires, or conscious reasoning processes to the “system” in question. It sufficesthat its behavior can be more readily explained and predicted by regarding it as if it had

4 Dennett (1984).5 Although Dennett discusses this conception in many places, my account is based primarily on his (1973)

and (1987).

88 PART I I

these; consequently, such a stance may be taken toward animals, computers, thermo-stats and the like, as well as human beings.6

Although it plays a relatively minor role in Dennett’s own accounts, for presentpurposes the most interesting feature of the intentional stance is its first-personapplication.7 Echoing Hampshire, Sellars, and others, he admits that this stance isineliminable from the first-person standpoint (“I cannot help but have a picture ofmyself as an Intentional system”8), and he infers from this the impossibility of thecomplete elimination of the intentional, which, he is careful to point out, does notpreclude the possibility of a universal mechanistic theory. And, finally, he remarks that“As an Intentional system I have an epistemic horizon that keeps my own future as anIntentional system indeterminate.”9 Thus, once again, it is the indeterminateness thatpertains inescapably to a finite intentional system’s knowledge of itself that is essentialto Dennett’s understanding of freedom.While Dennett does not explicitly make any such claim, he certainly gives the

impression he believes that this conception of an intentional stance, together with theassociated analysis of “acting under the idea of freedom,” expresses in a familiarnaturalistic idiom all that is of philosophical significance in the Kantian dictum. But,despite its initial plausibility, at least two features of this account give one pause. One isthe obvious tension between the affirmed ineliminability of the intentional stance fromthe first-person point of view and the insistence on its purely heuristic status. Withregard to oneself at least, the “choice” of stance does not appear to be a merelypragmatic matter and it is certainly not a matter of facilitating prediction. Instead, itseems that one is rationally necessitated to adopt some such stance toward oneself, evenif one endeavors to explain one’s own behavior mechanistically; for the act of expla-nation cannot itself be understood mechanistically, at least not when taken under thatdescription.The other is the adequacy of the notion of an intentional system to characterize our

conception of ourselves as rational agents. Setting aside first-person worries, we cangrant that some such conception may be adequate to govern our stance toward otherpersons, at least insofar as our goal is to predict and manipulate their behavior as meansto the attainment of our own ends. But, as Dennett himself recognizes, this does notexhaust our relation to others; and this leads him to introduce a distinct personal stance,which presupposes but goes beyond the intentional stance in that it encompasses theexplicitly moral dimension of our relation to others.10

6 See, for example, Dennett (1973), 162–5.7 The reason why first-person considerations do not loom large in Dennett’s discussion is clear from his

explicit commitment to what he terms “the third-person, materialistic perspective of contemporary science.”See Dennett (1987), 7.

8 Dennett (1973), 181. Although Dennett is here characterizing the view of D. M. MacKay, it is clearthat he endorses it. See also Dennett (1987), 27.

9 Dennett (1987), 182.10 Ibid., 165–7.

WE CAN ACT ONLY UNDER THE IDEA OF FREEDOM 89

Given this, it might appear that the correct question to ask is the adequacy of thepersonal rather than the intentional stance. But apart from the matter of the first-personapplicability of this stance (a topic about which to my knowledge Dennett has nothingto say), two considerations call into question its adequacy as well. First, since it too isjust a stance, its adoption remains a matter of choice rather than of fact or rationalnecessity.11 Second, although Dennett does connect it with the idea of moral commit-ment, his discussion suggests that it functions largely in the context of moral appraisal,that is, the assignment of responsibility.12 Accordingly, it does not seem adequate toground the conception of persons as moral claimants inherently worthy of respectrather than merely as potential objects of love or hatred, praise or blame, in short, thereactive attitudes in Strawson’s sense.13 And, assuming that the former conception isworth preserving, the case for a fresh consideration of the Kantian texts appearscompelling, if only to determine whether they might still contain some philosophicallessons that have been overlooked in the rush to naturalize.

IILet us begin by considering Kant’s claim that freedom is an idea, that is, a product ofreason that can never be encountered in possible experience. Otherwise expressed, it isa thought that we bring to the conception of ourselves, insofar as we conceive ourselvesas agents, not a fact that we might discover about ourselves through introspection orempirical inquiry. This denial of factual status is clearly something that Dennett wouldapprove, and so, too, is the purely regulative function that Kant assigns to ideas ofreason in general. Much like Dennettian stances, Kantian ideas regulate inquiry byprojecting a framework that determines the kind of explanation to be sought ratherthan what will be encountered in the course of experience.Nevertheless, Kantian ideas, unlike Dennettian stances, are not merely heuristic.

Although they function regulatively, as products of reason, they express conceptualnecessities. Moreover, the idea of freedom differs from the other Kantian transcenden-tal ideas in that its function is practical and therefore has no empirical use. What itregulates is the non-empirical conception of ourselves as rational agents, that is, ascenters of thought and action, rather than the empirical investigation of a domain ofinquiry. And, for this very reason, it is also constitutive (and therefore again hardlyoptional), not of some putative object—the self—but of one’s conception of oneself as anagent. That is to say, freedom is not simply a property that we may attribute toourselves as rational agents on heuristic grounds; it is rather the defining feature ofthis very conception.If the above indicates that conceiving oneself in accordance with the idea of freedom

is something more than assuming a stance, the content of this idea suggests that it

11 Dennett (1987), 167. 12 Ibid., 173–80. 13 See Strawson (1962).

90 PART I I

cannot be captured by the notion of an intentional system. Central to the Kantianconception is the thought of the spontaneity of the subject and the inseparability of thisthought from a subject’s exercise of its rational capacities, which is precisely the featurethat has been emphasized by John McDowell.14 In the Sellarsian language adopted byMcDowell, the conceptual sphere is characterized as the “space of reasons” and to placesomething in this space is to subject it to normative requirements of justification asopposed to causal explanation. Freedom, or, more properly in this epistemic context,spontaneity, enters the picture because movement within the space of reasons (judg-ment and inference) is governed by normative principles accepted by the subject ratherthan by the mere reception of sensory data. Appealing to such principles must bethought as something that the subject does for itself and thus spontaneously, rather thanas something that occurs independently of its self-conscious activity. And this stronglysuggests that placing a subject (including oneself) in the space of reasons involvesconsiderably more than assuming an intentional stance toward that subject, which, asalready noted, may be assumed toward sub-human systems.In my own work, I have attempted to express the epistemic role of spontaneity in a

more explicitly Kantian manner by analyzing judgment as an activity of “taking as.”The basic idea is that sensibility may suffice to present the mind with x’s that are F’s, butthat one cannot take an x as F (in Kant’s terms recognize it in a concept) withoutactively taking it as such. Moreover, as I have also emphasized, this activity requires thatin some sense one “knows what one is doing,” and that this peculiar mode of cognitiveself-awareness is precisely what Kant means by “apperception,” which is just a con-sciousness of spontaneity.

IIIAlthough this focus on epistemic spontaneity serves to underscore the differencebetween regarding oneself as an intentional system in Dennett’s sense and as a cognizerin the Kantian sense, it obviously is not sufficient to understand and evaluate Kant’sclaim that we can act only under the Idea of freedom. Nevertheless, it does put us inposition to deal with this issue, once we factor in the unity of reason. In other words,I am suggesting that Kant’s contention that it is one and the same reason that differsonly in its application provides the key to understanding his conception of freedom.15

The deep issue for Kant is not whether reason in its practical function is spontaneous(this is entailed by the thought that it is reason), but rather whether pure reason has apractical function, that is, is capable of determining the will. Since in the GroundworkKant identifies will with practical reason, this is equivalent to the question of whetherbeings such as ourselves with the capacity to think, also have wills.

14 McDowell (1994), esp. 3–22. 15 See Allison (1990), 36–7 and (1996a), 102.


An interesting, but frequently overlooked feature of Kant’s position is that he regardsthis question as unanswerable by theoretical means. In fact, he seems to have enter-tained, at least as an epistemic possibility, the idea that we might be moved by instinctrather than reason, which would render the belief in freedom illusory. Expressed inDennettian terms, this amounts to the view that our apparent practical rationalitymight be ultimately tropistic or “sphexish.”16 But while Kant does not preclude such astate of affairs on theoretical grounds, he does deny the practical possibility of acceptingany such thesis on the grounds that it is not a thought on which one can deliberate oract. To take oneself as a rational agent is to assume that one’s reason has a practicalapplication or, equivalently, that one has a will. Moreover, one cannot assume thiswithout already presupposing the idea of freedom, which is why one can act, or takeoneself to act, only under this idea. It constitutes, as it were, the form of the thought ofoneself as a rational agent.Much of what Kant has to say about freedom follows directly from this. As rational

agents, we operate within the space of practical reasons and these reasons must be subjectto objective norms (ultimately the moral law), which can alone justify action on thebasis of them. Thus, from the Kantian perspective, the necessity of acting under thisidea has an essentially normative force, and is therefore quite distinct from anyassumption regarding the openness of the future. As Kant puts it in an importantfootnote in the Groundwork, the point is that, “the same laws hold for a being thatcannot act otherwise than under the idea of its own freedom as would bind a being thatwas actually free” (GMS 4:448n).17

The normative force of this idea can be brought out by considering the practicalanalogue of the “myth of the given.” As McDowell notes, one important form of this“myth” maintains that sensory content plays a foundational role in cognition byproviding an epistemic link between concepts and the world. This assumption isdeemed a myth because the given can play no such role, since of itself it cannotprovide a (justifying) reason for belief. At best, it can enter into a causal explanation thatprovides an excuse for an erroneous belief. The epistemological problem explored byMcDowell is to explain how, in light of this result, it is possible to avoid a radicalcoherentism, such as he attributes to Davidson, which would undermine the thoughtthat experience places a rational constraint on cognition. The solution proffered is thatthe space of reasons extends to the sensory, since conceptual abilities (spontaneity) areinextricably involved in receptivity and therefore in experience. In a word, “theconceptual goes all the way down.”

16 Tropism or “sphexishness” is a recurring theme in Dennett, with his emphasis on imperfect rationality.See, for example, (1973) 169–72, and (1984), 10–13, 18, 22–5, 29, 31, 34–6, 38, 43, 46–9, 64, 70, 87.

17 Although Kant`s formulation is admittedly ambiguous, I take his point to be that the validity of theselaws for a being acting under the idea of freedom is not dependent on the condition that this freedom can beestablished theoretically.

92 PART I I

What is of interest here, however, is not this account of cognition, which hasdubious Kantian credentials in any event, but its practical analogue, which doesrepresent Kant’s view. The analogy I have in mind is as follows: just as no bare sensoryimpression constitutes a reason to believe, so no inclination or desire of itself constitutesa reason to act. It can become such only insofar as it is subsumed under some practicalprinciple that rationally endorses action on its basis. As Kant puts it in the canonicalformulation of this claim (which I have labeled the “incorporation thesis”):

[F]reedom of the power of choice [Willkür] has the characteristic, entirely peculiar to it, that itcannot be determined to action through any incentive except so far as the human being hasincorporated it into his maxim (has made it into the universal rule for himself, according to whichhe wills to conduct himself); only in this way can an incentive, whatever it may be, coexist withthe absolute spontaneity of the power of choice of freedom). (RGV 6: 24)

Here we find the precise practical analogue of the cognitive “taking as;” while therejected assumption that an incentive might determine the will of itself, independentlyof a spontaneous act of “incorporation,” is just the practical version of the myth of thegiven. And, continuing the analogy, it likewise follows that “justification goes all theway down.” In other words, one cannot appeal to a desire, no matter how strong, injustification of a course of action, and, indeed, for essentially the same reason as in thecognitive case: to do so is to remove oneself from the space of reasons; it is to take one’sbehavior as caused (and therefore as perhaps excused), but not as rationally justified. Inthat case, one is no longer acting under the idea of freedom, because one is no longeracting. One is a patient rather than an agent, or at least that is how one takes oneself.It should be obvious that this is not an empirical thesis. Kant is no more committed

to the proposition that I must be able to catch myself incorporating incentives intomaxims, than he is committed to the view that I must be explicitly aware of myself inevery act of thinking. In both cases, it is a matter of the necessity of a possibility. Thus,just as “It must be possible for the ‘I think’ to accompany all my representations” (KrVB131), for otherwise I could not represent anything through them, that is, take them asproviding reasons to believe, so, too, it must be possible for the “I take” to accompanyall my [potential] incentives, for otherwise I could not regard them as providing reasonsto act.

IVAlthough this takes us well beyond anything that might be teased out of Dennett’snotion of an intentional system and the corresponding stance, it still does not addressthe broader issue of whether this essentially normative Kantian conception can itself benaturalized. And this seems to turn on the question of whether the governing ideaof spontaneity can be mapped unto the framework of nomological explanation.McDowell, for one, thinks that it cannot, and he thus rejects the attempt to “domesti-cate spontaneity” by bringing it within the realm of law. He attributes the latter


procedure to what he terms “bald naturalism,” which he characterizes as a style ofthinking that would simply dismiss any “fuss over spontaneity.”18 At the same time,however, McDowell nicely illustrates Stroud’s thesis regarding the seductive charms ofnaturalism by insisting that none of this is incompatible with a “naturalism of secondnature,” which he also characterizes as a “naturalized Platonism.”19 Thus, despite hisinsistence on spontaneity, McDowell feels compelled to march under the banner ofnaturalism, even while claiming, in the spirit of “Wittgensteinian quietism,” to bedissolving rather than solving traditional philosophical problems.20

The inspiration for this extended naturalism is explicitly Aristotelian and it has twoaspects. One is the uncontroversial claim that the spontaneity in question is attributed to anormal human organism.21 The other is the idea that human beings are brought into thespace of reasons, and thereby enabled to actualize their capacities for spontaneity, by aprocess of Bildung.22 It is this entirely natural process that produces a veritable secondnature by initiating individuals into the realm of rational norms. Moreover, on McDo-well’s view, it is precisely because Kant lacked a “pregnant” or “serious” concept of asecond nature that he spoiled his deep insight regarding spontaneity by embedding it in atranscendental story, according towhich a supersensible subject is affected by an unknow-able supersensible reality. This spoils the insight because it leads to a subjective idealism inwhich empirical thought is no longer subject to any rational constraint from reality.33

Lacking such a concept, while recognizing the inadequacy of a bald naturalism, Kantallegedly saw no alternative except his disastrous transcendental story.23

The first point to note here is the extremely narrow reading of Kant that this analysisreflects. If one considers Kant’s essays in the philosophy of history, the Anthropology, andthe thirdCritique, all of whichMcDowell conveniently ignores, it is far from obvious thatKant lacks any such concept. To the contrary, in these writings he provides a rich,nuanced picture of human reason as a “naturalized,” historically conditioned faculty,which he attempts to integrate into the “critical” framework in the appendix to theCritique of the Teleological Power of Judgment. Granted, this amounts to little more than asuggestive sketch when compared to the grand Hegelian account, which McDowellevidently prefers, but the latter would hardly have been possible without the former.The real problem with McDowell’s Kant critique, however, is the interpretation of

transcendental idealism on which it rests, an interpretation which he seems to havetaken over whole cloth from Strawson. Having written extensively on this topic overthe years, I find that I have not much that is new to add at present.24 Nevertheless,something must be said; and, in the spirit of the occasion, I would like to suggest thatthe Strawsonian and similar dismissive treatments of Kant’s idealism suffer from an

18 McDowell (1994), 72. 19 Ibid., 85–6, 91–2, 109–10, 178.20 Ibid., esp. 86–7, 92–5. 21 Ibid., 84.22 Ibid., 41–4. 23 Ibid., 96–104.24 I have provided a full account of my more recent views on the topic in Allison (2004) and in (2006b),

which is included in this volume.

94 PART I I

acceptance of what might be called the “myth of the noumenal.”25 This myth consistsin the assumption that Kant is committed to the doctrine that the “real” is equated witha supersensible (and hence noumenal) realm, of which we can know nothing save thatit somehow affects us and that the cognition of appearances of which we are capable ismerely a pale substitute for genuine knowledge. And if this were indeed Kant’sdoctrine, who could blame McDowell (and many before him) for endeavoring torescue what may be of value in Kant from contamination by it?The mythical nature of this view stems from its complete failure to appreciate the

“critical” nature of Kant’s idealism. This idealism is critical in a two-fold sense. First, itis based on a reflective analysis of the conditions and boundaries of human cognition(the very idea underlying the original project of a “critique of pure reason”). Second, itemerges as a form of idealism primarily through the demonstration of the inadequacy,indeed, incoherence, of the opposing transcendental realism. Expressed in Hegelianterms, Kant’s idealism may be regarded as the “determinate negation” of such realism.Kant’s position on this issue is epitomized in the claim that, “Were we to yield to theillusion of transcendental realism, neither nature nor freedom would remain” (KrVA543/B571).This suggests a strategy of approaching transcendental idealism indirectly, by way of

a consideration of the transcendental realism to which it is opposed.26 The problemwith such a move, however, is that Kant tends to treat these two forms of transcen-dentalism as mutually interdefinable, making it a tricky matter to characterize tran-scendental realism in a way that might help one attain, rather than already presuppose,an understanding of the contrasting idealism. Nevertheless, the task is not hopeless; forit is at least clear that transcendental realism cannot be equated with any of the familiarforms of realism, since it includes phenomenalisms such as those of Berkeley andHume within its scope. For this reason, I have argued that it is best construed as ametaphilosophical or methodological “standpoint” (albeit one that is seldom madeexplicit) rather than a substantive metaphysical or epistemological doctrine. Moreover,this entails that transcendental idealism, must have the same status.Given the above, what I would now like to suggest is that transcendental realism

may be characterized in methodological terms as the view that the “real world” (thepurported object of genuine cognition) is something “noumenal” or “supersensible” inthe sense that the truth conditions of claims regarding it are to be determinedcompletely independently of any consideration of the subjective conditions ofhuman cognition (the forms of sensibility and the schematized categories). By contrast,transcendental idealism insists that these conditions must be “factored in” in order todetermine the conditions of empirical truth and that this, in turn, necessitates thedistinction between things regarded as subject to these conditions and the thought ofthe same things apart from them. In short, it affirms the necessity of drawing a

25 For a general critique of McDowell’s Kant-interpretation along similar lines, see Bird (1996).26 I have adopted this strategy in a number of places, most notably Allison (2004) and (2006a).


transcendental (not an ontological) distinction between things as they appear (subject tothe conditions of human cognition) and the same things as they are in themselves(thought independently of these conditions). Moreover, this distinction provides thebasis for the “critical,” demythologized conception of the noumenon, which is just thatof a limiting concept, “the function of which is to curb the pretensions of sensibility”(KrV A255/B311).Although the above characterization of transcendental realism as a form of nou-

menalism may seem tendentious, it is merely the consequence of taking seriously itsmethodological status. By rejecting in principle any determining role for necessarysubjective conditions of empirical truth, transcendental realism is committed to thenormativity for human cognition of a “God’s-eye” view of things. Moreover, if suchrealism is to avoid skepticism, it must assume the possibility of at least a partial access tothings so considered, which is just what transcendental idealism denies. But insofar astranscendental realism is committed to this, it is likewise committed to a form of themyth of the given, since it is then assuming that what can alone satisfy the truthconditions of human knowledge must be something to which the mind has accessindependently of its own conditions of cognition. And, if McDowell wishes to denythis consequence, as he surely must, it is difficult to see how he can do so withoutappealing, at least tacitly, to a form of idealism.27

VAfter this brief but necessary digression, let us return to our topic. To begin with, giventhe nature of the transcendental distinction and the correlative conception of anoumenon as a limiting concept, we should no longer be tempted to take Kant asclaiming either that we really (noumenally) are free and only appear (phenomenally) tobe causally determined or that we really are causally determined mechanisms, butnonetheless (for heuristic reasons) may think of ourselves as free. Both readings fall preyto the myth of the noumenal. The first, which accords with McDowell’s interpreta-tion, is admittedly suggested by much of Kant’s language; but it erroneously ontolo-gizes the transcendental distinction, thereby elevating a contrast between two ways ofconsidering things into one between two modes of being (real and apparent). Thesecond is suggested by Kant’s notorious “as if ” terminology in his discussion of theideas, and would probably be viewed sympathetically by domesticators of spontaneitysuch as Dennett; but it also inflates Kant’s empirical realism into the transcendentalvariety by granting supreme ontological status to the phenomenal world. And, perhapsmore importantly, by regarding the idea of freedom as a fiction, capable of only aheuristic justification, it deprives it of its normative force and practical necessity.

27 McDowell’s idealism is noted by Friedman (1995), 105.

96 PART I I

The task is therefore to preserve the normative force and practical necessity of this ideawithout making freedom into a transcendentally real property of a noumenal being.Moreover, the key to accomplishing this clearly lies in the above-mentioned limitingfunction of the noumenon; for it is precisely by curbing the pretensions of sensibility toencompass everything conceivable (as opposed to everything cognizable) that the conceptcreates the conceptual space necessary for the thought of a genuinely spontaneous agency.Perhaps Kant’s clearest expression of this view is contained in a passage in the

Groundwork, in which he states that “The concept of a world of understanding is . . .only a standpoint [Standpunkt] that reason sees itself constrained to take outside appear-ances in order to think of itself as practical” (GMS 4: 458). Even though the term“freedom” is not mentioned, the concept is certainly at work, since we have seenthat the idea of freedom is inseparable from the thought of reason as practical.Moreover, it is clear that, here at least, the intelligible world (or world of understand-ing) to which freedom is assigned is neither a transcendent metaphysical domain nor amerely heuristically adopted stance (ein Standpunkt is not a stance). It is rather aperspective that is rationally (not merely pragmatically) necessitated by the assumptionthat one’s reason is practical. In fact, the intelligible world is here precisely whatMcDowell apparently would like it to be, namely, the Kantian equivalent of thespace of reasons. It is a normative and conceptual, not an ontological space.It is also in this purely normative and conceptual sense that the “standpoint” is “outside

appearances.” Otherwise expressed, it refers not to a distinct, non-natural realm of truebeing, but rather to a conceptual space in which reason frames a model of agency that isnecessary in order to conceive itself as practical. And it is precisely for this reason that it isalso “outside nature” or the naturalistic standpoint. To be sure, by appealing to a process ofBildung, a “naturalism of second nature,” one may explain how we human beings havecome to occupy such a conceptual space and to conceive our agency in terms of the ideaof freedom. Nevertheless, the fact remains that this conception cannot itself be broughtwithin the naturalistic framework of explanation, at least not if this framework isunderstood to involve anything more determinate than open-mindedness.Still, at this point one can almost hear Barry Stroud and others retort that even open-

mindedness has its limits. The complaint is that, while the preceding analysis may help tocorrect certain misunderstandings of transcendental idealism, it in no way addresses thedeep objections against idealism überhaupt. In particular, the analysis of cognition asnecessarily involving subjective conditions (epistemic conditions) and a spontaneous“taking as” by the subject seems to fall prey to the common charges that idealism(whether transcendental or otherwise) either confuses knowingwithmaking or, equallydisastrously, conflates how we must think about the world with how it must be.28

Although I do not think that either objection applies to Kant’s view, I cannot argue forthat here. Instead, I shall close by simply asserting two claims to which I am deeply

28 On this latter point, see Stroud (1994), 231–51.


committed. First, as McDowell suggests, it is necessary to reject the domestication ofspontaneity in order to preserve the conception of ourselves as cognizers and agents.Second, and here I part company with McDowell, I believe that this is possible only onthe basis of transcendental idealism; for only such an idealism can ground the right to theconceptual space that we have come to occupy through a process of Bildung.29

29 I wish to thank Wayne Martin for his extremely helpful comments on an earlier version of this address.

98 PART I I

Essay Six

On the Very Idea of aPropensity to Evil

Perhaps nothing Kant wrote has proven as shocking to his contemporaries and asperplexing to present-day readers as his account of radical evil in Religion within theBoundaries of Mere Reason. The very idea that there could be such a thing as a propensityto evil in human nature proved shocking to his contemporaries because of its sugges-tion of the doctrine of original sin, which is the feature of Christian orthodoxy that wasmost inimical to the views of the Enlightenment. In our times, the puzzlement derivespartly from the apparent metaphysical excesses of Kant’s account, particularly the ideaof a timeless choice of disposition, and partly from the ethical rigorism underlying hisanalysis, which seems offensive to modern sensibilities.Despite its language, I do not believe that Kant’s account need be understood in a

metaphysically objectionable way. Moreover, I have argued that it constitutes adeepening of, rather than a break with, the moral psychology of the Groundwork andthe second Critique.1 Recently, however, Allen Wood has challenged this reading,offering a fresh anthropological interpretation, which, by equating radical evil withunsociable sociability, avoids both the metaphysical and moral-psychological problemsassociated with traditional readings, while closely linking the conception to centralthemes in the social philosophy of the Enlightenment. And, in the course of hisanalysis, he dismisses my earlier treatment of the problem as a misguided and unprom-ising trivialization of the Kantian conception.2

What follows is an attempt to respond to Wood’s critique and, more generally, to apurely anthropological interpretation of Kant’s conception of radical evil. AlthoughI acknowledge the presence of an important anthropological dimension in Kant’saccount, I deny that is the whole story. More specifically, I maintain that this accountis grounded in a conceptual analysis of the conditions of imputation and that it yieldsa deeper understanding of evil than can be provided by a naturalistic approach. Thediscussion is divided into four parts.3 The first provides an analysis of the concept of amoral propensity in general and a propensity to evil in particular. The second examines

1 See Allison (1990), 146–61. 2 Wood (1999), esp. 287 and 402.3 This is a development of the response to Wood’s “anthropoligization” of evil offered in Allison (2001c),

esp. 605–10.

the nature and grounds of Kant’s claim that this propensity is innate in human natureand therefore universal. The third responds to two objections to this claim and attemptsto show how this propensity is related to the anthropological conception of unsociablesociability, without, as Woods holds, being reducible to it. Finally, the last part offers abrief assessment of the significance of the concept of a propensity to evil for Kant’sconception of virtue.

IPart One of Religion within the Boundaries of Mere Reason contains an extended reflectionon the concept of moral evil and the conditions of its possibility in beings possessed ofboth a rational and a sensuous nature. This reflection may also be described as aconceptual analysis based on the underlying assumption that, “The human beingmust make or have made himself into whatever he is or should become in a moralsense, good or evil” (RGV 6: 44). Accordingly, freedom of the will is presupposed andis taken to entail that the locus of both moral goodness and evil lies in the will.Although Kant locates the ultimate ground of evil in a propensity that is somehow

both freely chosen and innate in our species, it should not be thought that he is offeringthis as a metaphysical explanation. In fact, he explicitly denies the possibility of any suchexplanation (RGV 6: 25 and 74). Instead, I think it best to read Kant as engaged in a kindof thought experiment, the aim of which is to spell out just what we are committed to, ifwe take seriously the idea that evil is to be imputed. So construed, Kant’s positing of aninscrutable propensity as the ground of evil remains within the parameters of an analysisof the concept of evil as a product of freedom. More precisely, it marks the limit of suchan analysis, the point beyond which there is nothing more to be said.The twin foundations of Kant’s conceptual analysis are his ethical rigorism and

his account of the human predispositions. Rigorism, as Kant here presents it,involves an uncompromisingly bivalent view of the moral life, according towhich every imputable act and morally responsible agent must be characterizedas either good or evil. This is contrasted with latitudinarianism, which allows for amoral middle ground in the case of both acts and agents (RGV 6: 22–4). Kantadmits that experience suggests latitudinarianism, but he mounts an a prioriargument for rigorism, which turns on two key points: the claim that respect forthe law serves as an incentive for every agent to whom acts are imputable, andwhat I call the incorporation thesis.4 The former entails that a choice contrary tothe moral law involves a positive resistance to it. The latter indicates that theresistance takes the form of the incorporation into our maxim of a deviantincentive. Together they entail that any act that fails to conform to the law ispositively evil rather than merely lacking in goodness or moral worth. And since this

4 See Allison (1990), 5–6 and passim; (1996a), 118–23, 130–5, 139–42; and (2006b), which is included inthis volume.

100 PART I I

failure is, ex hypothesi, an expression of freedom, the agent must likewise be regardedas evil.Even though rigorism entails that an act or agent is evil simply in virtue of not being

good, it is not to be confused with the Platonic-Enlightenment understanding of evil asignorance or simple lack of goodness. On the contrary, in Religion within the Boundariesof Mere Reason, Kant understands evil as involving something positive, in the sense of aposition, an active resistance to, or a turning away from, the moral law, which isprecisely what makes it both imputable and evil. Accordingly, despite his rationalism,Kant’s conception of evil is in some ways closer to that of the later Augustine than tostandard views of the Enlightenment, not to mention the intuitions of many present-day philosophers, including numerous Kantians.In order to understand Kant’s account, we must begin with his conception of the

predispositions or Anlagen of a human being, which he defines as “the constituent partsrequired for it as well as the forms of their combination that make for such a being”(RGV 6: 28). Simply put, they are constitutive features of human nature or essentialcapacities as they relate to the faculty of desire, without which a fully human existence isimpossible. According to Kant, there are three such predispositions: (1) the predisposi-tion to animality, which concerns a human being as a living being; (2) the predispositionto humanity, which concerns a human being as a rational animal whose reason is in theservice of inclination, and whose supreme principle is therefore self-love; and (3) thepredisposition to personality, which concerns a human being as an accountable moralagent, a being for whom reason is practical of itself (RGV 6: 26–28). The predispositionto personality is just the capacity to be motivated by respect for the law and it is in virtueof this capacity that we are moral agents, capable of either good or evil.Kant characterizes all three predispositions as predispositions to good and, somewhat

paradoxically, it is in virtue of this status that they turn out to be essential for under-standing the possibility of evil. Kant suggests that they are predispositions to good notmerely in the negative sense that they do not as such resist the moral law, but also in thepositive sense that “they are predispositions to the good (they demand compliance withit)” (RGV 6: 28). By this Kant means simply that these predispositions are the source ofduties, since we have duties both to ourselves and others qua sensuous beings andrational animals, as well as qua autonomous moral agents (see MS 6: 421–73).For present purposes, however, what is important about these predispositions is the

fact that they rule out two possible ways of construing evil and point to a third andmore adequate alternative. To begin with, they rule out both the familiar view that thesource of evil lies in our sensuous nature and the more exotic possibility that it lies in a“corruption of the morally legislative reason,” a kind of diabolical revolt against the goodas such (RGV 6: 35). The first is precluded by the inclusion of our animal nature and, inparticular, our inclinations in the predisposition to the good; though it is also crucialfor Kant that, as parts of our given biological nature, sensuous inclinations cannotbe imputed. The exclusion of the second is a more complex and controversial matter,but the main point is that such a corruption or perversion is incompatible with the

ON THE VERY IDEA OF A PROPENSITY TO EVIL 101

predisposition to personality. Since the moral law would then not function as anincentive, an agent with such a will would be a non-person in the moral sense, anunfortunate product of nature rather than an accountable being.5

This leaves the predisposition to humanity, our nature qua rational animal, whosegoverning principle is self-love. Inasmuch as this is likewise part of the general predis-position to good, it cannot be viewed as inherently evil. But since, unlike the purelysensuous side of our nature, it is connected with a use of reason and therefore freedom, itis subject to a misuse or perversion that does generate evil. Given the incorporationthesis, this perversionmust take the form of the adoption of an evil maxim. And since, asa condition of being either good or evil, an agent adopting this maxim must acknow-ledge (at least implicitly) the moral law as an incentive, its evilness can consist only in animproper ranking of the incentives. Instead of subordinating the claims of self-love tothose of morality, an evil maxim reverses this order, letting the principle of self-lovewear the trousers, so to speak, and assigning merely a subordinate role to moralconsiderations. Unlike a putative diabolical agent, however, a morally evil agent doesnot repudiate the moral law as such, but takes it as yielding a sufficient reason to act, orrefrain from acting, only on the condition that this does not conflict with self-love.The obvious problem here lies in the claim that there is a propensity in human

nature to evil, so understood. But this must be seen as the specification of an evendeeper problem with the very concept of a moral propensity whether to good or evil.A propensity orHang, much like a predisposition, does not seem to be the kind of thingfor which an agent may be faulted. Moreover, this concern is hardly assuaged by eitherKant’s characterization of a propensity as “the subjective ground of the possibility of aninclination . . . insofar as this possibility is contingent for humanity in general” or his useas an illustration the supposed propensity of savage people to intoxicants (RGV 6: 28).Clearly, what is needed in order to carve out conceptual space for a moral propensity

is the establishment of a possible connection between a propensity and freedom. Kant’sproposed solution to this conceptual problem is itself appropriately conceptual innature. It turns on the introduction of the concept of an intelligible act [That],which refers to the act of choice [Willkür], through which an agent’s suprememaxim is adopted (RGV 6: 36). The basic idea is that we are to conceive of theordinary, first-order maxims of rational agents as governed by a single fundamental orsupreme maxim, which applies to the whole use of freedom. Following Philip Quinn,we may call this a “meta-maxim” in virtue of its second-order status.6 Construed indispositional terms as a deliberative tendency to rank in a certain way the basicincentives of morality and self-love in their incorporation into first-order maxims,the meta-maxim may be viewed as a propensity. Since evil has already been located inthe subordination within a maxim of moral requirements to those of self-love, itfollows that by a propensity to evil must be understood an agent’s meta-maxim to

5 This is noted by Wood (1970), 210–15. 6 See Quinn (1988), 110.

102 PART I I

order the incentives in just this way in the adoption of its first-order maxims. More-over, as Kant suggests, the meta-maxim is itself formal in nature, since it concerns theform of an agent’s first-order maxims, or the ranking of incentives, rather than theirspecific contents (RGV 6: 36).Although Kant characterizes this purely intelligible act through which the meta-

maxim is chosen in dauntingly metaphysical terms, suggesting timeless noumenalagency, it need not be taken in this way. The point is merely that in order to givecoherence to an agent’s moral life, we must view the agent’s actions as expressions ofsuch a maxim, which determines the moral orientation of the will. Unlike ordinary,first-order maxims, however, the meta-maxim or propensity cannot be thought of asself-consciously adopted at a particular point in time. On the contrary, it is found to bealready at work when moral deliberation begins and must be presupposed in order toconceive the possibility of immoral actions in beings for whom the moral law providesan incentive. It is in this sense alone that it is to be viewed as timeless and intelligible.7

IIThe preceding considerations spell out and provide conceptual space for the concept ofa moral propensity and, a fortiori, for a propensity to evil. They do not, however, sufficeto show why Kant thought it justifiable to attribute a propensity to evil to humanbeings and to claim that it is entwined so deeply in humanity as to be characterizable asinnate (RGV 6: 3). Kant’s underlying rigorism seems to require merely that a personhave a propensity either to good or evil, since a mixed propensity is incoherent andsomeone lacking either propensity would not be a moral agent at all, in which case themoral law would not serve as an incentive. But this appears to leave as alternativepossibilities both a universal propensity to good and, more plausibly, a scenario inwhich some have a propensity to good and some to evil.Presumably, the concept of a propensity (as contrasted with a predisposition) to

good is ruled out by a combination of the preceding analysis and an appeal toexperience. Since evil first-order maxims presuppose an evil meta-maxim or propensi-ty to evil and universal human experience provides abundant evidence of evil, therejection of a universal propensity to good is justified on empirical grounds. Moreover,though he does not pose the problem in these terms, it may seem that Kant thoughtthat experience also rules out the second alternative, for he claims notoriously that,“We can spare ourselves the formal proof that there must be such a corrupt propensityrooted in the human being, in view of the multitude of woeful examples that theexperience of human deeds parades before us” (RGV 6: 32–33). In other words, theuniversality of evil revealed by experience appears to rule out the view that some

7 I discuss this point in Allison (1990), 47–53, 138–45, and 153–4.


individuals have a propensity to good, thereby leaving us with a universal propensity toevil as the default position.Nevertheless, inasmuch as the kind of universality that this anthropological survey

provides is incapable of showing the necessity of conceiving human nature as essentiallycorrupt, I believe it reasonable to take Kant’s appeal to anthropological experience as arhetorical ploy, which is designed to enable him to bracket the deeper questions posedby the problem. Moreover, Kant does not deny the possibility of a formal proof; hemerely says that the evidence spares him the necessity of providing one. This at leastleaves open the possibility that Kant thought he had such a proof and makes theattempt to reconstruct it on the basis of materials provided by the text worthwhile,albeit somewhat speculative.By characterizing the missing proof as “formal,” Kant presumably intended to

indicate its a priori nature. As such, it must proceed on the basis of concepts involvedin the analysis of the concept of evil and the propensity thereto. Furthermore, given theunderlying rigorism, it follows that any proof of the universality in human nature ofthe propensity to evil must be at the same time a denial for the same human natureof the possibility of a propensity to good and vice versa. Accordingly, the best availablestrategy for such a proof is to establish the incompatibility of a propensity to good withhuman nature. The claim is not that the concept of a propensity to good is itself self-contradictory, but that it conflicts with an invariable feature of human nature or, as wemight also put it, a condition of human agency.The first step is to determine what could be understood by a propensity to good.

Although Kant never explicitly characterizes such a propensity, rigorism requires that itbe the contrary to the propensity to evil, namely, the meta-maxim or dispositionautomatically to subordinate the incentive of self-love to the moral incentive. Assuch, it would consist in a spontaneous preference for the impersonal requirements ofmorality over an agent’s need as a sensuous being. For such an agent there could be nopossibility of temptation and no thought of the law as constraining, which, in turn,means that the law would not be viewed as an imperative and its requirements as duties.It seems clear, however, that Kant thought human nature incapable of such a

disposition, since it would mean that the will is beyond the need for rational constraintand thus beyond the twin thoughts of duty and respect for the law. This does not meanthat Kant thought it impossible to subordinate self-love to duty, since this is preciselywhat the moral law requires. It is only that we cannot do so spontaneously, without, asit were, giving self-love a hearing. But if this is beyond the capacity of human natureand we must attribute to it a moral propensity of some sort, it follows on a rigorist viewthat we must attribute to it the contrary propensity to evil.In order to complete the “deduction,” it remains necessary to explain why a propensity

to good is supposedly incompatible with, or beyond the capacity of, human nature, sinceit cannot be a simple consequence of having a sensuous nature. The answer lies in therecognition of the status of happiness as a necessary end for human beings and, presum-ably, for any finite rational beings conceivable by us. Once again, it is not that this

104 PART I I

precludes choosing duty over inclinationwhen they conflict; it is rather that we cannot doso spontaneously, without first considering the claims of happiness. As a necessary end,happiness must be accorded its rights, which entails that there will always be resistance tothe demands of morality when they infringe upon them.

IIIAs presented so far, the conception of an innate propensity to evil seems open to at leasttwo major objections. The first, which is not so much a problem for the presentinterpretation as it is for the Kantian theory itself, concerns the compatibility of such apropensity with the presupposed freedom of choice. The problem initially arose inconnection with the concept of a moral propensity, which, ex hypothesi, must beassumed to be imputable. As we saw, Kant’s solution was to introduce the conceptof an intelligible act, through which such a propensity, understood as a meta-maxim, isadopted. Kant further suggested that the appeal to this foundational act was not to beviewed as an attempted metaphysical explanation, but rather as the introduction of alimiting concept, an indication of the itself inexplicable presupposition or postulate atwhich an analysis of the concept of evil must terminate. The problem in its presentform is distinct from this and more directly related to familiar issues regarding freedomof choice. Simply put, by making an agent’s fundamental choice of the evil meta-maxim unavoidable, as is required by the strong universality claim, it appears that thefreedom supposedly underlying the whole analysis of evil is denied at the mostimportant point. This need not be a problem on certain compatibilist views, accordingto which freedom does not require the capacity to do otherwise, but this hardly seemsto apply to the Kantian conception.Although an adequate response to this problem would require a detailed consider-

ation of Kant’s theory of freedom, the central point is simply that Kant characterizesfreedom in terms of a causality of reason rather than a general capacity to do otherwise.This enables him to regard a holy will as free, even though it has no such capacity.More importantly, in the case of a finite will, for which the moral law presents itself asboth incentive and imperative, freedom is understood positively as the capacity to doour duty in the face of the strongest contrary incentives. It is only within the context ofa conflict between duty and inclination that Kant insists on a capacity to do otherwise,understood as doing what the moral law requires under given circumstances. More-over, it is only in light of this that we can understand Kant’s insistence that, “had notthe moral law already been distinctly thought in our reason, we should never considerourselves justified in assuming such a thing as freedom” (KpV 5: 4n).It follows from this that our inability to adopt an originally good meta-maxim, as

opposed to a subsequent virtuous disposition, would only be incompatible withfreedom, if there were a duty to have such a propensity. But, from what we havealready seen, this would amount to a duty to be beyond duty, which is clearly absurd.Consequently, even though the dice may be loaded and make the result unavoidable,


our adoption of the evil meta-maxim is still rightfully imputed to us. The adoption ofthis maxim, like its theological analogue, original sin, must be regarded (though notexplained) as resulting from an originary use of freedom.The second problem, which seems to be exacerbated by the reading sketched above,

concerns what we might term the evil of radical evil. Even though it follows fromKant’s rigorism that a bare propensity to subordinate moral considerations to thosestemming from self-love counts as evil, it seems to be so only in a minimal, almosttrivial sense. Such a criticism is irresistible if, as is still sometimes done, we assume thatradical evil is intended to designate extraordinary evil.8 But it might still arise, even afterrecognizing that Kant uses the term “radical” in the etymological sense to indicate theroot of all moral evil. Indeed, this may be part of what lies behind the trivialityobjection raised by Wood, who is well aware of what Kant meant by the term.9

What must therefore be shown in order to answer the triviality objection is not somuch that radical evil is evil, as that it is radical and that the propensity to evil, itselfinexplicable and inextirpable, must be regarded as the root of the more palpable formsof evil encountered in experience. But given the apparently vast gulf between the mereranking of self-love over the moral law and the truly horrendous acts that haveoccurred throughout recorded history, this is far from obvious. Indeed, it seemsdifficult not to side with John Silber on this issue, when he criticizes Kant for a failureto acknowledge diabolical, apparently selfless forms of evil, which phenomenologicallyspeaking, appear to have little to do with any such ranking.10

The Kantian response to this challenge, insofar as one can be gleaned from the text,consists of two parts. The first is the brief discussion of the three levels or grades of thepropensity to evil: weakness or frailty, impurity, and wickedness, which are depicted asthree successive stages in the development of this propensity (RGV 6: 29–30). Ofparticular interest here is the significant role that Kant gives to self-deception, whichenables evil agents to avoid having to admit their evil to themselves.11 Largely becauseof its focus on such self-deception, this account occupies an important, if not suffi-ciently appreciated, place in Kant’s moral psychology. In particular, it shows that Kant’sconception of evil is far more nuanced than his underlying rigorism might suggest.Nevertheless, this cannot be the whole story, since it is limited to an immanentdevelopment of the propensity to evil, which even in its final stage of wickednessneed not express itself in manifestly evil or vicious acts.The second part of the Kantian response, which has been emphasized by Sharon

Anderson-Gold and Wood, appeals to unsociable sociability as the key social factorunderlying all manifest forms of evil.12 The basic idea, which is clearly inspired byRousseau and which Kant sketches in his brief discussion of the predisposition to

8 I discuss this issue in Allison (1996a), 169–82. 9 See Wood (1970), 284.10 See Silber (1960).11 I discuss this issue in Allison (1990), 158–61 and passim; and (1996a), 173–5 and 178–81.12 See Gold (1984) and Wood (1970), 286–96.

106 PART I I

humanity, is that because of our social nature self-love inevitably assumes a compara-tive form. Thus, we tend to judge ourselves happy or unhappy only in comparisonwith others, which leads us, in turn, to a desire to acquire worth in the opinion ofothers. Kant states that this is originally merely a desire for equality, but that within thecompetitive social context in which everyone else has similar aims, it graduallybecomes transformed into the craving for superiority, which leads inevitably tojealousy and rivalry. These then give rise to the so-called “vices of culture,” whichare only possible in a cultured or civilized state and themselves serve as further spurs toculture. Finally, at their most extreme, these become the “diabolical vices,” of envy,ingratitude and spitefulness, or Schadenfreude, which are the highest forms of evil forKant (RGV 6: 27).Given this analysis, which amounts to a socialization of evil, it seems possible to

show how the bare propensity to evil can account for extreme evil as a culturalphenomenon, without assuming anything like an inherently diabolical will of thesort insisted on by Silber. Evil is rooted in self-love insofar as it refuses to limit itselfby the moral law, even though because of our unsocial sociability it takes forms in itsadvanced stages that are hardly commensurate with the pursuit of self-interest, asordinarily understood.Admittedly, it remains an open question whether even this enriched account of evil,

which Kant himself does little more than hint at in Religion within the Boundaries of MereReason, is really adequate to deal with evils such as the Holocaust. But setting aside thatlarge, perhaps unanswerable question, there remains on the table a far more prosaic,though nonetheless important, exegetical issue: How are we to understand the connec-tion between the original propensity to evil, which has been our central concern allalong, and the unsociable sociability that has now emerged on the scene? This questionhas been directly addressed by Wood, who offers a strongly reductionist reading,according to which radical evil is explicitly identified with unsociable sociability. Bydoing so, he makes such evil into an entirely empirical-anthropological concept,thereby leaving no room, or need, for anything like the conceptual analysis offered here.Although this reading is certainly suggestive and provides an appealing way to avoid

some of the previously noted difficulties in Kant’s account, it is not tenable. In fact,Wood’s error is the direct opposite of the error of philosophers who construe Kant’saccount as an attempt at some kind of metaphysical explanation of the genesis of evilthrough a mysterious atemporal choice. It is as if seeing the utter implausibility of anysuch metaphysical account, Wood opted for a reductive naturalistic-empiricism as theonly viable option.I believe that there is another alternative, however, one that preserves a distinction

between what for Kant is the purely intelligible propensity to evil and the properlyanthropological trait of unsociable sociability. The propensity pertains to the domain ofpure moral philosophy and the anthropological trait to that of the empirical knowledgeof human nature. At the risk of being charged with the vice of attempting to explainthe obscurum per obscurius, we might think of the difference in terms of the distinction


that Kant draws in the first Critique between the empirical and the intelligible characterof the will.13

According to Kant’s account, the former is the object of an anthropological investi-gation of human behavior and is the primary factor to which appeal is made in thecausal explanation of intensional action. As such, it is deterministic, though notphysicalistic, since it gives a prime role to psychological factors. The intelligiblecharacter is thought as the “transcendental ground” or cause of the empirical character;but it is not an explanatory ground, since all explanation of intentional action for Kantmust be in terms of the empirical character. Rather, it may be described as the“imputational ground,” in the sense that it is that to which appeal is made, if oneendeavors to impute the empirical character to the agent whose character it is. I thinkit clear that unsociable sociability is the generic empirical character of human beingsfor Kant and I think it also clear that Kant thought that this character should beimputed. This, then, is the function of the propensity of evil in its intelligible chara-cter (the propensity to subordinate the moral incentive to that of self-love). As such,it is what makes it possible to regard our unsociable sociability as something tobe imputed, instead of viewing it purely naturalistically, as either a socially or bio-logically conditioned trait, perhaps an unfortunate byproduct of our evolutionarydevelopment.

IVIn addition to its function as the ground of imputation and connection with unsociablesociability, the concept of a propensity to evil also plays a crucial role in Kant’s accountof moral development and his conception of virtue. As Kant puts it, even though the“thesis of innate evil” has no role in “moral dogmatics” since it is not the source of anydistinct duties, it is significant in “moral discipline,” for it entails that:

We cannot start out in the ethical training of our conatural moral predisposition to the good withan innocence which is natural to us but must rather begin from the presupposition of a depravityof our power of choice in adopting maxims contrary to the original ethical predisposition; and,since the propensity to this [depravity] is inextirpable, with unremitting counteraction against it.(RGV 6: 50–51)

I have argued elsewhere that this passage shows that the doctrine of radical evil, in theform of an original propensity to evil, not only defines our moral condition but also sets

13 This is a substantial change from the original version of this paper, where I attempted to distinguishbetween the two levels of analysis in terms of the contrast between the “pure” and the “schematized”concept of the propensity to evil, with unsociable sociability being identified as the latter. I now believe thatKant’s contrast between the two conceptions of character is much more germane to the task at hand,particularly since both unsociable sociability and the propensity to subordinate the moral incentive that ofself-love may be described as “character traits” for Kant, which rather than being in conflict with each otherapply at different levels of analysis. I discuss the empirical intelligible-character contrast in Allison (1990),29–53 and (1998b), which is included in this volume.

108 PART I I

the moral agenda for imperfect beings such as us.14 This agenda consists in thecontinual striving for an unobtainable ideal of holiness, in the course of which wemay attain virtue, understood as a kind of autocracy or self-constraint grounded inmoral principles.Admittedly, however, this agenda has appeared to many as both unduly demanding

and deeply puzzling. Must we really strive in apparently Sisyphean fashion after anunattainable ideal of holiness merely in order to become virtuous? Indeed, it seemsdifficult to avoid hearing Kant’s pietistic parents speaking here, rather than the authen-tic voice of pure reason. Moreover, our perplexity only increases, once it is realized thatthe first item in this agenda, and the condition of even the pursuit of holiness, mustbe a radical break with the propensity to evil, which Kant characterizes as requiringnothing less than a revolution in our cast of mind, or Denkungsart, and describes bymeans of the biblical images of “rebirth” and “new creation” (RGV 6: 47). But how,we may ask, is such a revolution possible, if, as Kant assumes, our disposition is alreadycorrupted? Surely, at the very least, this would require a bootstrap operation ofmonumental proportions. Unfortunately, Kant himself is of little help on the matter,since he candidly admits that such a revolution is inexplicable, while also insisting that itmust be possible, since we ought to undertake it (RGV 6: 44–48).Although much of this remains deeply obscure and problematic, it embodies a moral

intuition that cannot be easily dismissed. Reduced to its simplest terms, it is theseemingly paradoxical idea that the moral life essentially involves a struggle of freedomwith itself, that, morally speaking, each of us is our own worst enemy. It is this thatrequires that the first step on the road to virtue be a radical break with an agent’s pastself, a dispositional change that can only be adequately described in quasi-religiousterms such as conversion. Anything less, say a resolution to try harder next time, wouldamount to a continued capitulation to evil.Finally, it is also this idea that underlies Kant’s conception of radical evil. If his

conception of unsociable sociability links him with Rousseau, this idea places himsquarely within the tradition that reaches roughly from Augustine to Kierkegaard.Moreover, despite its obscurity at key points, it gives to Kant’s account of moral life adepth and seriousness that is lacking in the popular contemporary neo-Aristotelianaccounts of virtue as human flourishing. Failing to flourish, for whatever reason, iscertainly not a good thing; but it is also distinct from being evil. Accordingly, I think itfair to say any account of virtue that minimizes this difference is not only un-Kantian,but seriously deficient in its own terms.

14 See Allison (1990), 162.


Essay Seven

Kant’s Practical Justificationof Freedom

Practical justification in Kant comes in two distinct forms, which reflect the differentsorts of proposition that require justification. One concerns the justification of practicalpropositions and encompasses both the categorical imperative, as the fundamentalprinciple of morality, and particular categorical imperatives such as “Never lie.”Practical justification in this sense might also be termed “justification of the practical”and is a task that is generally viewed as required of any moral theory. Kant attempteda justification of the categorical imperative as the fundamental principle of moralityin the third section of GMS in the form of a deduction that is loosely modeled onthe transcendental deduction of the categories in the first Critique. In the secondCritique this project is abandoned and replaced with the claim that the categoricalimperative (or moral law) requires no deduction, since it is validated by the conscious-ness of the moral law as a “fact of reason.” By contrast, the Metaphysics of Morals,Kant’s last systematic work in moral theory, is devoted largely to the justification ofparticular duties.The second type of practical justification concerns certain theoretical propositions.

For Kant these are propositions regarding God, immortality, and freedom, whichin the second Critique he refers to as “postulates of pure practical reason” becausemorality requires that they be presupposed, even though they are not susceptible of atheoretical justification. It is this type of practical justification with which I shallbe concerned here. Or, more precisely, I shall be concerned with one aspect of suchjustification.The situation is complicated, however, in two ways. The first is by a distinction that

Kant draws between the kind of practical justification that he finds available for thepostulation of God and immortality, on the one hand, and freedom, on the other.Although in the Dialectic of the second Critique Kant includes freedom along withGod and immortality among the postulates of pure practical reason (KpV 5: 132–3), hedoes not treat it merely as a postulate in that work, since in the Analytic he offers adeduction of freedom from the moral law as a fact of reason (KpV 5: 47–50). Thus,while God and immortality are postulated as necessary conditions of the highest goodon the dual grounds that we have a duty to promote the highest good and the familiar“ought implies can” principle, Kant holds that freedom “in the strictest, that is, the

transcendental sense,” is (again by the ought implies can principle) a necessary condi-tion of morality in general (KpV 5: 29).1

The second complication concerns the difference in the way in which Kant endeavorsto provide a practical justification of freedom in various texts, particularly in the Ground-work and the second Critique. In the former work Kant links the justification of freedomwith a reflection on the consciousness of being a rational agent, that is, of having thecapacity to act in accordance with one’s representation of laws or on principles, withoutany special reference to moral laws or principles. In some of his later writings, however,most notably the second Critique and Religion, Kant maintains that it is only our con-sciousness of standing under the categorical imperative (as “the fact of reason”) that assuresus, from the practical point of view, of our freedom. In the first two parts of this essay I dealbriefly with each of these attempts at a practical justification of freedom, while in the thirdI offer some reflections on the metaphysical issues they pose.

IEarly inGroundwork 3 Kant remarks that “to every rational being possessed of a will wemust also lend the idea of freedom as the only one on which he can act” (GMS 4:448).2 In his lectures on metaphysics from about that time, Kant makes essentially thesame point, claiming that, “Freedom is a mere idea and to act according to this idea iswhat it means to be free in the practical sense,” to which he adds: “Freedom ispractically necessary—man must therefore act according to an idea of freedom, andhe cannot act otherwise” (Met-M 29: 898). Our initial task is to uncover and evaluatethe line of argument underlying this sweeping claim.3

I shall begin with a brief look at Kant’s first published statement of this line ofargument in a review of a work by Johann Heinrich Schulz, which appeared in 1783(two years before the Groundwork).4 In response to Schulz’s speculative determinism,which Kant characterizes as fatalism and compares to the doctrines of Priestley, Kant

1 In second Critique Kant defines freedom in the transcendental sense as “independence from everythingempirical and so from nature generally . . . ” (KpV 5: 97). Kant generally contrasts it with the “comparativeconcept of freedom,” by which he understands the Leibnizian conception. It is the latter that he contemptuouslydismisses as “nothing better than the freedom of a turnspit” (KpV 5: 97). In theDialectic of the firstCritiqueKantdefines transcendental freedom as “the power [Vermogen] of beginning a state spontaneously [von Selbst]” (KrVA533/B561). There, and in his lectures on metaphysics, Kant generally contrasts transcendental with practicalfreedom, which he defines in a number of different ways. For my analysis of this issue, see Allison (1990), 54–70.

2 Translations from GMS are my own. Citations from other works of Kant use the translations from TheCambridge Edition of the Works of Immanuel Kant.

3 In Allison (1997a), which is included in this volume, I dealt with this thesis from a more thematic, lesshistorical point of view, arguing against thinkers such as Dennett and McDowell, who endeavor, albeit in quitedifferent ways, to give a naturalistic reading to this claim. By contrast, the present account is more sharply focusedon the Kantian texts.

4 Schlulz’s work, the first part of which likewise appeared in 1783 and was the only part that Kantreviewed, was entitled: Versuch einer Anleitung zur Sittenlehre für alle Menschen, ohne Untersschied der Religion,nebst einem Anhange von den Todesstrafen [Attempt at an Introduction to a Doctrine of Morals for all Human Beings,regardless of Religion. together with an Appendix on Capital Punishment].

KANT’S PRACTICAL JUSTIF ICATION OF FREEDOM 111

makes two main points. The first turns on a sharp contrast between the speculative andpractical points of view. According to this line of thought, when confronted with thequestion of what ought to be done, even the speculative fatalist, for whom the future ispresumably already decided, will unavoidably proceed as if the decision were up tohim, that is, he will “always act as if he were free” (RSV 8:13). The second is a parallelbetween the freedom to think, that is, the idea that one’s judgment is based on objectivegrounds, which the cognizer takes as reasons to believe, rather than being the productof “subjectively determining causes,” and the freedom to choose, which is presumablylikewise based on objective grounds (RSV 8: 14).The restatement of this line of argument inGroundwork 3 is contained in a paragraph

with the heading: “Freedom must be presupposed as a property of the will of all rational beings”(GMS 4: 447–8), which Kant evidently regarded as a preparation for his centralargument rather than as a constituent of it.5 The overall task of Groundwork 3 is toestablish the validity of the moral law or categorical imperative, which has already beenshown to be a synthetic a priori practical proposition. Building on the previouslyestablished connection between the concepts of the categorical imperative and theautonomy of the will, Kant begins the section by arguing for what I have termed the“reciprocity thesis,” that is, the proposition that “a free will and a will under moral lawsare one and the same” (GMS 4: 447).6 Given this thesis, the natural strategy for Kant topursue would be to argue that we are free, from which the claim that we really aresubject to the categorical imperative would follow immediately. But since Kant wasprecluded from adopting such a strategy by the teaching of the first Critique thattheoretical reason is incapable of proving (or disproving) the reality of freedom inthe requisite transcendental sense, he was led to adopt the fallback strategy of arguingfor the necessity of presupposing freedom and contending that for practical purposesthis is equivalent to proving that we really are free.Kant prefaces his account by emphasizing the necessity of establishing the strict

universality of the attribution of freedom. In order to serve the purpose of Kant’sargument, freedommust be attributed to the will of all rational beings or, as he also putsit, to every rational being with a will. This follows from the combination of theunderlying methodological assumption of GMS and the reciprocity thesis. The as-sumption is that inasmuch as morality expresses a law for every rational being as such,its principle “must be bound up (fully a priori) with the concept of the will of a rationalbeing as such” (GMS 4: 426). And since the reciprocity thesis maintains that freedom isthe property of the will from which subjection to moral requirements is derived, itfollows that the legitimization of morality, which for Kant means the categoricalimperative, depends crucially on the premise that freedom is a property of the will ofevery rational being. Although the restriction of the attribution of freedom to rational

5 See Kant, GMS 4: 447.6 I analyze the reciprocity thesis, which also appears in slightly different form in the second Critique, in

Allison (1986), (1990), 201–13, and (2011), 273–300.

112 PART I I

beings with a will may seem trivial, since in order to have a free will it is necessary tohave a will, we shall see that it is crucial to Kant’s argument.Our immediate concern, however, is with Kant’s actual argument, which contains

two steps, each of which is presented by Kant in the form of a bald assertion:

(1) Now I say: Every being that cannot act other than under the idea of freedom, isfor that very reason actually free in a practical respect, that is, all laws that areinseparably bound up with freedom are valid for it, just as if its will had beenshown to be free in itself and in theoretical philosophy. (GMS 4: 448)

(2) Now I assert that to every rational being that has a will we must necessarily lendalso the idea of freedom, under which he acts. (GMS 4: 448)

As Kant makes clear in a footnote, the point of the first step is “to free ourselves fromthe burden that pressures theory” (GMS 4: 448), which is to say that even if thetheoretical question regarding the reality of freedom remains unsettled, the necessity ofappealing to the idea of it suffices to establish the validity of the categorical imperative.In evaluating this assertion, it is crucial to determine what is meant by acting under theidea of freedom and the grounds for the alleged necessity of so acting. To begin with, itseems clear that if we take acting under the idea of freedom to mean something likebelieving that one is free, the argument lacks plausibility; for it appears to be open tothe determinist simply to deny that he really believes that he is free, even if hemight grant that he sets aside his determinist convictions when engaged in practicaldeliberations.But if acting under the idea of freedom does not mean acting with the belief that one

is free, what does it mean? The answer lies in the nature of an idea for Kant, which,simply put, is a normative principle that is a product of reason. Otherwise expressed, itis a thought that one necessarily brings with oneself, when one takes oneself to bedeliberating or acting, not a fact that we might discover about ourselves throughintrospection or in some other manner. In Kantian terms, it is a regulative idea thatgoverns our conception of ourselves as agents. Moreover, as such, it has normativeforce. In Sellarsian language, to act under this idea is to place oneself in the logical spaceof (practical) reasons and therefore to take oneself as subject to rational norms of both amoral and prudential sort.This explanation of Kant’s first claim puts us in a position to understand the second,

which was already articulated in the review of Schulz and which turns on what he theretermed “freedom to think.” Kant does not mean what is usually understood by thatexpression, namely, the freedom (independence from external constraint) to holdwhatever beliefs seem warranted. Indeed, “freedom” is not the appropriate term,since what Kant has in mind is really the spontaneity that for him is an ineliminableingredient in discursive cognition. The basic idea is that the human understanding isspontaneous in the sense that it is not a mere receptacle of sensory data (this is how Kantregards sensibility); rather, it is an active faculty that takes what is sensibly given and

KANT’S PRACTICAL JUST IF ICATION OF FREEDOM 113

brings it under concepts.7 Schematically expressed, the understanding takes [judges] xas F, where the “taking” is an act of spontaneity, something that the subject does, ratherthan, like association, a causal process that occurs in the subject’s mind/brain. As Kantputs it in yet another bald assertion:

Now I assert that one cannot possibly think a reason that in its own consciousness is directed fromoutside with regard to its own judgments; for in that case the subject would attribute thedetermination of his power of judgment to an impulse rather than to reason. (GMS 4: 448)

This thesis, which Kant expressly limits to the first-person point of view (“in its ownconsciousness”), sets the stage for completing his argument for the necessity of lendingthe idea of freedom to the will of every rational being that acts, by extending, as he hadalready done in the Schulz review, the conclusion from reason in its cognitive to itspractical capacity. Thus Kant writes:

It [reason] must regard itself as the author of its principles independently of alien influences;consequently, as practical reason or as the will of a rational being, it must be regarded by itself asfree, that is, the will of a rational being can be a will of its own only under the idea of freedomand must therefore in a practical respect be attributed to all rational beings. (GMS 4: 448)

Kant is here affirming a symmetry between theoretical and practical reason or, moreprecisely, between epistemic and practical spontaneity, both of which are expresslylinked to a first-person point of view and speak to how a (thinking or acting) subject isrationally constrained to regard itself, qua cognizer or agent. Just as one cannot regardoneself as a cognizer without conceiving of oneself as a subject who judges somethingto be the case by bringing sensory data under concepts on the basis of normativeprinciples (principles of the understanding), so one cannot regard oneself as a rationalagent without taking oneself as incorporating the data of volition (incentives) intoone’s maxims (subjective principles of action) under the direction of normative rules(hypothetical and categorical imperatives).8

Even assuming the general framework of Kant’s account of cognition, there arethree problematic features of this line of argument, which may have been recognizedby Kant and help to explain his relegation of this relatively straightforward approach toa merely preparatory status and led him to embark on an extremely complex deduc-tion, which makes extensive use of the concept of an intelligible world and his “two-

7 Elsewhere I characterize this as “epistemic spontaneity” to contrast it with the “practical spontaneity”that Kant attributes to the will. See Allison (1990), 36–8 and passim and (1996a), 129–42.

8 I am here referring to what I term the “incorporation thesis,” which I take to lie at the heart of Kant’sconception of rational agency. In its canonical formulation it states that “freedom of the power of choice[Willkür] has the characteristic, entirely peculiar to it, that it cannot be determined to action through anyincentive except so far as the human being has incorporated it into his maxim (has made it into a universal rule forhimself, according to which he conducts himself ); only in this way can an incentive, whatever it may be,coexist with the absolute spontaneity of the power of choice” (RGV 6: 24). For my discussions of this thesis,see Allison (1990), 5–6, 40, 47–8 151, 189–90; (1996a), 118–23, 130–5, 139–42; and (2011) 112–20 andpassim.

114 PART I I

standpoint” doctrine. The first we may term the “bindingness problem.” By raising thespecter of a “hidden circle” (GMS 4: 453), Kant seems to be suggesting that theargument from the necessity of presupposing freedom is insufficient to account forthe claim that the categorical imperative is really binding on all rational agents, evenwhen strengthened with the thesis that whatever laws are valid for a being whosefreedom can be established theoretically are also valid for one who can act only underthe idea of freedom.9 Although Kant is notoriously unclear at this point, his worryseems to be that the most that the above argument can prove is that if we mustpresuppose that we are free, then (by the reciprocity thesis) we must likewise presup-pose that we are subject to the categorical imperative, which falls short of showing thatwe really are bound by it, as a deduction presumably requires.A second source of possible concern is the assumption of a strict symmetry or

isomorphism between cognition and volition with respect to their spontaneity. Infact, we have evidence from both the first Critique and the Reflexionen that Kant wasagnostic or at least ambivalent regarding the move from theoretical to practicalspontaneity, with the latter denoting a causality through reason.10 Moreover,I believe that Kant can be taken as acknowledging the problematic nature of thismove in Groundwork 3 through the restriction of the scope of his arguments to rationalbeings with wills. This reflects Kant’s recognition of the conceptual possibility ofrational beings without wills, that is, cognizers who are not agents, or, what amountsto the same thing, beings with reason but not practical reason.11 Although one mightthink that Kant could dismiss such a worry as idle, on the grounds that we are consciousof our agency and it can have no effect from the practical point of view, the fact is thathe did not.Finally, an additional problematic feature of Kant’s move from epistemic to practical

spontaneity involves the relationship between two different senses of freedom recog-nized by Kant: spontaneity and autonomy or, as he also describes them, a negative anda positive conception of freedom (GMS 4: 446). By the former Kant understands theopposite of natural necessity, that is, a causality that is effective “independently of aliencauses determining it” (GMS 4: 446). By the latter he understands “the property of thewill of being a law to itself ” (GMS 4: 447). Although Kant says at one point that the

9 For my analysis of the circularity problem, see Allison (1990), 218–21, and (2011), 301–30.10 These texts include KrV A547/B575, A548/B576, and A557/B582, where Kant expresses uncertainty

about whether reason has causality, which I take to be equivalent to the question of whether reason ispractical or whether rational beings have a will; and R 5442: 18: 183, where Kant distinguishes betweenlogical and transcendental freedom and denies the possibility of inferring the latter from the former.

11 Kant entertains such a possibility in the teleological reflections that he interjects in support of his claimthat the good will is the only thing that is good without restriction. The idea is that if reason had been given usonly for the purpose of better enabling us to attain happiness, which is a belief to which those who denyKant’s claim about the good will are supposedly committed, then it would not have chosen the best means,since that could be more easily attained through instinct. For our purposes, what is relevant about this is thatKant held open the possibility of beings who are cognitively rational, but whose choices are governed byinstinct rather than practical reason (see GMS 4: 395–6).


concept of the latter “flows” [fleisst] from that of the former (GMS 4: 446), this is notobviously the case.12 In other words, it seems perfectly possible that a will might be freein the contra-causal sense of not being causally necessitated by antecedent conditionsand yet ineluctably heteronomous in the sense that its menu of incentives (or motives)all stem ultimately from its sensuous nature.13 Moreover, since it is quite clear that thebindingness of the categorical imperative requires autonomy, it follows that, even if itwere successful, the argument from epistemic to practical spontaneity would beinadequate to establish the goal of the deduction. But since analyzing that argumentis not my concern here, I shall abandon GMS at this point and consider Kant’s verydifferent strategy for a practical justification of freedom in the second Critique.14

IIKant’s practical justification of freedom in the secondCritique reverses the strategy adoptedin the Groundwork. In both cases the argument presupposes the reciprocity thesis; butwhereas in theGroundworkKant began with an argument for presupposing freedom that isbased on a general conception of rational agency, rather than on any specifically moralpremises, and proceeds from this to a deduction of the categorical imperative, in the secondCritique he begins by appealing to the notorious “fact of reason,” that is, a presumably self-certifying consciousness of the moral law as supremely authoritative, and proceeds fromthis to a deduction of freedom. As Kant puts it, “the moral law, which itself has noneed of justifying grounds, proves not only the possibility but the actuality [of the power offreedom] in beings who cognize this law as binding upon them” (KpV 5: 47).The obvious problem with this strategy is the seemingly question-begging appeal to

a fact of reason, which Kant characterizes differently in different places and in his morecareful moments describes as a “fact as it were” [gleichsam als ein Faktum].15 Unfortu-nately, both the reasons for what has been termed Kant’s “great reversal” and thecogency of his appeal to the fact of reason are extraordinarily complex and contentiousissues with which I cannot deal adequately with here.16 Nevertheless, I shall saysomething about how I believe this “fact as it were” is best understood and its relevanceto the justification of freedom.I shall begin with the distinction drawn by Lewis White Beck between a “fact for”

and a “fact of pure reason.”17 By the former Beck understands a pre-given value that iscognized by pure reason. On this reading, pure reason is regarded as a theoretical

12 I discuss this issue in Allison (1996a), 137–8. A similar point is made, albeit in different terms by Hill(1992), 93–4, 106–10.

13 For my discussions of this possibility, see Allison (1990), 59–70 and (1996a), 109–14.14 For my analysis of the deduction see Allison (1990), 218–29 and (2011), 331–62.15 In Allison (1990), 321–32, I cite Kant’s eight characterizations of this “fact” in KpV.16 I offer my analysis of the fact of reason in Allison (1990), 230–49. The expression “great reversal” was

used by Karl Ameriks to characterize Kant’s change of approach to the issues of the justification of thecategorical imperative and freedom. See Ameriks (1982), 226.

17 See Beck (1960), 168–70, and (1965).

116 PART I I

capacity or, more precisely, a capacity for intellectual intuition, which is a capacity thatKant denies being possible, not only for humans, but for finite rational beings ingeneral. Thus, in addition to its question-begging nature, which it shares with otherforms of moral intuitionism, such a Platonic view stands in direct contradiction with acentral tenet of Kantian epistemology, which seems more than sufficient to precludetaking it as a serious interpretive option.Beck suggests, however, and I believe him to be correct, that things look consider-

ably different if we take Kant to be referring to the fact of pure reason.18 Beckcharacterizes the latter as the “fact” that pure reason is practical, that is, that it can ofitself determine the will; not, to be sure, in the causal sense, but in the normative sensethat it provides the will with a principle that is objectively necessary and, as such, bindsor obligates finite rational agents.19 Faced, for example, with the Humean thesis that“Reason is, and ought only to be the slave of the passions, and can never pretend to anyother office than to serve and obey them,”20 and its many modern variants, one may betempted to dismiss this Kantian claim as likewise question-begging. Moreover, it couldbe argued that, since the purported goal of the second Critique is to show that purereason is practical, by simply appealing to it as a fact of “fact as it were” Kant is makingthings rather too easy for himself.Tempting as it may appear, however, this dismissal of Kant’s procedure is itself too

easy. To begin with, the appeal to the fact of reason, so understood, reflects Kant’srealization, presumably brought about by the recognition of the failure of the deduc-tion in Groundwork 3, of the impossibility of proving that pure reason is practical by anargument that relies on extra-moral premises. And if this is the case, then arguably theonly viable strategic alternative left is to establish that pure reason shows itself to beactually practical, which is just what Kant claims to have accomplished by the appeal tothe fact of reason.21 In other words, much like the affirmation of the necessity of actingunder the idea of freedom in theGroundwork, the turn to the fact of reason may be seenas a fallback position to which Kant was led by the recognition of the restricted optionsleft open to him by his underlying critical commitments.Kant’s procedure is perhaps best described as phenomenological, where the phe-

nomenon in question is the ordinary moral consciousness. As such, it occupies somecommon ground with Groundwork 1, where Kant was likewise concerned with theordinary moral consciousness or, as he there characterizes it, “common rational moralcognition.” The difference is that whereas there this consciousness was used as a pointof departure for a Socratic-like analysis of what is implicit in it, Kant’s concern here is

18 I say “the fact of ” because Kant refers to it as “the sole fact of pure reason” (KpV 5: 31).19 In GMS Kant repeatedly emphasizes that what is objectively necessary as a prescription of reason is

subjectively contingent for finite beings with a sensuous as well as a rational nature. Accordingly, to claim thatwe are bound by the categorical imperative is not to say that we necessarily obey it.

20 Hume (2000), 266.21 See KpV 5: 42. Kant here characterizes the “fact” through which pure reason shows [beweiset] itself to

be practical as “autonomy in the principle of morality.”


with the express commitments of this consciousness, which amount essentially to therecognition of the supremely authoritative status of a moral principle that fits thegeneral description of the categorical imperative. Perhaps the most phenomenologicalmoment in theGroundwork, however, is Kant’s account of what we find if we attend toourselves in the transgression of duty. As he describes the situation, rather than actuallywilling that our (impermissible) maxim become a universal law, since that is impossible,we will rather that its opposite (the morally permissible maxim) should remain the law,“yet we take the liberty of making an exception of ourselves, or (even only for thisonce) for the advantage of our inclinations” (GMS 4: 424). Otherwise expressed, werecognize the authority of the moral law even in violating it and therefore recognize aneed to justify this violation to ourselves (and perhaps to others). Save perhaps for thecomplete sociopath, I do not think that this is a bad bit of moral phenomenology andI also think that it sheds a good deal of light on how Kant understood the fact of reason,that is, as the consciousness of an authoritative demand that stems from one’s own willrather than from an external source such as the will of God.The major question, of course, is whether we are to regard this putative fact of

reason as a genuine fact of reason, rather than, for example, of the Nietzschean will topower or the Freudian superego. Although I am not sure how Kant would address thischallenge, I believe that his best line of response would be a variant of his critique of theproject of grounding morality in human nature, which is in fact what Nietzsche andFreud did, albeit on the basis of very different conceptions of human nature than Kanthad envisaged. Simply put, the point would be that approaches of this type cannotaccount for the kind of necessity, with which moral requirements address us. In Kantianterms, the latter is an objective necessity, which bespeaks an origin in reason, ascontrasted with a subjective necessity that is to be understood in psychological oranthropological terms, which, expressed in Kantian terms, is what Nietzsche and Freud(among others) have to offer.Be that as it may, the central question for us is the move from this fact of reason to

freedom, which Kant himself describes as “the deduction of freedom as a causality ofpure reason” (KpV 5: 48). Although it is clear that the principle underlying thisdeduction is the ubiquitous “ought implies can,” the issue is made more complex bythe nature of the ought. The crucial point is that the categorical imperative requires notmerely a capacity to do what reason (or duty) requires, independently of, and evencontrary to, one’s needs and interests as a sensuous being, but also (and primarily) acapacity to do what reason requires precisely because it requires it. In the terminology, ofthe Groundwork, it requires not merely that one’s action accord with duty, but that it befrom duty; otherwise, the accordance with duty would be a contingent matter and one’smaxims would lack moral content or worth. And this requires not merely negativefreedom or independence from causal determination in accordance with laws ofnature, but also positive freedom or autonomy, which involves a capacity to determineoneself to action on the basis of principles of reason that make no reference to ourneeds as sensuous beings. Moreover, what the fact of reason supposedly shows is not

118 PART I I

simply the necessity of presupposing autonomy, but its actuality as a capacity to bemotivated by purely by moral considerations, a capacity of which we are consciouseven if it is never exercised.22

Finally, this difference in the conceptions of freedom makes it possible to avoid anapparent contradiction between Kant’s accounts in the Groundwork and in the secondCritique and later works. As we have seen, in the review of Schulz and the GroundworkKant treated freedom as an idea that is presupposed by the conception of oneself as arational agent, quite independently of any specifically moral considerations. Accord-ingly, the claim that we can act only under the idea of freedom is not to be understoodas maintaining that we can act morally only under this idea, but rather that it is onlyunder this idea that we can exercise rational agency in any of its dimensions. Bycontrast, in the second Critique Kant tells us that,

[W]hereas freedom is indeed the ratio essendi of the moral law; the moral law is the ratio cognoscendiof freedom. For, had not the moral law already been distinctly thought in our reason, we shouldnever consider ourselves justified in assuming such a thing as freedom (even though it is not self-contradictory). (KpV 5: 4n)

If Kant meant the same thing by freedom in this and similar texts as he did in GMS,when he insisted that we can act only under the idea of freedom, he would not onlyhave directly contradicted his earlier view, but have affirmed a seemingly implausiblethesis; for no matter where one may stand on the free will question, it does not appearto make much sense to claim that, apart from the recognition of being morallyobligated, we would have no basis for assuming anything like freedom. After all, ifI am or take myself to be a free agent in something like the manner that Kant affirms intheGroundwork and other earlier texts, am I not free in matters of prudence as well as inmatters of morality?23 But if, as I believe to be the case, Kant is referring in the secondCritique and other later texts, where our awareness of freedom is specifically linked toour consciousness of being bound by the moral law, to freedom as autonomy, thenthere is no contradiction. Indeed apart from this consciousness, it is difficult to conceive

22 Kant’s clearest illustration of this point is his oft-cited contrast between someone who asserts that wheninclination and opportunity are present he finds the desire irresistible and someone of whom a princedemands on pain of immediate execution that he testify falsely against an honorable man. Of the first,Kant asks rhetorically whether if a gallows were erected in front of the house where he finds the opportunityto satisfy his lust and he was certain of immediate execution upon satisfying it, he would be able to resist hisinclination. Of the second, Kant remarks, far more circumspectively, that, “He [the one put to the test]would perhaps not venture to assert whether he would do it [sacrifice his life by refusing to testify falsely] ornot, but he must admit without hesitation that it would be possible for him. He judges, therefore, that he cando something because he is aware that he ought to do it and cognizes freedomwithin him, which without themoral law would have remained unknown to him” (KpV 5: 30).

23 For example, in the first Critique Kant remarks with reference to the causality of reason, which heequates with freedom: “Now that this reason has causality, or that we can at least represent something of thesort in it, is clear from the imperatives that we propose as rules to our power of execution in everythingpractical” (KrV A547/B575). I take it that the imperatives to which Kant refers here include the hypotheticalas well as the categorical variety. Similarly the “rules” are both moral and prudential and “everythingpractical” is not limited to moral matters.


how an agent could have an awareness of a capacity to govern itself by pure reason,independently of any empirical interests or desires that the agent might have.

IIIOne of the questions that inevitably arise regarding so-called practical justifications ofostensibly theoretical claims concerns their metaphysical implications. This is especiallytrue for Kant, since much of what he says on the topic suggests that he is offeringsolutions from the resources of practical reason to questions that are necessarily posedby, yet unanswerable for, theoretical reason. In the case of freedom, Kant insists atseveral points in the second Critique that the moral law, through the fact of reason,establishes not merely the (logical) possibility but the actuality of freedom, whichcertainly gives the impression that he regarded his claims as having metaphysicalimport.24 Nevertheless, both my overall interpretation of transcendental idealism asmethodological or epistemological rather than ontological in nature and my under-standing of Kant’s use of this idealism to resolve the Antinomies have made mesuspicious of a metaphysical reading.Before turning to this issue, however, I wish to emphasize that Kant usually qualifies

his claims regarding the moral grounding of freedom (as well as the postulates of Godand immortality) with the caveat that they hold only for practical purposes. AlthoughKant expresses this caveat in a number of ways, I believe that they all come down to thesame thing, namely, that the claims are valid merely from the standpoint of the agentwho is concerned with the question: “What ought I do?”With regard to freedom, thefact of reason purportedly shows that we are really bound by the moral law and, assuch, are beings whose actions are imputable. For Kant, on my view, this is neither atheoretical truth about the nature of our noumenal agency of which we somehowbecome aware through our consciousness of standing under the moral law, nor aheuristic fiction that we adopt when we view ourselves as subject to moral require-ments.25 It is rather a standpoint that we are compelled (by reason) to adopt in virtue ofour consciousness of ourselves as bound by the moral law.26

My interpretation of transcendental idealism has been criticized from a variety ofdirections, but with respect to the question at issue the most germane is that of KarlAmeriks. According to Ameriks, the problem with non-metaphysical versions oftranscendental idealism, such as my own, is that they “give no reason to think that

24 See, for example, Kant, KpV 5: 6, 42, 105, 134, 143.25 The classical formulation of the fictionalist view is by Vaihinger (1935), esp. 271–301. For my criticism

of this view, see Allison (2011), 305–6.26 Kant states that, “The concept of a world of the understanding [Verstandeswelt] is only a standpoint

[Standpunkt], which reason is compelled to take outside of appearances in order to think itself as practical” (GMS4: 458). In my view, Kant argues for essentially the same thesis in KpV, except that reason is not requiredmerely to conceive itself as practical, but also (in view of the fact of reason) as bound by the moral law.

120 PART I I

the non-ideal has greater ontological status than the ideal.”27 Although Ameriks iscorrect in pointing out that on my reading (and perhaps others) the non-ideal has nogreater ontological import than the ideal, I question his further claim that this isincompatible with Kant’s deepest philosophical commitments. On the contrary,I think that the denial of greater ontological import to the non-ideal is in accordwith these commitments and that this is nicely illustrated by Kant’s treatment offreedom.To begin with, let us note the well-known difficulties in which one finds oneself

enmeshed in the endeavor to place Kant’s account of freedom in the ideal–realmetaphysical framework, Setting aside the variety of two object (or world) and two-aspects readings, at the end of the day, one who considers Kant’s account of freedom asa metaphysical thesis is forced to choose between two alternatives: (1) to take Kant’sview to be that we really are free and only seem to be causally determined; or (2) to readhim as maintaining that the noumenal self is free and the phenomenal one causallydetermined. Unfortunately, neither view seems particularly attractive. The formerundermines Kant’s empirical realism and the latter commits him to a seeminglyincoherent doctrine of two selves and to a highly counter-intuitive view of moralresponsibility. As Beck once put the latter point, “We assume the freedom of thenoumenal man, but hang the phenomenal man.”28

This suggests that the fundamental problem confronting any attempt to attribute ametaphysical status to Kant’s conception of freedom is that it tacitly assumes that theremust be some fact of the matter regarding freedom. Moreover, this applies equally tocritics of this conception, who affirm a hard determinism, or the favored contemporaryview of a “soft-determinism” or compatibilism, as well as to Kantian naturalizers whoregard Kantian freedom as something one experiences, that is, a sort of psychological oranthropological fact, which is where one is unavoidably led if one wishes to preserve afactual status for freedom, while rejecting any attempt to locate it in the “noumenalworld.”29

Admittedly, as with all the transcendental questions posed by Kant in the Dialectic,but particularly the Antinomies, it does seem as if there must be a fact of the matter.After all, is it not the case that the will or, if one wishes to avoid “faculty talk,” thehuman being, is either free or not free in whatever sense of “freedom” one wishes toaffirm or deny. Indeed, must not this be the case even if, as Kant seems to suggest, weare incapable of determining by theoretical means which party is correct?

27 Ameriks (1992), 334.28 Beck (1987), 42–3.29 See Guyer (1993). In fairness to Guyer, it must be admitted that on occasion, particularly, but not

exclusively, in pre-critical texts, Kant does speak of a sense of freedom that is empirical. And, of course, in hispolitical writings he is concerned with “outer freedom” or freedom of action. Nevertheless, as a matter ofinterpretation, I find it difficult to see how one can deny that Kant’s deepest thoughts regarding freedomconcern the transcendental dimension of the conception.


As I have argued elsewhere, however, though perfectly natural, this way of viewingthe situation reflects a virulent combination of transcendental illusion and transcen-dental realism.30 Although the illusion is both natural and unavoidable, indeed, it is theinevitable outcome of reason’s engagement in its inherent project of attempting tograsp the whole (the totality of conditions for any conditioned), transcendental realism,which is the counterpart of transcendental idealism, is both avoidable and the source ofthe metaphysical errors that Kant methodically diagnosed in the TranscendentalDialectic.31 Transcendental idealism does not remove transcendental illusion, thatbeing impossible (for example, it still seems natural to assume that there must besome matter of fact regarding freedom); but by discrediting transcendental realism, itmakes it possible to avoid being taken in by this illusion. In this respect at least, thefunction of transcendental idealism is largely therapeutic.32 Consider the problem ofthe age and size of the world, which is the concern of the First Antinomy. It certainlyseems as if the spatiotemporal world must either have a first beginning in time and anouter limit in space or be infinite in both respects, which is to say that there appears tobe a fact of the matter regarding the age and size of the world, independently of thequestion of how we are to determine it.According to Kant’s analysis of the Antinomy, however, this is an illusion, albeit a

natural and highly persistent one that takes a transcendental critique to uncover; for ifone assumes that the world (the totality of appearances) exists in itself, independently ofthe conditions of its cognition, which is the view to which transcendental realism iscommitted, then it follows that this whole must be regarded as being either finite orinfinite in the relevant respects. Moreover, even though Kant characterizes the ThirdAntinomy as dynamical rather than mathematical and claims that this entails that bothsides could be correct, I believe that the same principle is at work, namely, that thesource of the seeming conflict is the assumption that there must be some fact of thematter regarding freedom. Notice, if one continues to make this assumption in the faceof Kant’s official resolution of the Antinomy, which appeals to transcendental idealism,one is led either to endorse an outright contradiction (we are both free and not free atthe same time) or, more plausibly, we find ourselves back in the situation where wemust affirm either that we really (noumenally) are free and only appear (phenomenally)to be causally determined or, equally unattractive, that there is one self that is really freeand another that is causally determined. Notice also, however, that none of theseconsequences arise if we take the point of the resolution to be that there is no fact of thematter regarding freedom, but merely two regulative principles, each with its ownsphere of validity.

30 See Allison (2004), esp. 307–448.31 The definitive account of this issue is by Grier (2001).32 I am not claiming that the function of transcendental idealism is entirely therapeutic, but simply that this

is it major role in the Dialectic. Obviously, it also plays a constructive role in the Aesthetic and Analytic.

122 PART I I

Where, then, does this leave us with regard to Kant’s practical justification offreedom? In my judgment, it is not with any ontological thesis regarding freedom;nor, I might add, does it push us in the direction of the fictionalist view associated withVaihinger, according to which acting under the idea of freedom means acting as if wewere free, even though we know that we are not. As the label suggests, the basicproblem with this view, which has many contemporary variants, is that it too assumesthere to be a fact of the matter, albeit a negative one. Rather, adopting the language ofDummett, Putnam, and others, we end up with what I call a doctrine of “warrantedassertability from a point of view,” where the point of view is practical.33 In otherwords, qua rational agent engaged in a process of deliberation regarding what I ought todo (in either a moral or prudential sense), I must consider myself as free in the negativesense of being independent of determination by natural causes; while qua bound by themoral law I must regard myself as free in the positive sense of possessing autonomy. Theformer is the thesis of the Groundwork and the latter of the second Critique. Togetherthey constitute Kant’s practical justification of freedom, which, once one is liberated bytranscendental idealism from the illusion that there must be some fact of the matterregarding freedom, I believe to be sufficient for “practical purposes.”

33 See Allison (2004), 48 and (2006a), 18.


Essay Eight

The Singleness of the CategoricalImperative

The aim of this essay is to defend, or, more precisely, to lay the foundation for a defenseof Kant’s thesis that the categorical imperative, namely, “Act only according to that maximthrough which you can at the same time will that it become a universal law,” is “only a singleone” (GMS 4: 4216–9). That the singleness thesis stands in need of defense is evidentfrom at least three features of the argument of Groundwork 2. The first is that Kant’sapplications of the various formulas to the different types of duty seem to yield differentresults, which suggests that they cannot be equivalent. The second is Kant’s claim thatthe three designated formulas, namely: FLN, FH, and FA/FRE, are related to threeessential features of maxims, namely, their form (universality), their matter (an end),and their complete determination. The third is Kant’s suggestion that this involves aprogression, which he compares with that in the categories of quantity from unity, toplurality, to allness or totality (GMS 4: 43623–8).

1

Even though Kant also says that “the three ways of representing the principle ofmorality are at bottom only so many formulas of the very same law, one of which ofitself unites the other two in it [deren die eine die anderen zwei von selbst in sich vereignt]”(GMS 4: 436 8–10), and that the difference between them is “subjectively rather than[eher] objectively practical” (GMS 4: 43610–11), it remains difficult to see how theseformulas could be regarded as expressions of a single principle.2 Accordingly, it is

1 I use FLN to represent what is usually referred to as “the formula of the law of nature,” FH for “theformula of humanity” or “the formula of ends in themselves,” FA for “the formula of autonomy,” and FREfor “the formula of a realm [or kingdom] of ends.” I connect the latter with FA because, as will be explainedbelow, I do not regard it as a distinct formula. I render the so-called “formula of universal law” as FUL, eventhough, as will likewise be explained below, I view it as referring to the categorical imperative itself ratherthan to one of its formulas.

2 There is a disagreement amongst the English translators and commentators regarding the proper renderingof each of these claims, which have a significant bearing on the interpretation of Kant’s position. The first of theseconcerns the clause: “deren die eine die anderen zwei von selbst in sich vereignt.” Grammatically, it can be renderedeither as “one of which” or “any one of which” [unites the other two in itself]. Of the seven English translationsI have consulted (Abbott, Beck, Ellington, Gregor, Paton, Wood, and Zweig), four of them (Abbott, Beck,Gregor, andZweig) take Kant to be affirming the latter; two (Ellington and Paton) preserve the ambiguity;whileWood renders it as unambiguously claiming the former. Speaking in favor of the latter or broader reading is thefact that it appears to be more in accord with Kant’s fundamental claim that all of the formulas are of “the verysame law;”while the fact that Kant claims that they involve a progression analogous to that within the categoriesof quantity from unity to plurality to allness or totality (GMS 4: 43626–9) seems to require the former. For this

tempting to agree with Allen Wood, when he dismisses any claim of a strongequivalence between the formulas and suggests instead that the later formulas are“less inadequate” than the earlier ones and that theGroundwork’s search for the supremeprinciple of morality should “be seen as ending only when we grasp the moral law as asystem of formulas.”3

At the same time, however, it is also evident that the singleness thesis is a centralrather than a peripheral feature of Kant’s moral theory. Perhaps Kant’s clearest state-ment of this is in his lectures on ethics, where, in criticizing Baumgarten for providing anumber of moral principles, Kant is cited as stating that, “Where there are already manyprinciples in ethics, there is certainly none, for there can be only one true principle”(MPC 27: 266; 59).4 And this leads us in the direction of Onora O’Neill, who, in sharpcontrast to Wood, insists that, “If the claim of equivalence [of the various formulas]cannot be sustained, the argument of the Groundwork, and more generally of Kant’sethics, is deeply disappointing.”5

We thus find ourselves pulled in opposite directions both by the Kantian texts andby two of the leading contemporary commentators on these texts. Moreover, howeveruncomfortable this impasse may make us feel, it is only appropriate; for the texts areambiguous, and I believe that each of these commentators makes an importantcontribution to the debate. Accordingly, I see my task as incorporating the elementof truth in their accounts, while rejecting what I take to be their errors. To anticipate,I believe that the truth in Wood’s reading lies in his emphasis on the progressive,developmental nature of Kant’s account of the categorical imperative and his error inconcluding that this precludes any strong thesis regarding its singleness. Conversely,I take the truth in O’Neill’s position to be her insistence on the importance for Kant’sethical thought of the equivalence of the formulas and its errors in her neglect of FLNand insufficient attention to the progressive nature of Kant’s account.6

The discussion is divided into three parts. The first contextualizes Kant’s variousclaims about the categorical imperative and its formulas by considering them in light of

reason I have adopted the first alternative, even though the second is more favorable to the singleness thesis. For adiscussion of the philological issue, see Timmermann (2007), 110–11 and notes 122 and 125. The secondcontentious translation concerns “eher.” This can be translated either as “rather than” in which case Kant isclaiming that the difference between the formulas is subjectively rather than objectively practical, or as“more . . . than,” in which case Kant is claiming that the difference between them is more subjectively thanobjectively practical, which suggests that it is, at least in part, also objectively practical. Abbott, Ellington, Gregor,and Paton opt for the former alternative, Beck, Wood, and Zweig for the latter. My own view is that the textrequires the former reading, since I cannot see how the latter can be made consistent with the singularity thesis.

3 Wood (1999), 186.4 This passage was called to my attention by Timmermann (2007), 110.5 O’Neill (1989), 127.6 It should be noted that O’Neill’s neglect of FLN applies only to Constructions of Reason, where, as far as

I can tell, it is not mentioned at all. Interestingly, in her discussion of applying the categorical imperative inher first book, she focuses considerable attention on FLN and, in the process, anticipates Korsgaard’sconception of a practical contradiction. Unfortunately, however, in that work she does not deal with theequivalency issue. See Nell (O’Neill) (1975), 60–93.

THE S INGLENESS OF THE CATEGORICAL IMPERATIVE 125

the aims and structure of Groundwork 2. The second examines the singleness thesis inlight of this contextualization and finds that, at first glance, it appears to supportWood’sview. The third argues against Wood that a relatively strong version of the singlenessthesis is compatible with the recognition of the progressive nature of Kant’s account.7

IKant entitles Groundwork 2 the “Transition from popular moral philosophy to the metaphysicsof morals.” He takes “popular moral philosophy” in both a general and a more specificsense. The former is the widespread view that morality must be grounded in anempirical analysis of human nature. The latter is the version of this position articulatedby Christian Garve in the commentary attached to his translation of Cicero’s de officiis.8

Although I believe that Garve’s Cicero exercised a profound influence on Kant’sprocedure in the Groundwork, I shall not belabor the point here, since it is not directlygermane to the task at hand.9

As we learn from the Preface and Groundwork 1, by a “metaphysics of morals” Kantunderstands a priori moral philosophy or, more specifically, that portion of a priorimoral philosophy which has the task of uncovering the supreme principle of moralityby means of an analysis of the concept of a rational agent as such.10 This procedurestems from Kant’s twin assumptions that, in order to possess the requisite necessity, anyviable candidate for supreme principle of morality “would have to be valid not merelyfor human beings, but for all rational beings as such” (GMS 4: 40815–16) and that this ispossible only if this principle can be derived from the “universal concept of a rationalbeing as such” (GMS 4: 4123–4).After a preliminary discussion of imperatives in general and of the difference

between the hypothetical and categorical varieties, in which he endeavors to explainthe possibility of the former and postpones a consideration of the latter until Ground-work 3, Kant gets down to the business at hand, namely, analyzing the concept of a

7 For purposes of this essay, I wish to note that I take Kant at his word in insisting that there are only threeformulas. The appearance of there being more stems partly from the fact that the third comes in two forms(FA and FRE) and partly from an unfortunate tendency in the literature to consider FUL as just anotherformula, on the level of the others, rather than, as I believe it to be, the categorical imperative itself. Inmaintaining that there are only three formulas (properly speaking) I am in agreement with Stratton-Lake(1993), 318. The view that there are four is advocated by Guyer (2000), 174; while Paton insists that there arefive (1958), 129–32.

8 Garve (1783).9 See Reich (1939).10 It has often been noted that Kant understands the term “metaphysics of morals” in at least three

different senses in the Groundwork: (1) as equivalent to pure or a priori moral philosophy; (2) as the projectedwork for which the Groundwork is intended to lay the foundation; (3) as that to which a transition is made inGroundwork 2 and from which one is made in Groundwork 3. Taken in first sense, the Groundwork is part of ametaphysics of morals; while taken in the second sense it is clearly not a part of such a metaphysics; and takenin the third sense a metaphysics of morals is part of the Groundwork. See, for example, Bittner (1989), 14–15;and Schonecker and Wood (2004), 10–11.

126 PART I I

categorical imperative. He begins by claiming that the content of such an imperativefollows from an analysis of its concept. This claim is highly controversial and doubtsconcerning it often translate into doubts regarding the imperative itself.11 But havingdiscussed this matter elsewhere, I shall not reconsider it here.12

Instead, I shall begin at the point at which this imperative is in place and Kant takesthe first of several steps that are integral to his account. This step consists in an attemptto illustrate the fecundity of the imperative by means of the four examples. In so doing,Kant shifts from an analysis of the concept of the imperative to an account of theconditions of its application. It is here that the first of three formulas (FLN) isintroduced, wherein the moral law is conceived in accordance with the model of alaw of nature.13 The significance of this formula stems from the fact that it provides themeans by which the imperative is used for the assessment of pre-given maxims, ratherthan, as is sometimes thought, the derivation of duties.Having thus specified the categorical imperative as the supreme principle of morality

and illustrated its applicability, it might seem that Kant has completed the limited taskset for Groundwork 2 and is ready to turn to the third section, where the categoricalimperative is supposedly validated. Instead of doing this, however, Kant embarks upona lengthy account of rational agency, in the course of which he introduces additionalformulas of what is purportedly the same categorical imperative.To the best of my knowledge, the first interpreter to focus on this feature of Kant’s

procedure was A. R. C. Duncan.14 Basing his analysis on the premise that theGroundwork was not intended as a work in ethical theory, but rather, as Kant himselfproclaims, is concerned merely with setting forth and establishing the supreme princi-ple of morality, Duncan argued that the portion of Groundwork 2, in which Kantpresents the various formulas of the categorical imperative, constitutes an “Ethical

11 See, for example, Hill (1985), 19, and Wood (1990), 135–6, 163–7, and (1999), 81–2.12 For my most recent treatment of this issue, see Allison (2011), 135–41 and 162–75.13 I refer to FLN rather than FUL as the first formula because this corresponds to Kant’s procedure at GMS 4:

43615–18, where FUL is not mentioned. My view, admittedly highly controversial, is that FUL should beunderstood as referring to the categorical imperative itself and its three formulas are, as Kant says, FLN FH, andFA/FRE. The obvious objection is that FUL is itself a formula, which appears to imply that the three designatedformulas are formulas of a formula. The point is noted by Timmermann (2007), 104, note 108, in his discussionof Kant’s use of the phrase “the categorical imperative and its formula” at GMS 4: 432 note35–7. I believe that theproblem can be dealt with, however, by noting an ambiguity in Kant’s use of the term “formula” [Formel],particularly in connectionwith the categorical imperative. Sometimes he uses in it in a narrow sense in which it iscontrasted with a principle; it is in this sense that FLN, FH, and FA/FRE are formulas. But he also uses it in anextended sense in which it encompasses the categorical imperative itself as the fundamental principle of morality.For example, after analyzing the concept of a perfectly good or holy will, which, as such, stands under objectivelaws but is not subject to necessitation and therefore to imperatives, Kant concludes that “imperatives [bothhypothetical and categorical] are only formulas expressing the relation of objective laws of volition in general tothe subjective imperfection of the will of this or that rational being” (GMS 4: 4148–10). In other words, animperative (whether categorical or hypothetical) is a “formula” in the sense that it expresses the form that anobjective practical principle assumes in relation to imperfectly rational beings. It is in this sense that FUL is aformula; though it might better be characterized as second-order or meta-formula, which describes the relationin which it stands to the three listed formulas.

14 See Duncan (1957), esp.167–82.


Interlude,” which is tangential to the central concerns of the work.15 And, in order toexplain the presence of this lengthy interlude, Duncan appealed to the work of KlausReich, according to whom the three formulas are intended as “critical” versions of thethree great principles of Stoic ethics in Cicero: (1) the idea of a law of nature and theassociated principle “live according to nature,” (2) the idea that a human being isworthy of respect simply because he is human, and (3) the idea of a universal society ofrational beings.16

Although Duncan is to be credited with both pointing out and attempting to explainthe surprising turn that Kant’s argument takes, I agree with the interpretive consensusin rejecting his contention that the bulk ofGroundwork 2 is a mere interlude rather thanan integral part of Kant’s project in the work.17 But this still leaves us with the task ofexplaining the purpose of that portion of the text in which Kant seemingly complicatesmatters by introducing these additional formulas. If we are to claim that it is integral toKant’s project rather than a mere interlude, we need to provide a clearer account of justwhat this project is.Essential to my account is Kant’s characterization of his project (at least in Ground-

work 2) as deriving the concept of the categorical imperative from the concept of arational agent as such. This requires that Kant spell out the latter concept with sufficientspecificity to allow for the derivation from it of the concept of this imperative; and inlight of this, my claim is that not merely in the portion that Duncan regarded as aninterlude, but in Groundwork 2 as a whole, Kant’s intent is to provide a completeconstruction of the concept of the categorical imperative on the basis of a progressiveanalysis of the concept of (finite) rational agency.By the complete construction of the concept of the categorical imperative, I understand

an account of the necessary and sufficient conditions of its possibility. I take Kant to haveshown that there are three conditions that a categorical imperative must meet. (1) Quaimperative, it must have a strictly universal form.18 (2) It must presuppose something (anend) of absolute value. (3) It must command unconditionally.19 Not coincidentally, eachof these requirements is supplied by one of the formulas. Thus, FLN supplies themomentof universality, modeled on the conception of a law of nature. Correlatively, FH rests on

15 See Duncan (1957), esp. 83–4 and 167. Duncan argues that the 90 paragraphs of Groundwork 2 can bebroken down into four parts, which he terms subsections A–D. Subsection C, which Duncan characterizes as“The Ethical Interlude,” is composed of paragraphs 33–75 (GMS 4: 4215–4374).

16 See Reich (1939) 458–9, and Duncan (1957), 173–8. Duncan’s account differs from Reich’s, however,in that he affirms the identity of the three ideas that Kant used in his subsidiary formulas with the three ideas ofCicero, whereas Reich sees Kant as exercising the principle of charity on behalf of Cicero, attempting toprovide a sense in which the latter’s principles are acceptable. Moreover, Duncan presents his reading as “nomore than a hypothesis,” which he entertains in order “to explain the more curious features in Kant’streatise” (1957, 178).

17 One notable exception to this consensus is Freudiger (1993), 25–31.18 I have added the qualifier “qua imperative” because all concepts, as such, share the form of universality.

The concept of a categorical imperative would be that of an imperative with universal scope, that is, onebinding on all finite rational agents as such.

19 Although she does not put it this way, the point is suggested by Baker (1988), 399–400.

128 PART I I

the assumption that there is something of absolute value, namely, rational nature orhumanity, which, as an end in itself, provides a limiting condition on the pursuit ofprivate ends and the means to attain them. Finally, FA gives the imperative its uncondi-tional status; for Kant argues that only a law that is self-imposed can command uncondi-tionally, that is, apart from any interest or incentive that an agent may have to follow itsdictates. Since these are individually necessary and jointly sufficient conditions of thepossibility of the categorical imperative, we can say, using Kant’s terminology, that theyprovide its “complete determination.”As suggested above, I further maintain that his complete construction (or determi-

nation) of the concept of the categorical imperative proceeds in tandem with aprogressive analysis of rational agency. The analysis is progressive because each of thethree stages, which precede the introduction of the corresponding formula, adds a newdimension to our understanding of the nature of rational agency.The first step is expressed in Kant’s famous dictum that, “Everything in nature works

according to laws. Only a rational being has the capacity to act according to the representa-tion of laws, that is, according to principles or a will” (GMS 4: 41226–8). In other words,unlike everything in nature, rather than merely blindly obeying laws, rational agentshave the capacity to take the principles on the basis of which they act as normative forthem, which means that they not only act according to their representation of laws,they endorse the laws (normative principles) according to the representation of whichthey act. And since laws have as an essential quality universality, this also means thatrational agents have the capacity to govern their actions by principles that lay claim touniversality, which is the feature of rational agency that underlies the categoricalimperative. Although it is often assumed that this dictum expresses the kernel ofKant’s conception of rational agency, we shall see that it fails to provide everythingnecessary for the complete, i.e., fully adequate, concept of the categorical imperative,which is why, pace Duncan and others, Kant’s analysis cannot terminate at this point.20

The second step in Kant’s account of rational agency occurs immediately after thecompletion of the application of FLN. Returning to the concept of the will, which hadpreviously been identified with practical reason, Kant writes: “The will is a capacity todetermine itself to act in accord with [gemäss] the representation of certain laws. . . . Now thatwhich serves the will as the objective ground of its self-determination is the end, and ifthis is given through mere reason must be equally valid for all rational beings”(GMS 4:42719–24). Apart from the reference to “self-determination,” which is already implicitin Kant’s initial account of rational agency, the first part of this new characterizationadds nothing to the first. The second part, however, enriches our conceptions of bothrational agency and the categorical imperative by pointing out that all action aims at an

20 Another commentator who believes that the argument of Groundwork 2 is completed with thestatement of the only possible categorical imperative is Robert Paul Wolff, who explains the existence ofmuch of the remainder of section two as due to Kant’s obsession with his architectonic. SeeWolff (1973), esp.94, 155, 156 note 23.


end, thereby paving the way for the need for an unconditionally valid end as a furthercondition of the possibility of the categorical imperative.Kant completes his analysis by introducing the concept of autonomy. Accordingly,

we learn not merely that rational agents act in accordance with their representation oflaws, which brings with it universality, and that their actions are all end-directed, whichleads to the idea of an unconditionally valid or necessary end, but also that the lawsaccording to which rational beings act stem from their own will. This is likewiseessential to the possibility of the categorical imperative, since it provides the conditionunder which a categorical (unconditional) command is alone conceivable. This isbecause an agent’s subjection to an unconditional practical law is conceivable only ifthe law is self-legislated; for only in this case is the agent’s relation to the law notmediated by any presupposed interest, which would undermine its unconditionality.Neither conceiving of agents as acting according to their representation of law (onmaxims) nor as directed toward self-selected ends is sufficient to capture this thought,since they do not preclude the possibility that the maxims and ends might all reflectprior interests of an agent.21

IIInsofar as it suggests that the three formulas are to be understood in connection with aprogressively richer conception of rational agency, this sketch of Kant’s procedureseems to support Wood’s view. At the heart of the problem lies an issue that has not yetbeen noted, namely, an apparent equivocation on Kant’s part regarding universaliz-ability. Kant introduces this notion with the initial formulation of the categoricalimperative (FUL) and it supposedly provides the test for the moral permissibility ofmaxims. In the application of the first formula (FLN), we are told to act in such a waythat the maxim of one’s action could become through one’s will a universal law ofnature (GMS 4: 42118–20). The universalizability requirement is neither mentioned inthe second formula (FH) nor used in its application. But it reappears in the canonicalformulation of the third: “Choose only in such a way that the maxims of your choiceare at the same time comprehended with it in the same volition as universal law” (GMS4: 44019–21). We shall see, however, that in its reappearance the universalizabilityrequirement does not have the same sense that it had in its initial occurrence and

21 Although FRE has been omitted from the discussion up to this point, I do not believe that thisconstitutes a significant problem for the interpretation offered here, since it is generally recognized that FREis intended by Kant as an alternative expression of FA rather than as a distinct formula. Indeed, Kant himselfsuggests as much when he asserts that the idea of a realm of ends is dependent on or follows from that ofautonomy (GMS 4: 43312–16). Specifically, I regard the idea of a realm of ends as a means whereby finiterational agents can represent to themselves their autonomous agency. In other words, to conceive oneself asacting under the idea of autonomy, that is, as legislating universally for all rational beings as ends inthemselves, just is to think of oneself as a legislative member of a kingdom of ends. See, for example,Moral M2 29: 610; where Kant is cited as claiming: “Man sees himself, in a system of rational beings, as alegislating member thereof; otherwise we are mere instruments.”

130 PART I I

that this is the direct consequence of the intervening conception of rational beings asends in themselves.Setting aside the complications introduced by the contrast between a contradiction

in conception and a contradiction in will, what is tested under FLN is merely thecompatibility of an agent’s own maxim, e.g., false promising when in need of funds,with a hypothetical universal law produced by the universalization of that maxim. Thequestion is whether the maxim can withstand its own universalization.22 In otherwords, at issue is whether the end embedded in the maxim and the means chosen for itsattainment (profiting through false promising) are compatible with the possibility ofthis maxim considered as a universal law of (human) behavior. Since it concerns merelythe self-relation (qua universalized) of the agent’s own maxim, it involves an intra-subjective universalizability, which I shall refer to as universalizability1.This mode of universalizability correlates nicely with Kant’s initial account of

rational agency. In order to be subject to the universalizability requirement, soconstrued, one must be a rational agent with a capacity to act according to one’srepresentation of laws. And, presumably, those affected by actions on the maxim (therecipients of a lying promise) must be regarded as agents in the same sense; otherwisethey could not serve as promisees. In other words, the concept of a promise is onlyapplicable when all parties to it are assumed to be rational agents, at least in the minimalsense expressed by Kant’s first characterization. Moreover, Kant’s illustrations of FLNshow that this conception of rational agency is sufficient to rule out false promising andmany other impermissible maxims. But since FLN requires only universalizability1, themoral standing of the promisees, that is, the question of whether they are autonomousbeings, simply does not arise.Things look considerably different, however, once the conception of rational

agency is thickened in the second formula (FH) to include the proviso that all suchagents are to be regarded as ends in themselves, who, as such, provide a limitingcondition on the choice of ends and of the means used to attain them. Although FHdoes not itself appeal to a universalizability requirement, it provides the basis for achange in the understanding of this requirement, which is reflected in the third formula(FA), where it is expressed in the idea of the will of every rational agent as makinguniversal law, that is, a law that is thought to apply to every rational being as such. Toconsider oneself as making universal law through one’s maxims involves more thanmerely asking whether one’s maxim, qua universal law, can subsist with itself. It alsoinvolves regarding oneself as legislating for all other rational agents, which, since theyare likewise autonomous beings, means acting on maxims that all other agents could

22 This is what Christine Korsgaard has termed the “practical contradiction interpretation,” which I favor.Under the contrasting “logical contradiction interpretation” what is tested is whether a maxim is logicallycompatible with its own universalization. In order to take account of Paton’s position, she also includes a“teleological contradiction interpretation.” For present purposes, however, nothing turns on this point. SeeKorsgaard (1996), 77–105.


regard as universally valid. Otherwise expressed, it is to ask whether the principles onthe basis of which one acts are universally endorsable. In contrast to the initialconception, this is an inter-subjective view of universalizability, which I shall call uni-versalizability2.In view of this difference in the understanding of the universalizability requirement, it

should prove instructive to examine a class of the often discussed counter-examples toFLN, namely, those involvingmaxims of coercion, cruelty, orwanton violence. Considerthe maxim: “I shall murder professors who demand a lot of work and give low grades.”This maxim, and many others of its ilk, seem to pass the contradiction in conception testfor FLN with flying colors, but are blocked by FH, as well as by what Kant terms“common rational moral cognition.” Although most interpreters focus on the questionof reconciling such counter-exampleswith FLN (or FUL), our present concern iswith thefact that they are readily ruled out by FH, since this brings to the fore the equivalence issue.How,wemay ask, can FLN and FHbe regarded as different expressions of the “very samelaw,” if, as seems to be the case, they produce such different results?I shall begin my treatment of this question by considering O’Neill’s attempt to deal

with it. O’Neill defends a strong version of the equivalence thesis, insisting that thevarious formulas yield not only the same results, which makes them extensionallyequivalent, but for the same reasons, which makes them intensionally equivalent aswell.23 With regard to the type of counter-example under consideration, O’Neill’sargument is clear and to the point. She claims that FUL of itself rules out maxims ofviolence or coercion on the grounds that “they undercut the agency of those whomthey victimize.”24 And since the same holds of FH, the two formulas are said to be bothintensionally and extensionally equivalent.25 More specifically, O’Neill maintains thatboth FUL and FH ask an agent’s question “What ought I to do?,” albeit from differentperspectives. As she puts it:

FUL addresses the question from the perspective of agents who acknowledge that others too areagents, and enjoins them to shun principles that could not be adopted by others, that is, thatcould not be universal laws. FEI [FH] addresses the agent`s question from the perspective ofagent`s who acknowledge that action affects others, and enjoins them to avoid damaging others’capacities to act.26

A striking feature of O’Neill’s account is her total neglect of FLN, which leads her totreat the equivalence issue as concerned largely with FUL. This is unfortunate,inasmuch as the deeply problematic false positives all result from the application ofFLN, which appeals to universalizability1. Simply put, the problem made manifest bythe false positives is that universalizability1 and universalizability2 do not always yieldthe same results. Accordingly, by failing to deal with this issue, O’Neill sidesteps ratherthan resolves the equivalence problem as she defines it.

23 O’Neill (1989), 131. 24 Ibid., 133.25 Ibid., 139. 26 Ibid., 128–9. O’Neill uses “FEI” to express what I refer to as “FH.”

132 PART I I

In fairness to O’Neill, however, it must be noted that Kant himself appears to havedone essentially the same thing, by dismissing, as otiose, the need to consider hisexamples under the formula of autonomy. Thus, in excusing himself from this task,Kant writes in a note attached to his account of autonomy: “I can be excused here fromproviding examples to illustrate this principle, since those that have already illustratedthe categorical imperative and its formula can all serve here for the same purpose”(GMS 4: 43235–37). Since a change in the conception of universalizability, such asI have noted, would presumably have led to significant differences in his treatment ofthese illustrations under the formula of autonomy, it is possible that Kant was not awareof such a change. Nevertheless, I retain my thesis that it is both implicit in his accountand reflected in his language.Returning to O’Neill, the above-mentioned gap in her account is evident from her

description of the relationship between the formulas. She is correct in pointing out thateach formula answers the agent’s question: “What ought I to do?” and that eachaddresses this question from a particular perspective.27 And I have no objection to hercharacterization of the perspective of the agent who addresses the question in light of thesecond formula. My quarrel is only with her contention that the first formula addressesthe question “from the perspective of agents who acknowledge that others too areagents, and enjoins them to shun principles that could not be adopted by others, that is,that could not be universal laws.” If my previous analysis is correct, this does not hold forFLN, at least not if this formula is understood in light of Kant’s initial and “thin”conception of rational agency. As previously noted, while the application of FLNrequires one to regard others as rational agents in the sense of beings who act accordingto their representation of laws, it does not require the stronger assumption that they areautonomous agents, who, as such, must be able to endorse one’s principles. This assump-tion only emerges later in the development of the conception of rational agency, whichbrings with it a corresponding strengthening of the universalizability requirement.Accordingly, while I believe that O’Neill is on the right track, I also believe that by

ignoring FLN and the peculiar problems that it raises, she has made things somewhattoo easy for herself; for when FLN is recognized as one of the three formulas, it is farfrom evident that the equivalence for which she argues is to be found. And thus ourproblem remains, albeit in a sharpened form: how can principles that appeal to the verydifferent conceptions of universalizability be regarded as equivalent expressions of the“very same law”?

IIIGiven the constraints of time, I can here do nothing more than to sketch the outlines ofa solution, which is based on my view of the overall structure of Groundwork 2. Two

27 O’Neill (1989), 128.


points here are essential. First, it follows immediately from the hypothesis that the threeformulas are steps or stages in the construction of the complete concept of thecategorical imperative that they are expressions of the “very same law.” Otherwise,they could not be involved in the construction of the same concept. Taken by itself,however, this sense of sameness is too weak, since it does not entail equivalence in eventhe extensional sense. For example, one might say that FLN, FH, and FA are expres-sions of the same law in the sense in which youth, maturity, and old age are stages in thelife of the same person, in which case there is no reason to assume that the applicationof each formula to similar cases would yield uniformly consistent results. For that veryreason, however, it is doubtful that this is how Kant wished his singleness thesis to beunderstood; for it seems reasonably clear that Kant, if not many of his interpreters andcritics, regarded these formulas as at least extensionally equivalent in the sense that theyyield the same results for the same cases.28

The second and crucial point is that Kant’s progressive-constructive treatment of theconcept of the categorical imperative makes possible, indeed requires, the consider-ation of each formula from two points of view: first, as it initially appears in the courseof the construction and second, as it is reconceived in light of the completelyconstructed concept. Moreover, this includes FUL, which I do not consider so mucha formula of the categorical imperative as the categorical imperative itself.29 And sinceit is evident from the already mentioned counter-examples that the formulas are notequivalent when considered from the first point of view, my proposal is that they maybe when considered from the second. I am not claiming that this is how Kantunderstood the matter. Indeed, I suspect that he may have been guilty of an equivoca-tion regarding the understanding of the universalizability requirement; butI nonetheless believe that this proposal opens up the prospect of a positive resolutionof the equivalence issue and that it is a line of argument that Kant could have pursued.Let us consider FLN. In its initial appearance, its near identification with FUL and its

use in illustrating the latter makes it seem as if it simply is the categorical imperative,clothed in a form suitable for the appraisal of maxims. At this juncture, Kant gives nohint that there is any other manner or formula in which this imperative might beexpressed. In the course of Kant’s analysis, however, we come to realize that FLNcorresponds merely to a moment or aspect of the concept of the categorical imperative,namely, its universality. Moreover, we have seen that in the constructive process theuniversalizability requirement not only loses its pretense of being the whole story, but isitself transformed from universalizability1 to universalizability2. And what is particularlynoteworthy, though hardly surprising, is that the universalizability2 requirement isequivalent to the requirement to treat rational agents (including oneself) never merely

28 I take this to be implied by Kant’s claim that the difference between the formulas is “subjectively ratherthan [my emphasis] objectively practical” (GMS 4: 43610–11), as well as by his decision not to consider hisexamples under the formula of autonomy because it would be redundant.

29 See note 13.

134 PART I I

as a means but also at the same time as an end. In fact, Kant says precisely this in apassage from his final overview of the argument, where he writes:

The principle [Princip] act with reference to every rational being (to yourself and others) so that inyour maxim it holds at the same time as an end in itself, is at bottom the same [im Grunde einerlei]as the basic principle [Grundsatz], act on a maxim that at the same time contains in itself its ownuniversal validity for every rational being. For to say that in the use of means to any end I am tolimit my maxim to the condition of its universal validity as a law for every subject is tantamountto saying that the subject of ends, that is, the rational being itself, must be made the basis of allmaxims of action, never merely as a means, but as the supreme limiting condition in the use of allmeans, that is, always at the same time as an end. (GMS 4: 43734–387)

Although the first of the two principles to which Kant refers in this passage ispresumably FH, the second is clearly not FLN; for the universal validity at issue isunderstood as holding for all rational agents, which amounts to a universal endorsa-bility. Rather, I take what Kant refers to as the basic principle [Grundsatz] to be thecategorical imperative itself, that is, FUL, not in its initial presentation, which leavesopen how the universalizability requirement is to be understood, but in its completelyconstructed form, in which the latter is taken as universal endorsability. Accordingly,I shall call this FUL2.

30

The most important point about the passage before us, however, is that it explicitlyaffirms the reciprocity between maxims that are universally endorsable and those thatconsider every rational agent not merely as a means but also at the same time as an end.This reciprocity is grounded in the fact that regarding every rational agent who couldbe affected by acting on it in this way is precisely what makes a maxim universallyendorsable. And from this it follows that every universally endorsable maxim necessar-ily regards every rational agent who could be affected by someone acting on it notmerely as a means but also, at the same time, as an end and vice versa. Otherwiseexpressed, a maxim is universally endorsable if and only if it regards every rational agentwho could be affected by one’s acting on it not merely as a means but also, at the sametime, as an end. Moreover, if this is correct, then those troublesome false positives,which threatened the singleness thesis and, therefore, the very foundation of Kant’smoral theory are ruled out by the completely constructed categorical imperative,which should at least help to mitigate the persistent worries about Kant’s claim thatthere is a single categorical imperative.Finally, I stated at the outset that I sawmy task as incorporating the moments of truth

in the highly divergent treatments of this topic by Onora O’Neill and Allen Wood,

30 I believe that FUL2 can be equated with what Kant at one point characterizes as the “universal formula”(UF), which states: “Act according to that maxim which can at the same time make itself into a universal law” andwhich Kant claims is the preferred formulation for moral appraisal (GMS 4: 43627–4371). The universalformula is usually equated with FUL, that is, to say, with the categorical imperative itself. This reading hasbeen challenged, however, by Wood, who equates it with FA. See Wood (1999), 187–90. I discuss Wood’sview and present my own analysis of the universal formula in Allison (2011), 251–7.


while avoiding what I take to be their errors. By showing that O’Neill is correct ininsisting upon equivalence but errs in overstating her case and making things rather tooeasy for herself by glossing over the problems posed by FLN and the ambiguity of FUL,while Wood is correct in emphasizing the progressive, developmental nature of Kant’saccount but errs in denying that this is compatible with any serious account of thesingleness of the categorical imperative, I believe that I have succeeded in this task.Whether this suffices as a defense of Kant is another matter.

136 PART I I

Essay Nine

Kant on Freedom of the Will

Although there can be no doubt regarding the centrality of the concept of freedom inKant’s thought, there is considerable disagreement concerning its proper interpretationand evaluation. The evaluative problem stems largely from Kant’s insistence thatfreedom involves a transcendental or non-empirical component, which requiresthe resources of transcendental idealism in order to be reconciled with the “causalityof nature.” There is also, however, a significant interpretive problem posed bythe number of different conceptions of freedom to which Kant refers.1 In addition to“outer freedom” or freedom of action, and a relative, empirically accessible or “psy-chological” concept of freedom, which admits of degrees, Kant distinguishes betweentranscendental and practical freedom, both of which seem to involve indeterminismin the sense of an independence from determination by antecedent causes. Moreover,within this sphere Kant conceives of freedom as both absolute spontaneity (negativefreedom), which is a condition of rational agency as such, and as autonomy (posi-tive freedom), which is a condition of the appropriate moral motivation (acting fromduty alone).Given this complexity, the present discussion must be highly selective. Specifically,

it will focus initially on the nature of and relation between freedom as spontaneity andas autonomy. But since both of these senses of freedom affirm (albeit in different ways)an independence from natural causality, this necessitates a consideration of the rela-tionship between freedom (in both senses) and transcendental idealism. And, in orderto situate Kant’s views in their historical context, I shall frame the discussion with abrief account of the treatment of free will by some of his predecessors, on the one hand,and his idealistic successors, on the other.

I Freedom of the Will in Kant’s PredecessorsIn the German context, the agenda for the discussion of freedom of the will in theeighteenth century was set by Leibniz, who approached the topic in terms of hisprinciple of sufficient reason. After Leibniz, the main participants in this discussionwere Christian Wolff and Christian August Crusius. The former developed and

1 For a systematic discussion of this issue, see Beck (1960), 176–208, and (1987).

systematized the Leibnizian position and the latter was its foremost critic. Accordingly,a consideration of the views of these three thinkers is essential to the understanding ofKant’s position.

Gottfried Wilhelm Leibniz: Leibniz’s philosophy is built on two great principles: contra-diction (or identity) and sufficient reason. The former states that “a proposition cannot beboth true and false at the same time”; the latter that “nothing happens without a reasonwhy it should be so rather than otherwise.”2Whereas thefirst governs (logically) necessarytruths, which hold in all possible worlds, the second governs contingent truths, whichhold in the actual world. In addition to factual truths and laws of nature, the latter includesthe basic propositions of Leibniz’s metaphysics.The principle of sufficient reason can have this metaphysical function, however,

only because it involves a certain kind of necessity. Since God is a supremely perfectbeing, it follows (according to this principle) that God could choose only the best of allpossible worlds. Already during his lifetime, Leibniz was attacked on this point fordenying divine freedom, subjecting God to an over-riding necessity or fate. Typically,he dealt with this problem by distinguishing between an absolute or logical necessityand a relative or hypothetical one: the former applying to necessary and the latter tocontingent truths. Accordingly, Leibniz denied that it is absolutely necessary for God tocreate the best of all possible worlds, while also admitting that there is a sense in whichhe must choose the best, since anything else would constitute a violation of theprinciple of sufficient reason.Leibniz applied the same general framework to the analysis of human freedom.

Thus, he denied that the voluntary actions of finite rational agents are absolutelynecessary, since their non-occurrence does not involve a contradiction, while insistingthat their occurrence is certain (and is known timelessly by God), since their non-occurrence would violate the principle of sufficient reason.So far it might seem that Leibniz had “saved” freedom only by contrasting contin-

gent truths about the occurrence of human actions with logically necessary truths suchas those contained in mathematics. This is obviously inadequate as an analysis offreedom, however, since (among other things) it fails to distinguish between voluntaryactions and other occurrences in nature, which are causally rather than logicallynecessary. But Leibniz was well aware of this problem and attempted to deal with itby appealing to two more necessary conditions of a free act: spontaneity and choice (orintelligence). As he puts it at one point:

Aristotle has already observed that there are two things in freedom, to wit spontaneity andchoice, and therein lies our mastery over our actions. When we act freely we are not beingforced, as would happen if we were pushed on to a precipice and thrown from top to bottom;and we are not prevented from having the mind free when we deliberate, as we would be if we

2 Leibniz (1956), 15. Leibniz refers to both principles, albeit often with somewhat different formulations,in virtually all of his philosophical writings.

138 PART I I

were given a draught to deprive us of discernment. There is contingency in a thousand actions ofNature; but when there is no judgment in him who acts there is no freedom.3

As this passage indicates, by “spontaneity” Leibniz understood the absence of compul-sion by any external cause, and by “choice” (or “intelligence”) the recognition (or atleast belief) that a course of action is, in given circumstances, the best. But, likecontingency, spontaneity alone is insufficient for freedom, since it characterizes someactions of inanimate objects (e.g., a ball which has been set in motion along a smoothtrajectory) as well as the behavior of non-rational animals.4 Thus, again followingAristotle, Leibniz thought that we can speak meaningfully of freedom only in the caseof voluntary actions, in which an agent makes a conscious choice based on theperception of some good.Although understanding freedom in this way enabled Leibniz to bring free actions

under the principle of sufficient reason, it entails that, given a motive and a specific setof circumstances, an agent will invariably choose to act according to what is perceivedto be the best. In other words, an agent could not have chosen otherwise under thesame circumstances. Rather than denying this implication, however, Leibniz at-tempted to reconcile it with freedom by appealing to his dictum that a reason ormotive “inclines without necessitating.”5 He does not mean by this that free agentshave a capacity to disregard their motives (that would constitute a violation of theprinciple of sufficient reason), but merely that being motivated to X does not renderthe performance of X anything more than hypothetically necessary. Consequently,freedom for Leibniz is compatible with a certain kind of necessity.

Christian Wolff: If Wolff may be said to have modified the Leibnizian conception offreedom at all, it is by emphasizing even more strongly its deterministic features and itsintellectualism or anti-voluntarism. The first of these is a consequence of his attempt toderive the principle of sufficient reason from the principle of contradiction.6 Since theformer supposedly governs all that exists, its demonstration on the basis of the principleof contradiction threatens to reduce everything to a matter of logical necessity in themanner of Spinoza. Indeed, the charge that he taught a universal determinism was oneof the main reasons for Wolff’s expulsion from Halle in 1723.7 Nevertheless, Wolff didnot think that the attribution of a logical necessity to the principle of sufficient reasonentails that everything based on this principle is itself logically or “absolutely” neces-sary. Accordingly, he retained Leibniz’s distinction between absolute and hypotheticalnecessity and subsumed free actions under the latter. And, also like Leibniz, Wolfflocated the distinctive feature of free actions in the kind of grounds they have, not intheir lack of sufficient determining grounds. Specifically, free actions for Wolff arethose that are performed on the basis of what the intellect perceives to be best. Thus,rejecting the characterization of freedom as the capacity to choose either of two

3 Leibniz (1951), 143. See also 303. 4 See Leibniz (1981), 176. 5 Ibid., 175.6 Wolff (1997), vol. 1, 17–18. 7 See Beck (1969), 259.

KANT ON FREEDOM OF THE WILL 139

contradictory things on the grounds of its violation of the principle of sufficientreason,8 Wolff defined it instead as “the ability of the soul through its own power ofchoice to choose, between two equally possible things, that which pleases it themost.”9

This conception of freedom must be understood in terms of Wolff’s anti-volunta-rism. Strictly speaking, there is only one mental faculty for Wolff: the cognitive. Theother two traditionally conceived faculties (will and desire) are subsumed under it asreflecting different degrees in the distinctness of one’s cognition.10 Although it followsfrom this that an agent will necessarily “choose” what seems to be the best in a givensituation, this does not undermine the freedom of choice. The latter is preservedbecause the action is determined intellectually by what is perceived (rightly or wrong-ly) to be the best rather than being the result of compulsion by external forces.Moreover, like many eighteenth-century thinkers (including Hume), Wolff insistedthat such a conception of freedom is not only compatible with, but required bymorality, because its alternative (the so-called “liberty of indifference”) effectivelydeprives an agent of any motive or reason to act.11

Christian August Crusius: As already noted, Crusius was the foremost opponent ofWolffian thought in Germany and, as such, exercised a major influence on Kant.12 Forpresent purposes at least, the focal points of his critique are the Wolffian understandingof the principle of sufficient reason and its intellectualism or anti-voluntarism. Not onlydid Crusius reject as spurious Wolff’s attempt to demonstrate the principle of sufficientreason, he also repudiated the intellectualization of the real, that is, the equation ofconditions of cognizing (or consistent thinking) with ontological conditions. Againstthis virtual collapsing of ontology into epistemology or logic, Crusius (anticipatingKant) distinguished sharply between ideal and real grounds, between the logicalrelation of ground and consequent and the real relation of cause and effect. The latterhas a kind of necessity, which Crusius never succeeded in explaining very well, but isquite distinct from the logical necessity based on the principle of contradiction.13

Armed with this sharp distinction between conditions of cognition and conditions ofbeing, Crusius thought that he had provided conceptual space for a genuine freedom ofthe will. This is not only because the necessity governing the real is not a logical necessity,which even the Leibnizians acknowledged, but also because the principle of sufficientreason governs our understanding of things rather than the things themselves. Thus, againanticipating Kant, Crusius claimed that the endeavor to comprehend freedom leads to anunavoidable conflict of principles: on the one hand, we cannot conceive an actionwithout a cause (which rules out freedom in a stronger than Leibnizian sense), while,

8 Wolff (1997), 312–13. 9 Ibid., 317. 10 Ibid., 544–6.11 Ibid., 313–14. For a discussion ofWolff ’s ethics that highlights this point, see Schneewind (1998), 435–42.12 The following discussion of Crusius is indebted to those of Heimsoeth (1967); Beck (1969), 394–402;

and Schneewind (1998), 445–56.13 See Beck, (1969), 396–7.

140 PART I I

on the other hand, wemust assume freedom (in a strong sense) in order to conceive of thepossibility of moral agency.14

Crusius’ basic claim is that the distinction between these two kinds of conditionsentails that our inability to understand or explain freedom does not preclude its reality.On the contrary, he insisted on the reality of freedom as a fundamental power of thesoul, which can be known to be actual, even though it is inexplicable. Within theframework of his epistemology, Crusius explained this in terms of a distinctionbetween two kinds of cognition: symbolic and intuitive.15 The former comprehendsthings in terms of their relations to something else and the latter consists in animmediate awareness. Accordingly, we can have a direct assurance of realities thatwe cannot understand. Freedom is one such reality.Crusius’ justification for this claim turns on his voluntarism. Indeed, to underscore

this point, he introduced a distinct science (Thelematologie) whose special provenance isthe will.16 In this context, Crusius defined the will as the power to act according toone’s ideas. His point is that a capacity for cognition does not entail a capacity to actaccording to its determinations.17 Since Crusius thought that the will, as the chiefpower of the mind, has an executive function that presupposes, but cannot beperformed by, the intellect, he denied that God would create a being with understand-ing but no will.18

Moreover, for Crusius, this function of the will presupposes freedom in a strongsense. Consequently, the Wolffian account of freedom will not do, since it is a thinlyveiled determinism, which reduces virtue to a matter of luck.19 Against this, Crusiusinsisted that freedom must involve a capacity to choose between given alternatives,since it is only on the basis of this assumption that acts can be imputed to an agent. Ashe puts it at one point, “A willing that one could in identical circumstances omit ordirect to something else is called a free willing.”20 And later, in defining the mostperfect concept of freedom, he writes:

Whenever we freely will something, we decide to do something for which one or several desiresalready exist in us. . . . Freedom consists in an inner perfect activity of the will, which is capable ofconnecting its efficacy with one of the currently active drives of the will, or of omitting thisconnection and remaining inactive or of connecting it with another drive instead of the first one.21

For Crusius, then, rather than being determined by its strongest desire or, on theintellectualist version, by what is perceived to be the best, the will is conceived assomehow standing apart from its desires, with the capacity to determine which, if anyof them, are to be acted upon. Although such a conception of agency makes no sensefrom the Wolffian standpoint, with its reduction of all the powers of the mind to

14 Ibid., 400. 15 Crusius (1964), 171–3. See also Beck (1969), 399–400.16 Crusius (1969), 1, 3–4. 17 Crusius (1964), 427, 826–7. See also Beck (1969), 401.18 Crusius (1964), 866–7, 885–6. 19 Crusius (1969), 46, Schneewind (1998), 449.20 Crusius (1969), 54, Schneewind (1998), 449.21 Crusius (1969), 54–5. See Schneewind (1998), 449.


cognition, Crusius thought that he was able to accommodate it by means of his sharpdistinction between intellect and will.

II Kant’s Conception of Rational AgencyAlthough Kant’s eventual understanding of freedom of the will has strong affinities toCrusius’, his initial account is Wolffian. Thus, in his first metaphysical venture, Kantdefends the distinction between absolute and hypothetical necessity against Crusiusand, in good Wolffian fashion, insists that the question of freedom concerns the natureof the necessitating ground rather than the kind or degree of its necessitation (ND 1:400). Accordingly, freedom of the will is said to consist entirely in its being determinedby “motives of the understanding” rather than by external stimuli (ND 1: 400).Appealing to Leibnizian terminology, Kant defines spontaneity as “an action whichissues from an inner principle,” and remarks that “When this spontaneity is determinedin conformity with the representation of what is best it is called freedom” (ND 1: 402).Remnants of this view are to be found in some of Kant’s lectures on moral

philosophy and associated Reflexionen, where he appeals to a relative, empiricallybased conception of freedom. In this context, he speaks of degrees of freedom,corresponding to degrees of rationality, and correlated with degrees of imputability.Indeed, as late as 1784–5, Kant is cited as claiming that, “The more a man can bemorally compelled the freer he is; the more he is pathologically compelled, though thisonly occurs in a comparative sense, the less free he is” (MPC 27: 268). Although it isdifficult to know what to make of such claims, particularly those stemming from theperiod after the initial publication of the first Critique, the facts that Kant usedBaumgarten as his text and that this remark (and many others like it) is taken from astudent’s notes, makes it plausible to assume that he was stating Baumgarten’s viewrather than his own.Be that as it may, in his metaphysical lectures of the seventies Kant repudiates the

Wolffian conception of freedom on essentially Crusian grounds. Thus, he nowdistinguishes sharply between an absolute and a relative or conditioned spontaneityof the kind advocated by the Leibnizians. The former is claimed to be essential tofreedom in the genuine or transcendental sense, while the latter is compared to that of awatch or turnspit (ML1 28: 267–8).22 In fact, at one point Kant seems to haveentertained a speculative proof of transcendental freedom that anticipates later idealisticaccounts. According to this proof, the very conception of oneself as a thinking beingproves one’s transcendental freedom. As Kant puts it:

When I say: I think, I act, etc., then either the word “I” is used falsely or I am free. Were I notfree, I could not say: I do it, but rather I would have to say: I feel a desire in me to do, which

22 Kant later used the comparison to the freedom of a turnspit to denigrate the Leibnizian conception inKpV 5: 97.

142 PART I I

someone has aroused in me. But when I say: I do it, that means spontaneity in the transcendentalsense. (M L1 28: 269)

Nevertheless, even at that time Kant did not regard the self ’s transcendental freedom asunproblematic. Unlike his later treatments, however, its problematic feature is found inits apparent conflict with our ontological status as dependent beings.23 The problem isto understand how such a being could have anything more than the relative spontane-ity recognized by the Leibnizians. His resolution of the problem at this point is basicallythat of Crusius: we know that we are free in the transcendental sense, but we cannotexplain how this is possible (ML1 28: 270–1).The “critical” Kant retained the doctrine of the incomprehensibility of freedom,

while denying its cognizability. This denial is a consequence of the limitation of ourcognition to phenomena, that is, to things as they appear in accordance with our formsof sensibility. Since the idea of transcendental freedom is the thought of an agency thatis not determinable by sensible conditions, we cannot be said to know that we possessit. Kant also insists, however, on both the possibility and necessity of thinking ourfreedom, understood as an absolute spontaneity. Freedom, so conceived, is a transcen-dental idea (a necessary idea of reason), which is required for the thought of ourselves ascognizers and as agents.The idea that even our capacity to think presupposes absolute spontaneity seems to

have its roots in the speculative proof noted above. As Kant develops it in his laterwritings, the basic point is that to consider oneself as a cognizer is to assume suchspontaneity. This is because to understand or cognize something requires not simplyhaving the correct beliefs and even having them for the correct reasons, it also involvesa capacity to take these reasons (whether rightly or wrongly) as justifying the belief. Inshort, the thought of ourselves as self-conscious cognizers is inseparable from the ideaof our absolute spontaneity.24

Kant recognized, however, that this is not sufficient to justify freedom in thepractical sense, that is, freedom of the will as it is usually understood. The problemstems from the fact that our epistemic spontaneity appears to be self-certifying in a wayin which our practical spontaneity is not. The first part of this story is familiar, albeithardly non-controversial. Since spontaneity in the above-mentioned sense is a neces-sary condition of thinking, I cannot think of myself as thinking without attributingsuch spontaneity to my mind. Expressed in Cartesian terms, I cannot coherently doubtthat I am a thinker because such doubt is itself an act of thinking. But since it does seempossible to doubt that one has a will, or, equivalently, that one’s reason is practical, thisline of argument cannot be directly carried over into the practical sphere. For all that

23 Kant, ML1: 268–70. Nevertheless, it should be kept in mind that Kant still wrestled with this problemin his “critical” period. See KpV 5: 100–4.

24 See Kant, GMS 4: 448; RSV 8: 14. I discuss this issue in Allison (1990), 37–9 and (1996a), 98–106,130–4, as well as in (1997a) and (forthcoming), which are included in this volume.


we know we might be nothing more than thinking automata: beings who are capableof thought, but whose actions are governed by instinct rather than practical reason.Here again, the influence of Crusius is evident. For both thinkers it is the separation

of intellect and will as distinct powers of the mind that opens up the possibility, whichis unintelligible from the Leibnizian point of view, that we might have the formerwithout the latter. Unlike Crusius, however, Kant does not deny such a state of affairson theological grounds, but seems to have held that it is a possibility that cannot beexcluded by the resources of theoretical reason.25

The main point, however, is that from the practical point of view, this possibility ismoot. At least from a first-person perspective, while engaged in deliberation regardingthe proper course of action, we necessarily presuppose our freedom. To take oneself asa rational agent capable of choice and deliberation is to assume that one’s reason ispractical or, equivalently, that one has a will. As Kant famously puts it, “Now I assertthat to every rational being having a will we must necessarily lend the idea of freedom,also, under which alone he acts” (GMS 4: 448).As we shall see below in connection with the Third Antinomy, the idea of freedom

is that of an uncaused cause, that is, of an agency capable of making an “absolutebeginning,” by which is understood the capacity to initiate a causal series that is notitself determined by any antecedent condition. In its application to the human will andits practical freedom, this means that we are rationally constrained to regard ourselves asspontaneous initiators of causal series through our choices. Otherwise expressed, wecannot, at least from the first-person point of view, regard these choices as thepredetermined outcomes of either the state of the world or of our own psychologicalstate, including our beliefs and desires.Perhaps Kant’s best statement of the conception of such freedom is in a passage from

Religion, where he writes:

[F]reedom of the power of choice [Willkür] has the characteristic, entirely peculiar to it, that itcannot be determined to action through any incentive except so far as the human being hasincorporated it into his maxim (has made it into a universal rule for himself, according to whichhe wills to conduct himself); only in this way can an incentive, whatever it may be, coexist withthe absolute spontaneity of the power of choice (of freedom). (RGV 6: 24)

Although this characterization of the freedom of the power of choice is part of Kant’sdiscussion of how the moral law can be an incentive for sensibly affected beings such asourselves, it is noteworthy that he claims that it applies to any incentive (includingthose based on inclination). Consequently, it is best viewed as providing a model forthe thought of free agency in general rather than merely moral agency. Nevertheless,its force is normative rather than descriptive. It is not the case that introspection

25 Nevertheless, there remains a teleological, if not theological, component in Kant’s approach to theissue. For example, in the Groundwork he speculates that if nature had aimed at our happiness rather than ourmorality, it would have left things up to instinct rather than to practical reason (4: 395).

144 PART I I

invariably shows that we never act on a incentive without first “incorporating it intoone’s maxim”; it is rather that we necessarily conceive our agency according to thismodel insofar as we take ourselves to be acting on reasons. This act of incorporationmay also be seen as the practical analogue of the spontaneity that we necessarilyattribute to our understandings in cognition. Just as reasons to believe cannot functionas reasons unless we take them as such, desires do not of themselves provide us with asufficient reason to act. They can become reasons only insofar as we freely assign themthis status, by subsuming them under a principle of action (maxim), which we likewisefreely adopt.26

If we put this central Kantian idea into its historical context, it may be seen as asuccessor to the Leibnizian dictum that motives incline without necessitating. Thedifference is that, under the influence of Crusius, Kant construed the distinctionbetween inclination and necessitation in much stronger terms than the Leibnizians.Whereas the latter meant by this merely that a strong inclination to X does not make itabsolutely necessary that one will X (though all things considered, it makes it certain),for Kant such an inclination does not, of itself, even give one a sufficient reason to X.

III Imputability and AutonomyKant’s account of the relationship between morality and freedom is complicated by thefact that it encompasses two issues: imputation and motivation. His treatment of theformer consists in a relatively straightforward application of the general conception ofrational agency to morally relevant acts. The basic idea is that the imputability ofactions presupposes freedom in the strong sense of absolute spontaneity. Accordingly,unlike the Leibnizian view, for which a merely relative spontaneity, understood as alack of external compulsion, suffices to ground responsibility, for Kant it requires thatthe agent is not predetermined at all. In other words, without violating the psycholog-ical continuity of the person, we must consider an imputable act as if it were an“absolute beginning.”Once again, this is close to the position of Crusius, who defended the traditional

view that freedom involves a capacity to have chosen otherwise in a given set ofcircumstances. Kant, however, gives a somewhat different twist to this thought in lightof his conception of morality. Rather than defining freedom simply as the capacity todo otherwise in the sense of an ability to choose either for against the dictates ofmorality, Kant typically appeals to the principle that “ought implies can.”Consequent-ly, the weight of his account falls on the idea that, no matter how dire one’scircumstances, one is aware through a consciousness of standing under the moral lawthat one can do what duty requires, simply because one ought to do so. Moreover, it is inthis sense that we must understand Kant’s claim, which may at first seem to conflict

26 Elsewhere I have termed this the “incorporation thesis.” See Allison (1990), 5–6 and passim, and(1996a), 118–23, 130–5, and 139–42.


with the previous analysis of rational agency, that without the moral law such freedomwould have remained unknown to the agent (KpV 5: 30).27 It is not that theconsciousness of the moral law first makes us aware of our rational agency, since insofaras we take ourselves to be acting we are necessarily conscious of that. It is rather that thisconsciousness makes us aware of a capacity to disregard all our inclinations, even ournatural love of life, when duty requires it.This conception of freedom rests on the assumption that moral considerations give

one a sufficient reason to act or, as Kant usually puts it, that the moral law serves as anincentive. To understand this, however, we need to consider the doctrine of theautonomy of the will. As introduced in the Groundwork, autonomy is defined as “theproperty of the will by which it is a law to itself (independently of any property of theobjects of volition)” (GMS 4: 440). It is contrasted with the principle of heteronomy,which denies that the will can give itself the law and assumes that the object (whateverhappens to be desired) must give the law to the will (GMS 4: 441). Consequently, toattribute heteronomy to the will is not to claim that it is causally determined, but ratherthat it requires some antecedent desire in order to have a reason to act.Kant claimed that the will’s heteronomy was presupposed by all previous moralists,

including voluntarists such as Crusius (KpV 5: 39). His basic objection is that it isincompatible with the possibility of the categorical imperative, since the latter not onlydetermines what our duty is in given circumstances, but requires us to act from duty,which is possible only on the assumption that the will is autonomous. Thus, unlikemost present-day conceptions of autonomy, Kant’s is an all or nothing affair: either thewill has it or it does not. Moreover, if it does not, morality must be rejected as aphantom of the brain (GMS 4: 445).In the Groundwork, Kant also argues that the positive conception of freedom

(autonomy) follows from the negative conception (spontaneity) and that given auton-omy, “morality together with its principle follows from it by mere analysis of itsconcept” (GMS 4: 447).28 Or, as he puts it in the second Critique, “[F]reedom andunconditional practical law reciprocally imply each other” (5: 529). Both formulationscome to the same thing, namely, that freedom (construed as autonomy) is not only anecessary, but also a sufficient condition of morality.29 Consequently, given thesignificance that Kant attributes to autonomy, it is no wonder that he devoted thethird part of theGroundwork, which is intended to establish the reality of the categoricalimperative, to a “deduction” of the autonomy of the will.

27 See also Kant’s earlier claim in the same work that the moral law is the “ratio cognoscendi of freedom” (KpV5: 4n).

28 Kant often contrasts a positive and a negative sense of freedom, albeit not always in the same terms. Inthe Groundwork freedom in the positive sense is identified with autonomy, while in its negative sense it isunderstood as the will’s capacity to exercise its causality independently of determination by alien causes (4:446). The former is contrasted with heteronomy and the latter with natural necessity.

29 I have termed this the “reciprocity thesis.” For my discussion of this thesis, see Allison (1990), 201–13and (1997a), 136–8 and passim.

146 PART I I

Although it is impossible to examine here this complex and difficult argument,which was replaced in the second Critique by an appeal to the “fact of reason,”30 it isnecessary to consider briefly Kant’s distinction between Wille and Willkür, which is acentral feature of his treatment of the will in his later writings. Whereas in theGroundwork Kant simply identified will (Wille) with practical reason, thereby equatingthe question of whether we have free will with the question of whether our reason ispractical, in these later writings he introduced a more complex account of the will ascontaining both legislative and executive functions.31

In addition to creating problems for the translator, since each of these terms can berendered as “will,” the situation is further complicated by the fact that Wille itself istaken in two senses: a broad sense in which it connotes the faculty of volition, or will asa whole, and a narrow sense in which it connotes the legislative function of this faculty.Accordingly, both Wille in the narrow sense and Willkür, which is perhaps besttranslated as “the power of choice,” are aspects of Wille in the broad sense.It is tempting to correlate these two aspects of will with the two conceptions of

freedom (spontaneity and autonomy). In fact, this works nicely in the case of Willkür,the freedom of which consists in an absolute spontaneity. The situation is morecomplex in the case of the connection of Wille and autonomy, however, since Kantdoes not seem to have been at one mind on the matter. Thus, in the published text ofthe Metaphysics of Morals, Kant states that only Willkür can be regarded as free, whereasWille must be thought to be neither free nor unfree, since it is directed to giving lawrather than to action (MS 6: 226). By contrast, in his unpublished preliminary notes(Vorarbeiten) for this work, Kant entertains the possibility that Wille might be free in adifferent sense thanWillkür because it is law giving rather than law following (VMS 23:249). Perhaps the best way to render Kant consistent on this point is to keep in mindthe distinction between the two senses of Wille. When Kant denied the Wille as such iseither free or unfree, he had in mind the narrow sense of the term. Wille, so construed,may not be thought to be free with regard to the legislation of the categoricalimperative, since this is its fundamental law. But, conversely, it does seem possible toattribute autonomy to Wille in the broad sense, since it is conceived as legislating toitself.

IV Freedom and DeterminismThe centerpiece of Kant’s account of freedom is the Third Antinomy and its attemptedresolution through an appeal to transcendental idealism. Like each of the four anti-nomies, the third is presented as a conflict between cosmological ideas, that is, betweenways of conceiving the world as a whole (as a totality of conditions). In this case, it is aconflict between the conception of the world as containing an infinite series of causal

30 My most recent and detailed discussion of this deduction is in Allison (2011).31 For my discussion of this distinction, see Allison (1990), 129–36.


conditions, each of which is itself conditioned by its antecedent condition, and theconception of this series (and therefore the world as a whole) as anchored in somethingthat is itself unconditioned. On Kant’s analysis, each side is capable of demonstrating acontradiction in the opposed view. But since it is assumed by both parties that thesealternatives are themselves contradictory, the refutation of one is seen as equivalent tothe demonstration of the other.Insofar as this dispute is explicitly concerned with a cosmological issue regarding the

need for (and possibility of ) a first cause, its connection with the question of free will isnot immediately evident. Kant’s explanation turns on the conception of freedom asabsolute spontaneity. As we have seen, to consider oneself as free in this sense is toconceive oneself as initiating through one’s choice a fresh chain of events or an“absolute beginning.” But the problem with such a conception of agency is that itappears to conflict with the principle of the Second Analogy: “Everything that happens(begins to be) presupposes something which it follows in accordance with a rule” (KrVA189).Kant’s general approach to the antinomial conflict is to suggest that the appearance

of a contradiction rests on a misunderstanding, which is a consequence of the tran-scendental realism assumed by both parties to the dispute. Since Kant regards suchrealism as the contradictory opposite of his own transcendental idealism, he defines it inrelation to the latter. Underlying the contrast between the two forms of transcenden-talism is the distinction between objects considered as they appear, that is, qua givenunder the subjective conditions of human sensibility (space and time), and these sameobjects considered as they may be in themselves, that is, qua thought independently ofthese conditions by some putative “pure understanding.” Whereas the transcendentalidealist limits human cognition to objects considered in the former way, the transcen-dental realist ignores this distinction and assumes that our cognition, even of thespatiotemporal objects of human experience, concerns objects considered in the latterway. In Kant’s terms, the transcendental realist treats mere appearances as if they werethings in themselves.32

According to Kant, this confusion leads directly to the misunderstanding underlyingthe antinomial conflict as a whole and can be avoided only by replacing transcendentalrealism with transcendental idealism. This is because the transcendental realist iscommitted to the assumption that the totality of conditions must be “given” (at leastfor God) independently of our piecemeal and successive cognition of them. Conse-quently, such a realist necessarily assumes that this totality consists of either a finite or aninfinite number of conditions. In the case of the Third Antinomy, the issue is whetherthere is a first, uncaused cause or an infinitely extended causal chain, every member ofwhich is itself causally conditioned.

32 For an analysis of the contrast between transcendental idealism and transcendental realism, see Allison(2006a), which is included in this volume.

148 PART I I

Things look rather different from the transcendentally idealistic point of view,however, since it is no longer assumed that there is some ultimate fact of the matter,not even one to be determined by God. The claim is rather that each position islegitimate, if relativized to a point of view. From the empirical point of view, everycondition must itself be conditioned, which leaves no room for an absolute beginningor uncaused cause; whereas from the intellectualist point of view, which is concernedwith the conditions of coherent thought rather than experience, it is necessary toassume some such cause in order to satisfy reason’s demand for completeness. Kant’sclaim is that transcendental idealism (unlike transcendental realism) is able to reconcilethese two points of view by introducing a distinction between conditions of experienceand conditions of thought. This creates logical space for the possibility that both partiesmay be correct: the determinist with respect to objects of possible experience and theindeterminist with respect to merely intelligible objects.In applying this schema to the human will, Kant invokes a distinction between

empirical and intelligible character.33 The will in its empirical character is described as“nothing other than a certain causality of . . . reason, insofar as in its effects in appear-ance this reason exhibits a rule in accordance with which one could derive the rationalgrounds and the actions themselves . . . and estimate the subjective principles of hispower of choice [Willkür]” (KrV A549/B577). Although it may appear strange to findthe will in its empirical character described as a causality of reason, Kant’s point is that,even at the empirical level, the voluntary actions of human beings exhibit a “character”that is distinct from that of merely physical or mechanical causes, since they reflect a setof underlying intentions. These intentions constitute the “subjective principles of thewill.” They are empirical insofar as they can be inferred from overt behavior and usedto explain past actions and predict future ones.The notion of an empirical character therefore involves a deterministic, though not

reductionistic, picture of human agency, and it is this picture to which we appeal whenwe are simply observing human behavior, “and, as happens in anthropology, . . . tryingto investigate the moving causes of his actions physiologically” (KrV A550/B578). It isalso, in all essential respects, the view of such agency affirmed by the Leibnizians andmost forms of compatibilism to the present day. Since agency, so conceived, is itselfpart of the natural order, there is no problem regarding its compatibility with this order.At least for Kant, however, the problem is that, under this assumption, there is also nofreedom.As Kant viewed the situation, freedom is required to account for the “ought” (both

moral and prudential). Since this involves considering human actions normatively inrelation to practical reason rather than descriptively in relation to the conditions of theirexperience and explanation, it requires a different conception of agency, one that

33 In addition to Allison (1990), 29–53, I discuss this distinction and related issues in (1998b), which isincluded in this volume.


allows us to conceive of the will as capable of an absolute beginning. The function ofthe notion of an intelligible character is to provide the requisite conception.Kant’s thesis that one and the same volition may be considered from these two

apparently conflicting points of view and assigned two such characters has beendeemed deeply paradoxical, if not outright incoherent, by many. Of particular concernis Kant’s attempt to illustrate this thesis by means of the notorious case of the maliciouslie. Faced with such an act, Kant suggests, we first enquire into its motive causes andthen seek to determine the degree to which the act and its consequences may beimputed to the agent. In considering the former question, we naturally appeal toexplanatory factors such as “bad upbringing, bad company . . . the wickedness of anatural temper insensitive to shame . . . carelessness and thoughtlessness, as well as toother occasional causes that may have intervened” (KrV A554/B582). In short, it isassumed that the act can be fully explained in terms of a combination of environmentalfactors and character traits. But in spite of this, Kant maintains, we still blame the agent.Moreover, we do not do so on the familiar compatibilist grounds that the act is theconsequence of the agent’s own bad character. Rather, we do so because we presup-pose that

[I]t can entirely set aside how that life was constituted, and that the series of conditions thattranspired might not have been, but rather that this deed could be regarded as entirely uncondi-tioned in regard to the previous state, as though with that act the agent had started a series ofconsequences entirely from himself. (KrV A555/B583)

If one is to avoid reducing Kant’s account to sheer nonsense, this claim must beconsidered with great care. First, we must keep in mind its context, which is that ofa critique of the attempt to conceive imputation solely in terms of a Leibnizian-typeview of agency. As Crusius had already claimed, this view is inadequate because itreduces one’s virtue or viciousness to a matter of luck. But if this is to be avoided, itdoes seem necessary to regard an agent as acting in way that is not determined entirelyby character and circumstance, that is, as capable of initiating an absolute beginning.Second, in spite of Kant’s language, we need not take him as affirming the utterly

implausible view that one’s past behavior, disposition, and circumstances play no role ingoverning one’s voluntary actions, as if one’s present self were discontinuous with one’spast self. This would amount to a form of the liberty of indifference, justly ridiculed bythe Leibnizians and many others. But Kant is not claiming that, all things considered, itwould be equally easy for the liar to speak the truth on that occasion. He is claimingrather that he could have done so. Or, perhaps better, that we must presuppose that hecould have done so, if we are to blame him for the lie.Third, we must keep in mind the status of freedom as a transcendental idea, which,

as such, has no explanatory role. It is not that in some cases we appeal to freedom of thewill in order to explain an action while in others we judge that the agent had no choicein the matter. Certainly, we often distinguish actions in these terms, considering theformer voluntary and the latter involuntary, and Kant has no problem with this

150 PART I I

distinction. His point is rather that it applies only within the explanatory framework ofempirical character and does not touch upon the transcendental question. The latterconcerns the very conception of a voluntary action, insofar as it is deemed imputable.And it is to resolve this question that transcendental idealism is required.Considered as a whole, Kant’s account may be seen as an attempt to reconcile two

apparently conflicting principles: (1) the deterministic principle of the Second Analogy,which holds that every occurrence (including the voluntary actions of rational agents)has an antecedent condition from which it follows according to a rule; and (2) the thesisthat the conception of ourselves as genuine agents to whom actions are imputedrequires the attribution to the will of freedom in a strong (indeterminist) sense.Given this problematic, transcendental idealism is presented as the only hypothesison the basis of which both of these principles can be maintained.Nevertheless, such a resolution appears vulnerable at three points. One is the

coherence of its proposed solution. If transcendental idealism is, as many critics charge,itself incoherent, then appealing to it to reconcile these principles is of no greaterimport than appealing to the concept of a round square would be with regard to thequestion of how a figure can be both round and square. The other two points concerneach of these principles taken singly. For if we abandon (or modify) either thedeterministic principle of the Second Analogy or Kant’s essentially Crusian conceptionof freedom, then the need to appeal to transcendental idealism disappears.Although some defenders of Kant have chosen the first route, which involves the

reduction of the causal principle to a merely regulative status,34 the usual move is totake the second, which amounts to an appeal to some form of compatibilism. Whetherat the end of the day this provides an adequate conception of freedom remains an openquestion that cannot be decided here. What should be clear, however, is that Kantrejected the compatibilist conception of freedom as he understood it. On this issue,Kant stands firmly with Crusius rather than the Leibnizians. Thus, while defending thegeneral thesis that freedom is compatible with causal determinism, which is thedefining mark of compatibilism, he rejected the conception of freedom in terms ofwhich this compatibility is usually understood. This is the source of both the complex-ity and much of the interest of Kant’s account.

V Freedom of the Will in Kant’s SuccessorsThis final section will discuss briefly the concept of free will in three of Kant’s idealisticsuccessors: Fichte, Hegel, and Schopenhauer. Although other thinkers, e.g., Schelling,undoubtedly could have been included, these arguably provide the most interestingcase studies of the development and criticism of Kant’s thought on the topic.

34 Most notably Beck (1987) and (1975), and Korner (1987), 52–8.


Johann Gottlieb Fichte: Whereas for Kant and his predecessors freedom is a problem(albeit a vitally important one) for philosophy, for Fichte it provides the foundation ofphilosophy. This is reflected in Fichte’s characterization of his own position as a“system of freedom,”35 which he also misleadingly describes as nothing more thanthe Kantian philosophy “properly understood.”36

Fichte’s creative reconstruction of the Kantian philosophy is articulated in variousversions of hisWissenschaftslehre and related writings, which he composed in the middleand late 1790s. It is based largely on two principles, each of which breaks withorthodox Kantianism. The first is that the absolute autonomy (independence) of theI is the mandatory starting point of philosophy in the sense that everything is to beexplained in terms of the I and its conception of itself, while this self-conception is notitself to be explained in terms of anything more fundamental.37 The second is that theI is not a thing or substance (a Cartesian res cogitans) but an activity. Specifically, it is theactivity of self-determining or self-positing, which Fichte, following Kant, viewed asessential even to the theoretical use of intelligence. Accordingly, consciousness of self isjust the consciousness of this activity, and the task of philosophy or Wissenschaftslehre isto spell out the necessary conditions and implications of this activity and the conscious-ness thereof.Since in order to posit itself the I must confront an objective world (the not-I),

which opposes and limits its activity, Fichte avoids what he takes to be the misguidedappeal of some Kantians to a pre-given realm of things in themselves.38 Instead, he“deduces” the reality of an external, physical world as a necessary condition of self-consciousness. Rather than a bizarre flight into metaphysical fancy, Fichte’s positionmay be seen as grounded in a radical reinterpretation of Kant’s distinction between twostandpoints and division of the philosophical terrain into transcendental idealism andtranscendental realism. As already noted, for Kant the contrast between the twostandpoints concerns two ways of considering things and events (including humanactions): as they appear under the spatiotemporal conditions of sensibility and as theyare thought through pure reason independently of these conditions. And for Kant atleast one key difference between the two forms of transcendentalism is that the formerallows for the distinction between the two standpoints, whereas the latter does not.Rejecting the appearance–thing in itself distinction as ordinarily understood, Fichte

regards the contrast between the two standpoints as one between the points of view ofthe philosopher and of ordinary consciousness. The latter is inherently and appropri-ately realistic (in the sense of Kant’s empirical realism), with the result that the I is

35 This is from the draft of a letter to Jens Baggessen, April/May 1795. It is cited by Daniel Breazeale in hisintroduction to his edition of Fichte (1994), vii, note.

36 Fichte (1994), 52.37 According to Breazeale, Fichte initially affirms this principle in his “Review of Aenesidumus.” See

Breazeale, editor’s introduction to his edition and translation of Fichte (1988), 14–16.38 In the “Second Introduction to the Wissenschaftslehre” Fichte denied that such a doctrine was actually

asserted by Kant himself. See Fichte (1994), 66–76.

152 PART I I

viewed as a being among beings. This may also be regarded as the naturalisticstandpoint assumed by science. For the philosopher, however (at least the idealisticphilosopher), the mandatory starting point is the I itself, of which the philosopherbecomes aware through a reflection on her own self-determining activity. Thus, it isfrom the standpoint of philosophy, and only from this standpoint, that primacy isassigned to this activity of the I and the objective world is viewed as existing only forand through it.This is closely connected with Fichte’s methodological dichotomy between idealism

and dogmatism, which replaces the Kantian dichotomy between transcendental ideal-ism and transcendental realism. Although clearly modeled on the latter, Fichte’sunderstanding of the fundamental division of the philosophical terrain is orientedmore to the question of the nature and status of the I than to the epistemologicalquestion of the conditions and limits of a priori knowledge. Put simply, whereasFichtean idealism takes the I as starting point, dogmatism, in its various forms, startswith a pre-given world of beings (including human beings), and its project is to explainthe possibility of the I, in both its cognitive and practical dimensions, on this basis. Notsurprisingly, Fichte asserts that this project fails because dogmatism cannot account forthe possibility of the I as self-reverting activity, whereas idealism, starting with thelatter, can account for the experience of an objective world of things with which thedogmatist begins and which defines the standpoint of ordinary consciousness.Even though the theoretical portion of the Wissenschaftslehre is devoted entirely to

demonstrating the latter thesis, Fichte readily admits that the dogmatist will never beconvinced by its argument.39 More generally, he held that the conflict betweenidealism and dogmatism is irresolvable at the theoretical level. Nevertheless, Fichtethought that he could overcome this impasse through a radicalization of the Kantianprinciple of the primacy of practical reason. He does this by insisting that all reason is atbottom practical, which for Fichte means that practical considerations, that is, thoseconcerning the conditions of the possibility of the I as self-reverting activity, constitutethe ultimate court of appeal in philosophy.Consequently, while Fichte agrees with Kant in denying the possibility of a theoretical

proof of freedom, he has quite different reasons for doing so. One of these is the statusassigned to freedom or self-determination, which, as inseparable from the thought of theI, itself serves as a first principle of philosophy and, as such, cannot be demonstrated.Although Kant took freedom (in the sense of absolute spontaneity) to be inseparable fromthe thought of the I and even, in proto-Fichtean fashion, at one time used this as the basisfor a demonstration of freedom, he never took the I as the first principle of philosophy inanything like Fichte’s sense. Another, evenmore un-Kantian, reason is Fichte’s pragmatic,even proto-existentialist, orientation. Anticipating themes developed in the past century,Fichte, with his doctrine of self-determination, not only regarded the existence of an I as

39 Fichte (1994), 15–16.


prior to its essence (what the I makes of itself), he also seems to have divided all people(including philosophers) into two classes: those who affirm and those who attempt todeny their freedom. Accordingly, he suggests that the kind of philosophy one adopts(idealism or dogmatism) reflects the kind of character one has.40 Idealists affirm theirfreedom, understood as the act of self-determination, while dogmatists deny it byconceiving of themselves as determined rather than as self-determiners. Thus, dogmatismis seen not merely as a defective philosophy, but as the sign of a character defect as well.41

Considered from the practical point of view, this self-determining activity takes twoforms and involves two conceptions of freedom, which correspond roughly to Kant’sdistinction between spontaneity and autonomy. In Fichte’s preferred terminology, theformer is characterized as “formal” and the latter as “material” or “absolute” free-dom.42 Moreover, each is made the central topic of a distinct work.Formal freedom is the concern of the Foundations of Natural Right (Grundlage des

Naturrechts nach Principien der Wissenschaftslehre) (1796), which contains the systematicstatement of Fichte’s legal and political philosophy. In the spirit of theWissenschaftslehre,he attempts to “deduce” such freedom as a necessary condition of self-consciousness.The work is concerned, however, with a specific form of self-consciousness, namely,that of oneself as a particular individual with determinate desires and ends. Since thisinvolves a conception of oneself as an end-setter, it is inseparable from the conscious-ness of one’s capacity to set ends and to strive to realize them in the external world.Fichte’s key claim here is that this consciousness is possible only insofar one findsoneself as a finite rational agent among others. As he puts it at one point, “The humanbeing . . . becomes a human being only among human beings.”43 Fichte develops thisthought in connection with his conception of a “summons” (Aufforderung), which canstem only from another rational being. Since it is only through such a summons to door refrain from a certain course of action that I can be aware of my capacity to choose, itbecomes a necessary condition of my awareness of myself as a free, self-determiningindividual. But since in being aware of a summons one must also be aware of the freeagency of the summoner, it follows that I can consider myself as free only insofar asI consider other finite rational beings as free in precisely the same sense, that is, as self-determining end-setters or free individuals.Nevertheless, Fichtean formal freedom is a limited conception of freedom, which fails

to do full justice to the self-determining activity of the I. This is because the kind of self-determination required for formal freedom consists merely in the setting of one’s ownends as a rational agent. Confronted with alternative courses of action, I myself choosewhich one to adopt. Although equivalent to freedom of choice as traditionally under-stood, from the Kantian standpoint it provides a merely heteronomous conception offreedom. The problem is that, while attributing to rational agents a certain independencefrom their desires and a capacity to determine which ones to act upon (spontaneity), it

40 Fichte (1994), 20. 41 Ibid., 90.42 I am here largely following the account of Neuhouser (1990), 121–31. 43 Fichte (2000), 37.

154 PART I I

leaves this choice as arbitrary rather than as norm-governed. Expressed in contemporaryterms, the limitation of this conception of freedom as the basis for an adequate under-standing of self-determination is that it does not “go all the way down.” One maydetermine on which desires one chooses to act, but if one does not also determine theprinciples governing one’s choice, one is not fully self-determined.Fichte’s account of material or absolute freedom, which is developed in The System of

Ethical Theory (Das System der Sittenlehre nach dem Principien der Wissenschaftslehre) (1798),takes the form of a reflection on the conditions of a complete self-determination (onethat does go all the way down). Consequently, it is based on the questionableassumption that the latter is intelligible, that one can speak meaningfully of the I asconstituting or determining itself, as it were, out of whole cloth. Fichte’s basic claim,which provides the foundation of his ethical theory, is that such self-determinationrequires governing one’s choice of maxims by a self-legislated principle. It turns out,however, that the only principle that qualifies in this respect is the demand todetermine our freedom solely in accordance with the idea of self-determination.44

Or, as he also puts it, “I am supposed to be a self-sufficient I.”45 This amounts aradicalization of the Kantian principle of autonomy, as the result of which autonomy,understood as complete independence of any thing or value that is not rooted in theI and its self-determination, is reconceived as an infinite task rather than as a constitu-tive feature of the will. As moral agents, we are obligated to strive to attain fullautonomy or self-determination even though, in virtue of our finitude, this cannever be completely attained.46

Georg Wilhelm Friedrich Hegel: Like Fichte, Hegel granted a foundational role to theconcept of freedom and equated it with self-determination. But whereas Fichte under-stood the latter as an apparently groundless act of self-positing, through which theI supposedly constitutes both itself and its other (the non-I), Hegel understood it inmore “concrete” terms as a “being-with-oneself-in-an-other” (Beisichselbstsein in einemAnderen).47 In terms of Hegel’s dialectical logic, this means that freedom is attainedthrough an overcoming of the otherness of the other, by which its otherness is negated,while its being is preserved. AsHegel shows graphically in his account of the life and deathstruggle in the Phenomenology, the latter is necessary because a simple, “abstract” negation(killing the other), does not leave an other in whom one’s freedom can be actualized.48

Consequently, the other must be negated in a way that preserves its being, that is, it mustbe “superseded” (aufgehoben). The famous master–slave dialectic in the Phenomenology,which immediately succeeds this struggle, is presented as the first and inherentlyflawed attempt to attain freedom so understood. Since the other (the slave) is consciousmerely of his total dependence on the master, that is, of his unfreedom, the master isnot able to find his freedom fully actualized in the slave’s consciousness. The basic idea,

44 Fichte’s Werke, IV, 59. 45 Fichte (2005), 212. 46 Ibid., 66, 149, 229.47 Hegel (1991), 54 (}23). The following discussion is indebted to Neuhouser (2000), 18–54.48 Hegel (1977), 109–10 (}175–}177).


which has clear affinities to Fichte’s conception of a summons, is that one can find one’sfreedom only if it is freely (not slavishly) recognized by the other. Moreover, since only afree being can freely recognize the freedom of another, this means that no one is fully freeunless all are free. Such a condition of universally recognized freedom is for Hegel thegoal of history, which, rather than being the infinite task it was for Fichte, Hegel believedto have been already attained (at least in principle) in the laws and institutions of post-revolutionary Western Europe.49

In the Philosophy of Right, this conception of freedom, the full attainment of which isidentified with humanity’s or “spirit’s” self-realization, is applied to an analysis of thehuman will. Thus, it is here that Hegel’s “speculative”-historical account makescontact with what is generally regarded as the “free will problem.” Although thisaccount involves the usual Hegelian obscurity, the basic goal is to analyze the problemin light of the conception of freedom as being-with-oneself-in-an-other. This analysisbegins with the concept of will, which, like Kant, Hegel identifies with practical reasonor intelligence.50 Also like Kant, Hegel asserts that the will, so conceived, is inseparablefrom freedom. Accordingly, he claims that “freedom is just as much a determination ofthe will as weight is a basic determination of bodies.” And, again, that “Will withoutfreedom is an empty word, just as freedom is actual only as will.”51

Hegel differs from Kant and stands much closer to Fichte in his understanding ofpractical reason. Whereas Kant begins with a separation of theoretical and practicalreason and endeavors to unite them, Hegel rejects the Kantian dichotomy and beginsinstead with the idea of their inseparability. According to Hegel, neither will norintelligence (thought) are possible apart from one another and the difference betweenthem is simply between theoretical and practical attitudes. Thus, on the one hand, willis itself merely “a particular way of thinking—thinking translating itself into existence,”while, on the other hand, it is only in thinking that one is with oneself and can,therefore, find oneself in another, which is the goal of the will as free.52

This interpenetration of thought and volition leads, in turn, to the most distinctivefeature of the Hegelian conception of a free will, namely, its unification of apparentlyconflicting conceptions into a single concrete idea or “concept.” In light of thisconcept, Hegel provides what amounts to a rational reconstruction of the concept ofa free will in the Introduction of the Philosophy of Right, which he then uses in the bodyof the work as the basis for an analysis of the distinct spheres of right (Abstract Right,Morality [Moralität], and Ethical Life [Sittlichkeit]), each of which is viewed as theactualization of a particular dimension of concrete freedom.The elements of this reconstruction are a set of interrelated dichotomies, which

express in somewhat different terms what may be described as the subjective andobjective poles of the concept of freedom. These include universality (or negativefreedom) and particularity; infinitude and finitude; form and content; being for-itself

49 This is the central theme of the various versions of Hegel’s Lectures on the Philosophy of World History.50 Hegel (1991), 35–7 (}4). 51 Ibid. 52 Ibid.

156 PART I I

and being in-itself. In each case, the first element stands for the moment of indepen-dence or indeterminacy. Thus, universality, or negative freedom, represents the inde-terminacy through which consciousness stands apart from and above its particularcontents. This indeterminacy is also expressed in the idea of the infinitude of thewill, understood as its opposition to everything finite (including an agent’s drives andinclinations). Similarly, this represents the “formal” side of willing, in contrast to theparticular content chosen. Finally, it is also the free will as it is for-itself in contrast tothis will as it is in-itself, that is, in its inherent nature, which consists in willingsomething determinate rather than in simply remaining indeterminate.Since the first or subjective side of the dichotomy corresponds to what is usually

thought to be the full or adequate characterization of free will, at least on an indeter-minist account, Hegel argues, in effect, that this sense of adequacy is illusory becausethe conception it embodies is inseparable from, and dependent upon, its polar oppo-site. Just as in the master–slave dialectic of the Phenomenology, the universality of the willturned out to be dependent upon its particularity, since the latter is the source of thecontent to be willed, the infinitude dependent upon its finitude, and so forth. Thelesson drawn from this is that an adequate understanding of freedom must integrateeach of these opposing moments. In other words, the “truth” of freedom (in contrast toits certainty) involves both universality and particularity, infinitude and finitude, formand content, and the will as it is both for and in itself. Thus, these first elements, whichmost previous philosophers have seen as both necessary and sufficient for freedom, areviewed by Hegel as merely necessary conditions, which, if not combined with theirdialectical opposites, constitute merely the appearance or form of freedom.This provides the justification for Hegel’s definition of freedom as being-with-

oneself-in-an-other, since it supposedly shows that a freedom that does not somehowincorporate otherness is nothing more than an empty abstraction. Accordingly, theproblem is to understand how otherness can lose its character of simple unfreedom andbecome dialectically transformed into an essential ingredient in freedom. Hegel’s claimis that this is possible only if this otherness is itself an expression of freedom. Only thendo we find freedom fully actualized as it is in-and-for-itself. Or, as Hegel also puts it,“The will in its truth is such that what it wills, i.e., its content, is identical with the willitself, so that freedom is willed by freedom.”53

By understanding freedom in this way, Hegel may be said to have changed thesubject, which is why here, as in other areas of philosophical inquiry, it is so difficult tojuxtapose his views in a straightforward way to those of other thinkers. Thus, ratherthan worrying, as previous philosophers (including Kant and Fichte) had done, aboutthe reconciliation of freedom with natural causality, Hegel’s analysis focuses on therelation between a formally free choice and its content. Although Hegel would nodoubt agree that in a sense he has changed the terms of the debate, he would also

53 Ibid., 52–3 (}21 addition).


contend that this is the result of a dialectical analysis of the inadequacy of the way inwhich the problem has traditionally been framed. Indeed, he might suggest that thisinadequacy consists precisely in an exclusive focus on the issue of the causal indetermi-nacy of a putatively free choice, thereby neglecting the substantive issue of the contentof such a choice. His position seems to be that if the latter is ignored, one will be leftwith nothing more than an empty, abstract freedom, which is not worthy of the nameand which corresponds to the so-called “liberty of indifference” that was dismissed bythe Leibnizians and many others in the rationalist tradition.Our present concern, however, is with Hegel’s use of this analysis against Kant.

According to Hegel, in virtue of his commitment to the categories of abstract under-standing, which reflects a failure to attain Hegel’s own speculative standpoint, Kant wasled to conceive of freedom merely as “arbitrariness” (Willkür) or as a “formal self-activity,” by which Hegel understood simply the freedom to do as one pleases.Although the most common idea of freedom, it is also the least adequate, since byviewing the content of the will’s choice (provided by competing drives and inclination)as given to it from without, it reduces the will’s freedom to a mere contingency. Thelatter is a moment of genuine freedom, but Hegel suggests that it is a delusion to take itas equivalent to freedom. And Kant, like all advocates of a “reflective,” i.e., non-speculative, philosophy is deemed subject to this delusion.54

Nevertheless, whatever the merits of Hegel’s positive account, he seems to haveseriously misrepresented Kant’s view of free will. Expressed in Kantian terms, whatHegel is doing is characterizing Kantian freedom solely in terms of freedom of choice,thereby ignoring the intimate connection between Willkür and Wille.55 In fact, prop-erly understood, freedom for Kant does not consist in the sheer arbitrariness of choice,but in a choice governed by rational norms stemming from Wille or practical reason.Much as in Hegel, then, genuine Kantian freedom may be seen as involving a“synthesis” of form and content, with the former stemming from Willkür and the latterfromWille. The basic difference consists in the location of the source of these norms: inthe autonomous pure practical reason of the agent for Kant, and in the objectivelyexisting laws and institutions of a society for Hegel. Although this is an importantdifference, it is one that cannot be considered here.

Arthur Schopenhauer: Schopenhauer’s views on free will are of interest here for threereasons. The first is his uncompromising determinism with regard to particular actions,on the basis of which he dismisses a compatibilism such as Leibniz’s as an inadequate,even duplicitous “middle way,” which endeavors to preserve the term “free will,”while emptying it of any sense.56 The second is the connection of this determinismwith a radical voluntarism, which effectively identifies ultimate reality (under the guiseof the Kantian thing in itself) with will. The third is his use of Kant’s contrast between

54 Hegel (1991), 48–9 (}15).55 This misrepresentation is pointed out by Wood in his editorial notes to Hegel (1991), 399 note 2.56 Schopenhauer (1960), 15.

158 PART I I

empirical and intelligible character, which Schopenhauer describes as “the greatest ofall achievements of the human mind,” to offer an alternative conception of freedom.57

Schopenhauer’s determinism is based on his appeal to the principle of causality,which he regards as one of the four distinct forms assumed by the principle of sufficientreason. As in Kant, nothing happens without a cause from which it follows necessarily.But rather than recognizing only the familiar mechanical, physical, and chemicalcauses, Schopenhauer assigned the causes of human actions to its motives insofar asthey determine the will.58 Accordingly, Schopenhauer insisted that motives are everybit as much causes as any others, since they involve the necessitation of their effects. Buthe also thought that this is not always recognized because of a confusion of the freewith the voluntary.The reason for this confusion is traced to self-consciousness, which makes one aware

of a capacity to do as one wills: to choose either A or B, if one so wills. Schopenhaueracknowledged the genuineness of this awareness, but rejected its identification with aconsciousness of freedom of the will. The latter concerns a supposed capacity to willwhat one wills, which he rules out on the grounds that it either directly violates theprinciple of sufficient reason (since it assumes that there could be a choice without areason) or leads to an infinite regress, whereby an agent must be thought as willing towill, ad infinitum.Although Schopenhauer regarded motives as causes, he did not view them as alone

sufficient to determine the will. Since they always work in conjunction with character,the same motives could lead people with different characters to act in different waysunder similar circumstances. Nevertheless, a person with a particular character willalways act in the same way under the same circumstances. Thus, to claim that I (HenryAllison) could have acted differently is to say that I could have been a different person.Unfortunately, this is impossible, since one can change one’s behavior but not one’scharacter, even though we can only discover this character after the fact by consideringwhat we have done. Like many other thinkers, then, Schopenhauer located theultimate determining ground of voluntary actions in a person’s character. It is the latterthat determines by what motives a person can be moved to act and to what extent thatperson is susceptible to these motives.Perhaps the most striking feature of Schopenhauer’s account lies in his virtual

identification of character and will, which he viewed as the true core of the self. Inother words, Schopenhauer is a voluntaristic determinist.59 Like other voluntarists, forexample, Crusius, he granted primacy to the will over the intellect. Indeed, Schopen-hauer regarded the latter merely as an instrument or tool of the former, in the sense thatits function is to determine which course of action is best suited to attain the ends

57 Schopenhauer (1965), 110. 58 Schopenhauer (1974), 70–2, and (1960), 36–7.59 Although some subsequent philosophers, most notably Nietzsche, who was himself deeply influenced

by Schopenhauer, fit this category, most of his predecessors on this score appear to have been theologians, forexample, Luther, intent upon defending the doctrine of original sin.


projected by the will.60 Unlike most voluntarists, however, he used this to denyfreedom of the will. In fact, according to Schopenhauer, it is precisely the prioritizingof intellect to will in philosophers such as Descartes and Leibniz that led to theirerroneous doctrines of free will. His point is that by making will subordinate tointellect, and even reducing volition to an act of thought, these intellectualists effec-tively reduced the question of what one is to what one knows, thereby assuming thatby increasing one’s knowledge one can change one’s character.61 Schopenhaueracknowledged that Spinoza did not come to such a conclusion, but he describedhim as a philosopher who reached the correct (deterministic) conclusion from falsepremises.62

Consequently, Schopenhauer’s denial of freedom of the will rests not only on theuniversal scope of causality and the conception of motives as causes, but also on his coredoctrine of the unchangeableness of character. According to Schopenhauer, our(empirical) character is something with which we are born and cannot change.63

Greater knowledge and experience may change how we act under given circumstancesby making it clear that a certain course of action is good or harmful for us, but it doesnot fundamentally change who or what we are.Nevertheless, Schopenhauer reconceived rather than denied freedom and, as noted

above, did so by appealing to Kant’s distinction between empirical and intelligiblecharacter. According to his reading of Kant, a person’s empirical character (or will asphenomenon) is the necessary consequence of a timeless choice by the intelligiblecharacter (or will as noumenon). The choice must be timeless in order to be freebecause the principle of sufficient reason governs what occurs in time, and such atimeless choice is conceivable because time pertains merely to the phenomenal realm.Although Schopenhauer attributed this doctrine to Kant, he also suggested that it wasanticipated, albeit in a mythopoetic manner, by Plato in his famous “Myth of Er” in theRepublic.64

Schopenhauer apparently also thought that this doctrine of a timeless choice ofcharacter was required to do justice to our sense of moral responsibility. Even thoughwe may know that our particular actions are determined by a combination of characterand circumstances and that this character is itself fixed before birth, we nonethelessrightly hold ourselves responsible for our deeds.65 But given the manner in which theissue is framed, the only way to save the freedom requisite for responsibility (“truemoral freedom”), is to regard ourselves as responsible for who we are, that is, for ourcharacter. And since the latter is innate and unchangeable, this means that we mustconceive of our empirical character as the result of a timeless choice.Although Schopenhauer, like Fichte, presented his view as a reformulation of Kant’s

position correctly understood, it is doubtful that Kant would have countenanced it. In

60 See, for example, Schopenhauer (1960), 101. 61 Schopenhauer (1958), vol. 1, 292.62 Ibid., 298. 63 Schopenhauer (1960), 54–5.64 Schopenhauer (1965), 113–14. 65 Schopenhauer (1960), 91–9 and (1965), 112–13.

160 PART I I

fact, in spite of some indications to the contrary, it is clear from his example of themalicious lie that Kant would have rejected Schopenhauer’s thesis that one choosesone’s empirical character out of whole cloth and that freedom consists entirely in thischoice. For Kant it is not only the case that such a person could have chosen not to be aliar but also that, given both his character and circumstances, he could at that momenthave chosen not to lie. Thus, Kant, unlike Schopenhauer, wished to preserve thefreedom of particular acts. An agent must be deemed able to have chosen differentlyunder the same circumstances because the categorical imperative dictates that he oughtto have done so. But since Schopenhauer rejected Kant’s conception of morality asbased on a categorical imperative, arguing instead for an ethics of sympathy, he hadneither need nor room for the Kantian conception of a free action.


PART III

Essay Ten

Is the Critique of Judgment“Post-Critical”?

Dualism, formalism, subjectivism: these for Hegel constituted the unholy trinity of theKantian philosophy, the grounds for characterizing it in Glauben und Wissen as a“philosophy of reflection,” as opposed to a speculative philosophy, which is alonecapable of doing justice to the nature of reason.1 In this samework, however, Hegel alsoclaimed to find in the third Critique, particularly in the conceptions of the reflectivepower of judgment and the intuitive understanding, an anticipation of the speculativeprinciple of the identity of subject and object, thought and being. Moreover, thisprivileging of the third Critique is not merely a passing phase of the young Hegel; forin his later writings he continued to regard the intuitive intellect as a genuine anticipa-tion of the concrete universal and he attributed to it the functions of grounding thepurposiveness of nature, the beautiful in art, and the ultimate unity of nature andfreedom.2 To be sure, despite his positive appraisal of certain aspects of the thirdCritique,Hegel remained highly critical of Kant’s procedure in that work as a whole, suggestingthat he failed to grasp the implications of his own insights and thus lapsed back into adisastrous subjectivism. Nevertheless, the fact remains that for both the young and themature Hegel the last of Kant’sCritiqueswas seen as post-critical in essential respects andas pointing the way to a new, more adequate mode of philosophizing.Recently this Hegelian or “speculative” reading of the third Critique has been

reaffirmed by Burkhard Tuschling. What is particularly interesting about Tuschling’sapproach, however, is that he does not impose the Hegelian critical apparatus on Kantin an external way. Instead, he argues that even from “the classical or authentic Kantianstarting point of the understanding,” it can be shown that “Kantian TranscendentalIdealism involves elements of the Hegelian or speculative understanding.”3 He alsoclaims that Kant himself came to recognize this, not only in the Opus postumum, butalready in the third Critique. On his view, by 1790 Kant was well aware of “substantial,

1 Hegel (1968), 325–46.2 See especially, Hegel (1959), }55 and (1971), 372–82. For a helpful discussion of these texts and the

whole issue of Hegel’s understanding of Kant’s conceptions of purposiveness of nature and the intuitiveintellect, see Düsing (1990), and Baum (1990).

3 Tuschling (1992), 111. 1 shall here be focusing primarily on Tuschling’s views in this article and in his(1990).

if not fatal shortcomings of his Transcendental Idealism” and intentionally made thisknown to his contemporaries.4

According to Tuschling, Kant’s innovations in the third Critique are to be seenwithin the context of his ongoing and never satisfactorily resolved struggle with theproblem of integrating the empirical, contingent features of particular experience into asystem of knowledge grounded in the a priori principles of the understanding.5 Thisstruggle, which culminates in the final reformulation of transcendental idealism in theOpus postumum, is a direct result of Kant’s break with the theocentric paradigm of therationalists and their ideal of a mathesis universalis grounded in the intuitive intellect ofGod. That is to say, it is a consequence of what is usually referred to as Kant’s“Copernican revolution” in philosophy, which consists in the replacement of thisdivine understanding with the discursive human understanding as the “lawgiver tonature.” This revolution is the source of all the dichotomies (e.g. form–matter,sensibility–understanding, a priori–a posteriori, analytic–synthetic, appearance–thingin itself), which characterize the Kantian philosophy and therefore of its deepest andmost intractable problems, paramount among which is the incompatibility of thecontingency of the brute, given features of experience with the universal lawfulnesssupposedly imposed by the understanding.The claim, then, is that Kant attempted to deal with this and related problems in the

Introduction to the third Critique through an appeal to a new “transcendental princi-ple,” the purposiveness of nature, which is itself founded on the concept of an intuitiveunderstanding. Moreover, this new principle is not introduced merely to ground thepossibility of an additional mode of experience (aesthetic), but rather to secure the veryresults that were supposedly established in the first Critique. In fact, Tuschling claimsthat without this principle, “the whole system of transcendental concepts and principles,centered around the concept of the original synthetic unity of apperception, wouldcease to be applicable.”6 Tuschling further contends that Kant’s appeal to this newprinciple entails the abandonment of the most basic assumption of the initial criticalposition, namely, “that the discursive human understanding is the final, legislative andsupreme ground of the unity of nature.”7 Accordingly, this new principle is “superor-dinate to apperception, categories and the principles of pure understanding,”8 whichgives to the new transcendental deduction (presumably of the principle of purposive-ness) a “speculative character” in the sense indicated by the young Hegel.9

It is this claim of radical revision that I propose to examine here. The question is notwhether there are any significant innovations or developments in the third Critique.I think it obvious that there are. Nor is it whether one can profitably read the thirdCritique through Hegelian spectacles in order to understand the central themes andclaims of German idealism.10 This, too, is readily conceded. At issue is only whether

4 Tuschling (1992), 112. 5 See Tuschling (1990), 174–80.6 Tuschling (1992), 115. 7 Ibid., 121. 8 Ibid., 119.9 Ibid., 118–19. 10 For an example of this approach, see Pippin (1995).

166 PART I I I

the third Critique contains an abandonment of the basic commitments and principles ofthe first. I shall argue that it does not and that the later work is best seen as buildingupon rather than attempting to reconstruct Kant’s original “critical” edifice.

IIn claiming that the purposiveness of nature is a new transcendental principle, firstintroduced in the third Critique, Tuschling is obviously assuming its absence from thefirst and in this respect at least he seems to be in agreement with Rolf-Peter Horstmannand others. Although Horstmann emphasizes the extreme ambiguity of the argumentof the Appendix to the Transcendental Dialectic in which Kant takes up the topics ofsystematicity and purposiveness, he insists that both the idea of the systematic unity ofknowledge and the purposiveness of nature (understood as a specific form of systematicunity) are there regarded as logical rather than transcendental principles.11 Similarly,Paul Guyer has argued that for the Kant of the first Critique all of the genuinelytranscendental work is accomplished in the Analytic, where it is shown that thecategories and the principles based upon them are themselves sufficient to groundthe unity of experience, even at the empirical level, that is, to ensure the applicability ofempirical concepts and laws to whatever may be given in sensible intuition.12 Guyeradmits that there are indications of a different view in the Appendix, since Kant thereseems to grant an essential regulative role to reason’s idea of systematicity. But he insiststhat Kant is there not really serious about such a position and that in the end he retreatsto the view that systematicity is a desideratum pursued by reason, which does not standin any essential connection with the operations of the understanding.13

By contrast, my own view is that the Analytic attempts to accomplish much less thanGuyer suggests and leaves ample room for the assignment of genuine transcendentalstatus to reason’s principle of systematicity.14 Consider Kant’s significant, yet generallyignored remark in the Second Analogy that he is concerned with determining merelythe “formal conditions of empirical truth” (KrV A191/B236). I take this to mean thatthe Transcendental Analytic is concerned mainly with the determination of a set of apriori rules (or epistemic conditions), which determine what can count as an object ofpossible experience. As “formal,” these rules serve to preclude certain (logically)possible scenarios, such as changes that are not alterations of enduring substances oralterations without causes. Consequently, they may be said to provide the norm towhich all empirical laws must conform in order to count as laws (KrV A128); for, asKant himself notes, no cognition can contradict them without “at once losing allcontent, that is, all relation to any object, and therefore all truth” (KrV A62–63/B87).Indeed, that is why they can be used as a “canon for passing judgment upon the

11 Horstmann (1989), esp. 165–8. 12 See Guyer (1990a), 29–30 and (1990b), 224.13 Guyer (1990a), 33–4 and (1990b), 227–8.14 For my fullest treatment of this topic, see Allison (2004), 423–37.

I S THE CRIT IQUE OF JUDGMENT “POST-CRITICAL”? 167

empirical employment of the understanding” (KrVA63/B88). It does not follow fromthis, however, that they suffice to ground any particular causal law or even to guaranteethat such laws are there to be found.15 In Kantian terms, that would give them thestatus of an organon rather than merely that of a canon.16

Armed with this view of the Analytic, the claim of the Appendix to the Dialecticthat reason has its own transcendental principle of systematicity seems somewhat lessmysterious. But before considering that matter it must be emphasized that we aredealing with what is purported to be a genuinely transcendental principle. To be sure,Kant introduces the “systematic unity of the manifold of [empirical] cognition” asprescribed by reason as a “logical principle,” the function of which is to assist theunderstanding by means of ideas in order to secure “unanimity among its various rulesunder one principle” (KrV A648/B676). And, considered as such, it is only subjective-ly necessary. In order to avoid a complete misunderstanding of Kant’s position,however, it is crucial to keep in mind that this characterization is presented as aprovisional conclusion from the initial account of the hypothetical employment ofreason, which is quite different from proclaiming tout court that it is in fact merely alogical principle. Thus, there is no contradiction involved when, after a brief analysis ofthe regulative function of the idea of a fundamental mental power, Kant goes on toargue that this logical principle, as articulated in the complementary demands forhomogeneity, specificity, and affinity, presupposes a transcendental principle (KrVA654/B682).But if there is no contradiction, there is certainly a major shift in emphasis, indeed,

one that has seemed opaque to many commentators. For Kant proceeds to claim that,as transcendental, this principle must be assumed to be necessary and objectively valid(KrV A651/B679), or at least to possess “objective but indeterminate validity” (KrVA663/B691).17 In short, the claim is that this principle is more than a mere methodo-logical recommendation to look for homogeneity, specificity, and affinity; that despiteits regulative function, it involves a presupposition about what is to be encountered inexperience. And this raises the puzzle about how a principle could have both transcen-dental status and a merely regulative function.

15 Admittedly, there is a sense in which the existence of causal laws is guaranteed by the causal principleitself or the mere concept of causality. As Michael Friedman has argued, since the causal relation for Kantinvolves strict universality and necessity, to say that A is the cause of B is to affirm a universal law to the effectthat “all events of type A are necessarily followed by events of type B.” See Friedman (1992). But it hardlyfollows from this that in experience we shall ever encounter more than a single event of each type andtherefore be in a position to discover empirical causal laws. In short, for all that can be inferred from theconcept of causality, the possibility remains open that there might be nothing more than what could betermed “instantaneous laws,” that is, laws with merely a single instance. Moreover, such “laws” could neverbecome empirically known as such. This latter point has been put clearly by McFarland (1970), 8–11.

16 On the distinction between a canon and an organon, see KrV A21/B26, A61/B85, and especially A63/B88, where Kant states that the “transcendental analytic should be used only as a canon for passing judgmentupon the empirical employment of the understanding . . . ”

17 See also KrV A656/B684, A660/B688, A664/B692, and A680/B708. Kant’s position is that each ofthe three maxims or logical principles presupposes a corresponding transcendental law.

168 PART I I I

The transcendental status assigned to this principle is based on its alleged necessitywith respect to empirical inquiry. In support of this necessity, Kant claims that withoutthe law requiring us to seek systematic unity, “we would have no reason, and withoutthat, no coherent use of the understanding, and lacking that, no sufficient mark ofempirical truth” (KrV A651/B679). And, in a similar vein, he remarks:

If among the appearances offering themselves to us there were such great a variety . . . that even themost acute human understanding, through comparison of one with another, could not detect theleast similarity (a case which can at least be thought), then the logical law of genera would not obtainat all, no concept of a genus, nor any other universal concept, indeed no understanding at all wouldobtain, since it is the understanding that has to do with such concepts. (KrV A654/B682)

In considering these (and similar) texts, it is instructive to contrast the characterizationof the transcendental principles of the Analytic as “formal conditions of empiricaltruth”with the Appendix’s view of systematicity as that without which we would have“no sufficient mark of empirical truth.”18 Clearly, Kant is not claiming that systema-ticity is, of itself, a sufficient mark or criterion of empirical truth, as if the systematicembeddedness of an empirical generalization or law in an overarching theory or set oflaws were sufficient to guarantee its truth. The claim is rather that systematicity isnecessary in order to have a sufficient mark of empirical truth and therefore a coherentemployment of the understanding or, as the second passage suggests, virtually anyvalid employment of the understanding at all.Unfortunately, as is frequently noted, Kant offers precious little in the way of argument

in support of these bold claims. But setting that large issue aside for the moment, I wouldlike simply to note several points that have emerged from our brief consideration of theAppendix to the Dialectic. First, as is clear from the passage cited above, already in the firstCritique Kant links systematicity with the possibility of forming empirical concepts;indeed, that is why it is necessary for understanding. Accordingly, we cannot acceptHorstmann’s claim that this is an insight unique to the third Critique.19

Second, given Kant’s equation of experience with empirical cognition, it followsfrom this that systematicity must in some sense be regarded as a necessary condition ofthe possibility of experience. This does not conflict with the Analytic, however, sincethis principle conditions experience in a different way and at a different level than thecategories and principles of the understanding.20 As formal conditions of empiricaltruth, the latter or, more specifically, the Analogies, serve as rules for the unification ofappearances in a single time. As such, they collectively condition experience in the

18 The contrast is noted by Mertens (1975), 38.19 See Horstmann (1989), 169–70. Although Horstmann connects empirical concept formation with

purposiveness rather than systematicity and notes that purposiveness has a rather restricted meaning in the firstCritique, I do not think that this affects the main point at issue.

20 The importance of paying attention to the different levels of Kant’s analyses, particularly with respect tothe contributions of understanding and reason, has been emphasized by Gerd Buchdahl. See particularly,Buchdahl (1969a), esp. 651–65, (1969b), and (1974).


sense of determining the forms of the thought of an objective temporal order (hencetheir “formal” status). For the same reason, they are also conditions of objects ofexperience, which accounts for their constitutive status. By contrast, systematicity isconcerned with the connectability, and hence the coherence, of the first-order empir-ical claims regarding objects falling under these formal conditions.21 Consequently,systematicity is not itself a direct condition of objects of experience, which is why it has aregulative rather than a constitutive function.Third, and most important, this regulative function is not incompatible with a

transcendental status. The common view that it is stems from the erroneous assumptionthat “regulative” means something like “merely heuristic” or optional. This is errone-ous, because the whole purpose of the Appendix is to argue for the indispensability ofreason and its ideas with respect to the empirical use of the understanding. To claimsuch indispensability is to claim a transcendental status, if not a constitutive function.And that is why the analysis culminates in the transcendental deduction of the ideas,which Kant describes as “the completion of the critical business of pure reason” (KrVA670/B698).22

IITurning from the Appendix to the Dialectic of the first Critique to the two versions ofthe Introduction to the third, we find both some common ground and some significantdifferences. The basic difference is the shift in focus from the hypothetical use of reasonto the reflective capacity of the power of judgment. In the first Critique, Kant does notseem to have recognized that this power might have such a capacity or function, whilein the third it is the main topic, providing the very raison d’etre for a critique. Also, thereflective power of judgment’s peculiar principle, the purposiveness of nature, is farbroader than reason’s concept of systematicity. And, perhaps most significantly, the firstCritique’s vague appeal to an “indeterminate objective validity,” which is claimed firstfor systematicity and later for the ideas, is replaced by the explicit acknowledgment ofthe subjective, reflective character of judgment’s principle of purposiveness. As Kantputs it in both versions of the Introduction, it is a matter of the “heautonomy” ratherthan the autonomy of the power of judgment, of this power legislating merely to itselfrather than to nature (see EE 20: 225; KU 5: 185).23

21 This reflects Kant’s view that the proper object of reason is the understanding, which is a central tenet ofthe Dialectic. See, for example, KrV A306–7/B363, A321/B378, A326/B383, A327/B384.

22 I take the goal of this deduction, which Kant admits is quite distinct from the deduction of thecategories, to be to establish a necessary connection between the three official transcendental ideas andreason’s governing idea of systematicity. Each of these ideas posits the “object in idea,” which must benecessarily presupposed in the quest for homogeneity, specificity, and affinity. This also explains theconnection between the two parts of the Appendix, a topic which is seldom discussed. I analyze thisdeduction in Allison (2004), 437–44.

23 The introduction of the notion of heautonomy is emphasized by Mertens (1975, 6) as the majordifference between the Appendix to the Dialectic and the third Critique.

170 PART I I I

Nevertheless, “logical purposiveness,” the form of purposiveness with which Kant isinitially concerned in the Introductions, does correspond to reason’s principle ofsystematicity. In fact, he explicitly links it with what in the first Critique were describedas the principles of homogeneity, specificity, and affinity (KU 5: 185). Moreover, Kantis emphatic regarding the transcendental status of this principle. Thus, in an importantfootnote in the first Introduction, Kant affirms that, despite its apparently logical andtautological character, this principle is both synthetic and transcendental, since itspecifies the “condition of the possibility of the application of logic to nature” (EE20: 211–12 note). And in the second Introduction, he insists that this principle requiresa transcendental deduction (KU 5: 182).Unfortunately, Kant is not so clear in his accounts of the epistemic function of this

principle. In particular, it is not clear whether it concerns the possibility of unifyingempirical laws into a system (theory construction), of formulating empirical laws in thefirst place, of forming empirical concepts, of classifying “natural forms” into genera andspecies, or of attributing necessity to empirical laws. Kant makes all of these claims atone point or another, without ever explaining the connection between them.24

This unclarity concerning the function of the principle of logical purposiveness orsystematicity carries over to its transcendental deduction, which is provided in bothversions of the Introduction, though it is only characterized as such in the second.Underlying this deduction are two key premises: (1) that because of their formalcharacter, the transcendental laws underdetermine the particulars falling under them,that is to say, these laws are compatible with any number of possible empirical laws andarrangements; and (2) that it is necessary to distinguish between a system of natureaccording to empirical laws and such a system that is graspable by the human mind.Kant’s position seems to be that, even though the necessary lawfulness of nature at thetranscendental level somehow guarantees that nature will also constitute a system oflaws at the empirical level, it does not guarantee that it will be one that the humanintellect can grasp. As he puts it in the first Introduction:

For the multiplicity and diversity of empirical laws could be so great that it might be possible forus to connect perceptions to some extent in accordance with particular laws discovered onvarious occasions into one experience, but never to bring these empirical laws themselves to theunity of kinship under a common principle, if, namely, as is quite possible in itself (at least as far asthe understanding can make out a priori ), the multiplicity and diversity of these laws, along withthe natural forms corresponding to them, being infinitely great, were to present to us a rawchaotic aggregate and not the least trace of a system, even though we must presuppose such asystem in accordance with transcendental laws. (EE 20: 209)

The deduction turns on the claim that the power of judgment, in its reflective capacity,cannot accept the possibility of such a state of affairs, but must instead presuppose that

24 The point is emphasized by Guyer, who claims that Kant affirms both a “taxonomic” and “explanato-ry” versions of systematicity and that they are unrelated. See Guyer (1979), 44–5.


nature is systematically organized in a way that is accessible to our cognitive require-ments. Since this is a subjective requirement of the power of judgment, the claim is notthat nature must be purposive—as if this power could impose the conditions of itssuccessful operation on nature—it is rather that it must necessarily regard nature aspurposive as a condition of engaging in empirical inquiry.The crucial question, however, remains why it is necessary to presuppose such a

principle at all. Is Kant perhaps here inflating a desideratum into an absolute require-ment? Or, even worse, is he claiming that success must somehow be guaranteed as aprecondition of engaging in inquiry? Presumably, the most that Kant is entitled toclaim is the necessity of presupposing that the possibility is not ruled out in advance; forthat would render the project of inquiry absurd. Accordingly, if the deduction is tosucceed, it must be shown that logical purposiveness or systematicity is a condition ofthe very possibility of acquiring empirical knowledge, and in this sense a condition ofthe possibility of experience. When this issue first arose in connection with theAppendix to the Dialectic I postponed a discussion of it. But if we are to understandKant’s claim in both Critiques regarding extra-categoreal, yet transcendental conditionsof empirical knowledge, the issue must be addressed.Like Hannah Ginsborg, I think it best to begin with a consideration of the question

of the conditions of the formation of empirical concepts.25 Since such concepts areobviously essential for empirical knowledge, if logical purposiveness or systematicitycan be shown to be necessary in this regard, its transcendental status would besufficiently established for the purposes of this deduction. And Kant, indeed, doesmake such a claim when, in the first Introduction, he characterizes the principle asstating “that for all things in nature empirically determinate concepts can found” (EE20: 211). The question therefore becomes why nature’s systematic ordering, theclassifiability of its constituents into something like genera and species correspondingto natural kinds, should be a necessary condition of the possibility of forming empiricalconcepts rather merely a desideratum, which would greatly facilitate the process.Clearly, some similarity is necessary if the understanding is to form general concepts,but it might seem that Kant is requiring too much.The key to the answer lies in Kant’s expression “empirically determinate,” which

I take to mean being determined by underlying, empirically real features things ratherthan merely by superficial features, which reflect the contingencies of the way in whichwe happen to encounter these things in experience. In other words, Kant’s claim is notthat systematicity is a condition of the possibility of forming any empirical concepts,since any common features would provide the reflective power of judgment withsomething to compare; it is rather a condition of forming empirical concepts that are

25 See Ginsborg (1990), 171–91. What follows owes a great deal to Ginsborg’s account, which I regard asby far the most insightful discussion in the recent literature of the issues regarding the epistemic role of theprinciple of purposiveness.

172 PART I I I

cognitively significant, that “carve nature at its joints,” if you will. Consider Kant’s oft-cited remark regarding Linnaeus:

Could Linnaeus have hoped to outline a system of nature if he had had to worry that if he found astone that he called granite, this might differ in its internal constitution from every other stonewhich nevertheless looked just like it, and all he could hope to find were always individualthings, as it were isolated for the understanding, and never a class of them that could be broughtunder concepts of genus and species[?] (EE 20: 215–16n)

Kant is here claiming that a classificatory system such as that of Linnaeus presupposesthat the observable similarities and differences on which conceptualization and classifi-cation are based correspond to real similarities and differences in the inner nature ofthings. Expressed in Lockean terms, the point would be that systematicity is required ifwe are to have knowledge of “real” as opposed to “nominal” essences. Or, adoptingKant’s distinction between judgments of perception and judgments of experience tothe case of concepts, one could say that unless we presuppose a systematic arrangementin nature, we would have to regard all our concepts as mere “concepts of perception”rather than “concepts of experience.”26

Moreover, as Ginsborg has shown, this analysis can readily be extended to empiricallaws and their necessity. For one thing, without assuming something like natural kindswe could not even begin to look for empirical laws or hope to distinguish such lawsfrom contingent regularities. For another, determinate empirical concepts presupposeknown causal laws, since the inner properties on the basis of which we conceptualizeand classify things must include causal properties. Finally, the necessity and nomologi-cal character of relatively specific laws, such as that of the solubility of gold in aqua regia,are a function of their derivability from higher level laws, such as those that hold at themolecular and atomic levels.27 For all these reasons, then, it is necessary to presupposethat “Nature specifies its general laws into empirical ones, in accordance with theform of a logical system, in behalf of the power of judgment” (EE 20: 216).Admittedly, much more would have to be said in order to make this fully convinc-

ing. But even this superficial sketch should suffice to show that there is no abandon-ment of “critical” principles in this account of the purposiveness of nature as atranscendental principle of the reflective power of judgment. In particular, there isno basis for Tuschling’s claim that Kant viewed this principle as in some sense“superordinate to apperception, categories and the principles of pure understanding.”Nor, may I add, should he have done so. To argue to the contrary is to conflate theinsufficiency of the transcendental principles of the understanding to account for the

26 Support for this thesis, at least as a reading of Kant, is provided by Kant’s anthropological-biologicalwritings, such as “On the Different Races of Men,” in which he sides with Buffon, and against Linnaeus,insisting on the necessity of finding and appealing to real rather than merely nominal types or species. Thus,Kant accepts Buffon’s definition of species in terms of real consanguinity (a causal property) ratherthan superficial similarities. For a useful discussion of this issue, see Zammito (1992), 199–213.

27 See Ginsborg (1990), 190.


possibility of empirical cognition with the inadequacy of the grounding of theseprinciples themselves. Nothing in the third Critique indicates that Kant had any secondthoughts about the latter and, as we have already seen, he had readily acknowledgedthe former in 1781. Otherwise expressed, it is not that Kant belatedly came to theconclusion that he could not account for the applicability of the transcendental lawsapart from the presupposition of the purposiveness of nature; it is rather that hecontinued to recognize that these laws, as “formal conditions of empirical truth,” donot of themselves yield a “sufficient mark of empirical truth.”

IIIIn light of these considerations, we are in a position to consider Tuschling’s thesis that inthe third Critique Kant abandoned his essential principle that the discursive humanunderstanding is the ultimate source of the unity of nature and that he granted ultimacyinstead to the intuitive intellect. Central to this reading is the assumption of an essentialconnection between purposiveness in its various forms and the idea of an intuitiveintellect, which is introduced in }77 as a contrast to the discursive variety. Whereas thelatter is defined by its dependence on given sensible data, which it subsequentlysubsumes under “analytic universals,” the former is characterized by its independenceof any such data. Thus, rather than having to subsume sensibly given particulars underabstract, analytic universals, a putative intuitive intellect would generate its own contentthrough its “synthetic universals.”Moreover, as a direct result of this, the contingency offit between universal and particular, which is endemic to a discursive intellect and theground of the dualities characteristic of the Kantian philosophy, would be overcome.As textual support for this thesis, we are given Kant’s remark in the published

Introduction that since the universal laws of nature (the formal, transcendental prin-ciples) have their ground in the human understanding, “the particular empirical laws,in regard to that which is left undetermined in them by the former [the universal lawsof nature], must be considered in terms of the sort of unity they would have if anunderstanding (even if not ours) had likewise given them for the sake of our faculty ofcognition, in order to make possible a system of experience in accordance withparticular laws of nature” (KU 5: 180). Assuming that this other and more thanhuman understanding is to be identified with the intuitive intellect of }77, it mightseem that the latter is being assigned the function of grounding the lawfulness of nature.And from this it is but a short step to the conclusion that this same intellect also groundsthe beautiful in art and organic unity in nature.Even setting aside the question of the identification of the two more than human

intellects, however, such a reading is hardly compatible with the merely regulativefunction Kant assigns to the idea of purposiveness. As a principle of the reflective powerof judgment, this prescribes how we (as finite, discursive cognizers) are rationallyconstrained to regard nature, not how it “really is an sich.” Moreover, although thisidentification seems natural and was certainly made by Hegel, it is actually quite

174 PART I I I

problematic. This has been clearly shown byManfred Baum, who notes that Kant linksthe idea of purposiveness specifically to our discursive form of understanding, while anintuitive intellect is by definition one for which this idea and the associated contingen-cy of fit has no place.28 Accordingly, it is difficult to see how the latter could function asthe ground of purposiveness on the Kantian account. Similarly, we cannot, as Hegelattempted to do, explain the regulative function assigned to the idea of purposivenessin terms of the merely ideal status of the intuitive intellect. On the contrary, if we couldassume the reality of such an intellect, there would be no room for purposiveness in anyform.29

What then is the function of this intuitive intellect and why does Kant appeal to twodistinct, perhaps even incompatible, conceptions of a more than human understand-ing? These are not simple questions and I can here do no more than sketch the outlineof an answer. To begin with, the idea of an intuitive intellect has two distinct tasks inthe resolution of the Antinomy of Teleological Power of Judgment. The first is astraightforward limiting function, namely, to underscore the claim that the need torepresent organic beings in accordance with the idea of purpose is a function of thediscursive nature of our intellect. From this, Kant concludes that it is a merelysubjective necessity, not applicable to every conceivable intellect, for example, anintuitive one, and is therefore without any ontological import.30 The second functionis to account for the possibility that mechanism and teleology are not merely logicallycompatible qua maxims, but that they are unifiable in a single overarching account. Forunless this were the case, “they could not subsist alongside one another in theconsideration of nature” (KU 5: 412). In other words, without such unifiability,the requirement that empirical laws cohere in a system would have to be abandonedin the biological domain, since mechanistic and teleological explanations could not becombined in a single research program. The ground for this unifiability is provided bythe idea of an intuitive intellect precisely because what for our discursive intellectrequires the idea of purpose would be grasped by such an intellect as necessaryindependently of any appeal to purposes.By contrast, the idea of the non-human understanding of the Introduction serves to

regulate the power of judgment in its reflective capacity rather than to limit reason in itstranscendent pretensions. Within the context of the Introduction’s concern withlogical purposiveness, this idea is necessary in order to conceive of the possibility of

28 See Baum (1990), esp. 167–73.29 Also relevant to this point is the similarity between the idea of an intuitive intellect and Spinoza’s God/

substance, which acts by virtue of the necessity of its nature rather than in light of an idea. After criticizingSpinoza on the grounds that one cannot account for the purposiveness in nature by deriving it from a firstcause that acts by virtue of the necessity of its nature (KU 5: 393–4), Kant could hardly endorse the derivationof such purposiveness from an intuitive intellect that proceeds in an analogous fashion. The similarity is notedby Thompson (1995), esp. 448. Tuschling traces the distinction between the discursive, finite intellect andthe non-discursive, divine intellect to Descartes, especially Principles I, 23, but this does not affect the issue.See Tuschling (1990), 175 n. 5.

30 See also my discussion of this issue in Allison (1991a), which is included in this volume.


the multiplicity of empirical laws constituting an order graspable by a human (discur-sive) intellect. Similarly, in the Dialectic of the Teleological Power of Judgment, theidea of a creative intelligence is supposedly needed for the representation by adiscursive intellect of the unique whole–part relationship, which is claimed in theAnalytic to be criterial for organic beings. In both cases, then, this intelligent cause isconceived according to the model of a discursive understanding, since it is viewed asoperating according to the idea of a purpose or design (something which is explicitlydenied of the intuitive intellect).This suggests that the idea of a divine or more than human intellect is characterized

as non-discursive (and therefore as proceeding non-purposively) when it functions as alimiting concept and as discursive and purposive when it functions regulatively. But nomatter how one stands on that question, it should at least be clear that Kant consistentlylinks purposiveness with the discursive rather than the intuitive intellect.Finally, it should also be noted that because of the way in which the intuitive

intellect is characterized, what Tuschling views as a breakthrough to a new speculativestandpoint would have been regarded by Kant as a regression to a pre-critical position.Tuschling is well aware of this problem with respect to the young Hegel’s own appealto the intuitive intellect. Accordingly, he insists that for Hegel this does not mark aregression because (i) it leads to a new logic of absolute thought, which denies theprinciples of non-contradiction and identity, and (ii) that Hegel’s metaphysics is not ananalytic “concept-metaphysic” in the manner of Wolffian rationalism, but ratherdemonstrates the concept as the “immanent structure of empirical existence.”31 Buthowever accurate this may be as a characterization of Hegel’s view, the fact remainsthat such an appeal to the intuitive intellect would have been regarded as a regressionby Kant. Moreover, let us not forget that Tuschling’s basic claim was that even theauthentic Kantian would have to acknowledge these results and it is only this claim thatI have challenged here.

31 Tuschling (1990), 187.

176 PART I I I

Essay Eleven

Reflective Judgment and theApplication of Logic to Nature:Kant’s Deduction of thePrinciple of Purposivenessas an Answer to Hume

Inhis analysis of the power of judgment in the two versions of the Introduction to the thirdCritique Kant introduces the purposiveness of nature as a new transcendental principlepertaining to this power in its reflective capacity and in the published version he provides itwith a transcendental deduction. Although this principle has been the topic of considerablediscussion in recent years, neither its role as a general condition of empirical cognition northe nature and force of its deduction have been adequately explored. Such an exploration isthe concern of this essay. I shall argue that this principle lies at the center rather than theperiphery of Kantian epistemology and that, even though Kant does not present it as such,its deduction may be viewed as his definitive answer to Hume regarding the rationalgrounding of induction broadly construed. The discussion is divided into three parts. Thefirst examines the principle of purposiveness, its source in the reflective power of judgment,and its multifaceted epistemic function. The second analyzes Kant’s deduction of thisprinciple, and the third evaluates its adequacy as an answer to Hume.

IIn both versions of the Introduction Kant describes the power of judgment’s principle asthe purposiveness of nature. In the first version this purposiveness is characterized morespecifically as “logical” (EE 20: 216–17) and in the second as “formal” (KU 5: 180–1); butin both cases it signifies the contingent agreement of the order of nature with ourcognitive needs and capacities. Moreover, in both versions Kant explicitly links thisprinciple with familiar formulas or maxims such as “Nature takes the shortest way”; “it[nature] makes no leaps, either in the sequence of its changes or in the juxtaposition ofspecifically different forms”; and “the great multiplicity of its [nature’s] empirical laws isnevertheless unity under a few principles” (KU 5: 182; see also EE 20: 210). As these

formulas suggest, the basic idea is that we look upon nature as if it had been designedwithour cognitive interests in mind; though we have no basis for supposing that this is in factthe case. In the formulation of the published Introduction Kant describes the principlethusly:

[S]ince universal natural laws have their ground in our understanding . . . the particular empiricallaws, in regard to that which is left undetermined in them by the former, must be considered interms of the sort of unity they would have if an understanding (even if not ours) had likewisegiven them for the sake of our faculty of cognition, in order to make possible a system ofexperience in accordance with particular laws of nature. (KU 5: 180)

Although this formulation in terms of a system of empirical laws (or, as it is often referredto in the literature, of “systematicity”1) is prevalent in both versions of the Introduction,it is not the only way in which this principle and its function are characterized. Forexample, in the first Introduction it is presented as the principle that “for all naturalthings concepts can be found that are determined empirically,”which is then glossed as“we can always presuppose nature’s products to have a form that is possible in terms ofuniversal laws which we can cognize” (EE 20: 211–12). In the published Introduction,however, Kant appears to argue that its main function is not simply to systematizeempirical laws but to ground their very necessity, that is, their claim to nomologicalstatus (KU 5: 183). In fact, in various places in the Introductions, Kant suggests that theprinciple of the purposiveness of nature is necessary for the formation of empiricalconcepts, the classification of “natural forms” into genera and species, the unification ofempirical laws into a system (theory construction), the formulation of empirical laws inthe first place, and the attribution of necessity to such laws.2

Nevertheless, it is possible to find some coherence in this variety of formulations, ifwe keep inmind that the essential function of the power of reflective judgment is to finduniversals for given particulars and that this quest can take the form of seeking eitherempirical concepts under which particulars can be subsumed for the sake of classificationor empirical laws in terms of which their behavior can be explained. Moreover, asHannah Ginsborg has pointed out, these two types of universal are themselves closelyconnected, as are a taxonomic classification of “natural forms” in terms of genera andspecies and a systematic organization of empirical laws. For one thing, without assumingsomething like natural kinds we could not even begin to look for empirical laws or hopeto distinguish such laws from contingent regularities. For another, determinate empiri-cal concepts presuppose known causal laws, since the inner properties in terms of whichwe conceptualize and classify things must include causal properties. Finally, the necessityand therefore the nomological character of relatively specific laws, such as that of the

1 See Guyer (1990a) and (1990b).2 The diversity of formulations and functions is emphasized by Guyer, who claims that Kant affirms both a

“taxanomic” and an “explanatory” version of logical purposiveness (or systematicity) and that they areunrelated. See (1979), 44–5.

178 PART I I I

solubility of gold in aqua regia, are a function of their derivability from higher level laws,such as those that hold at the molecular and atomic levels.3

Perhaps of greater immediate relevance, the same connections can also be spelledout in terms of Kant’s conception of judgment. To begin with, concepts for Kant serveas predicates of possible judgments (KrV A69/B94), which means that the wholepurpose of bringing intuitions under concepts is to make possible determinate judg-ments about their corresponding objects. The judgments in which Kant is interestedare, however, those termed in the Prolegomena “judgments of experience,” that is,objectively valid, grounded claims about objects of possible experience. Moreover, itseems clear that in order to qualify as such a judgment must either be itself a statementof empirical law or derivable from such a law. Accordingly, the search for empiricalconcepts that can serve as predicates in judgments of experience is inseparable from thesearch for empirical laws; and since, as suggested above, the latter is inseparable from ahierarchical organization of such laws, it follows that the quest for the conditions of thepossibility of empirical concepts and for the systematic organization of empirical lawsare best seen as a two poles of a quest for the conditions of the empirical cognition ofnature, qua empirical, or equivalently, for judgments of experience.In themore expansive account in the first Introduction, Kant insists that, even though

the principle of purposiveness is transcendental, it is “merely a principle for the logicaluse of the power of judgment,” which, as such, allows us to regard “nature a priori asqualified for a logical system of its multiplicity under empirical laws” (EE 20: 214). The“logical use of the power of judgment” is to be distinguished from its transcendental use,which, according to the first Critique, is to provide the schemata that are the sensibleconditions for the application of the categories. It consists in the formation of empiricalconcepts and their organization into genera and species, which makes possible thesubordination of these concepts in judgments and the connection of the judgments insyllogisms. Insofar as our concepts are orderable in a single set of genera and species theyhave the form of a logical system, and insofar as this order reflects the actual order ofnature it may be thought of as such a system under empirical laws.Although such a view of nature has merely the status of a regulative idea, Kant points

out that in light of it we can proceed to investigate nature either from the bottom up orfrom the top down. The former procedure begins with the classification of diverseparticulars as members of a single species, then distinct species are unified on the basis ofcommon properties into a genus, and different genera into a higher genera, etc. Ideally,the process culminates in the unification of all higher order genera into a single highestgenus. Conversely, the movement from the top down is one of increasing specifica-tion, wherein differentiations are continually introduced between items that wereinitially taken to be members of a single species.4

3 Ginsborg (1990), 190.4 Although Kant does not make the point here, in his discussion of the same issue in the first Critique he

points out that this process of specification is ideally infinite, since it is always appropriate to search for furthersub species (See A655–7/B683–6).

REFLECTIVE JUDGMENT AND THE APPL ICATION OF LOGIC 179

Kant’s view of the significance of such an ideal scheme for empirical cognition is bestexpressed in two footnotes in the first Introduction. The first of these is attached to thepreviously cited formulation of the principle as “for all natural things concepts can befound that are determined empirically.” In the note Kant argues that, even though thisprinciple may seem to be merely logical and tautologous, it is actually synthetic andtranscendental because it expresses nothing less than the “condition of the possibility of theapplication of logic to nature” (EE 20: 211–12) [my emphasis]. By “logic” Kant does notmean formal logic, but rather our discursive, conceptual abilities. The initial claim isthat some such systematic structure (approaching the ideal of a logical system) isnecessary for the successful exercise of this capacity because this exercise is based oncomparison (the logical act) and comparison requires something to compare. In light ofthis Kant, concludes that, as a condition of the possibility of its own logical activity, thereflective power of judgment,

Must . . . assume . . . that nature in its boundless multiplicity has hit upon a division of itself intogenera and species that makes it possible for our power of judgment to find consensus in thecomparison of natural forms and to arrive at empirical concepts, and their interconnection witheach other, through ascent to more general but still empirical concepts; i.e., the power ofjudgment presupposes a system of nature which is also in accordance with empirical laws anddoes so a priori, consequently by means of a transcendental principle. (EE 20: 212n)

Although at the end of the note Kant repeats the unexplained transition from empiricalconcepts to a system of empirical laws made in the text to which it is attached, the mainfocus is on the conditions of forming a set of empirical concepts that cohere with oneanother. Some degree of coherence is clearly necessary, if the concepts obtainedthrough comparison are to be connectable with one another in judgment, that is, ifthey are to function as concepts at all; and this is what is provided by their systematicordering in terms of the relation of genera and species. It would be a mistake, however,to regard such coherence merely as a supplemental requirement or desideratum ratherthan as a necessary condition of the possibility of the concepts themselves. On thecontrary, the necessity for a hierarchical ordering in terms of genera and species followsfrom the very nature of a concept.Consider, for example, the concept “gold,” understood as a yellow metal, soluble in

aqua regia. It is composed (in part) of these distinct concepts, which constitute itsintension, and it stands to each of them in the relation of species to genus (or logicalform to matter). Thus, “gold” designates a species of yellow objects, of metals, of thingssoluble in aqua regia, etc.; while at the same time it also functions as a genus under whichdifferent types of gold (or of things composed of gold) are to be distinguished as species.Moreover, this is not unique to the concept of gold, but is a feature of every empiricalconcept. In short, every concept (except for that of the highest genus5) is itself both a

5 In his Logic Kant defines the highest genus as that which is not a species and the lowest or infima speciesas that which is not a genus. But whereas it is necessary to assume a highest genus, the possibility of a lowest

180 PART I I I

species of the concepts contained in it and a genus for the concepts falling under it, sothat the very possibility of concepts as general representations presupposes a system ofconcepts subordinate to one another in terms of the relation of genera and species.It follows from this that a system of concepts hierarchically organized in terms of

genera and species is a necessary condition for the application of logic to nature, that is,for empirical judgment. It does not, however, follow that such a system is, as such, alsosufficient to account for the kind of judgment in which Kant was really interested,namely, judgments of experience, which, as I have argued, either themselves state orare derivable from empirical laws. Accordingly, it is worth exploring a second note inthe first Introduction in which Kant seems to go further. In this frequently discussednote he writes:

Could Linnaeus have hoped to outline a system of nature if he had had to worry that if he found astone that he called granite, this might differ in its internal constitution from every other stonewhich nevertheless looked just like it, and all he could hope to find were always individualthings, as it were isolated for the understanding, and never a class of them that could be broughtunder concepts of genus and species[?] (EE 20: 215–16n)

This goes beyond the preceding note bymaking explicit the requirement that a classifica-tory system reflect an underlying order of nature. Thus, whereas any number of suchsystems might be possible, the assumption is that there is one (and only one) that, as itwere, “carves nature at its joints.” Consequently, the goal of a systematizer such asLinnaeus is to provide the system that reflects this order (or at least comes as close aspossible to doing so). Moreover, since the classification of phenomena has to be based onobserved uniformities and differences, the operative assumption must be that outersimilarities and differences correspond to inner or intrinsic ones. To use Kant’s ownexample, objects with the observable features of granite must also be similar in their innercharacter; otherwise there would be no basis for inferring from the fact that an object hasgranite-like features that it will behave similarly to other objects with these features.In light of this, it is instructive to consider Kant’s cryptic account of the “inferences”

of reflective judgment in the Jäsche Logic. According to this account, there are twospecies of such inference, that is, two ways of inferring (empirical) universals fromparticulars, namely, induction and analogy. The former moves from the particular tothe universal according to what Kant terms the “principle of universalization [Principder Allgemeinmachung]: What belongs to many things of a genus belongs to theremaining ones too.” The latter moves from a similarity between two things withrespect to a particular property to a total similarity according to the corresponding“principle of specification: Things of one genus, which we know to agree in much,also agree in what remains, with which we are familiar in some things of this genus but

species is denied on the grounds of the generality of every concept (JL 9: 97; 595). In the Critique of PureReason (A290/B346) Kant identifies the highest genus as the “concept of an object in general,” which is sub-divided into the concepts of something and nothing. There, however, he is concerned with transcendentalrather than empirical concepts.


which we do not perceive in others” (JL 9: 133). Finally, these principles are themselvesspecifications of the higher order principle governing the inferences of the reflectivepower of judgment as a whole: “that the many will not agree in one without acommon ground, but rather that which belongs to the many in this way will benecessary due to a common ground” (JL 9: 132).Although Kant does not make the point, it seems clear that induction and analogy are

the inference-forms through which judgments of experience are grounded.6 In Hu-mean terms, they describe the inferential processes through which we move fromsomething observed to something unobserved. Or, in more contemporary language,they are the vehicles through which predicates are “projected”: either from someinstances of x to all x’s (induction) or of a given x on the basis of other predicates alreadyknown to pertain to that x (analogy). It is also clear that the principle on which they arebased and which licenses such inferences or projections of predicates is itself an applica-tion of principle of purposiveness as described above. For to claim that “the many willnot agree in one without a common ground” is just to claim that observable regularitiesare not merely incidental, but reflect an underlying ground or order of nature.Finally, it follows from this that the stakes involved in a deduction of the principle of

purposiveness are high indeed, amounting to nothing less than what is now termed the“vindication” of induction. More specifically, the issue in its Kantian form is whetherinductive procedures (construed in a broad sense to include reasoning by analogy aswell as induction proper) can be given a rational justification within the framework ofthe reflective power of judgment.7 For such a justification is required in order toprovide an answer to Hume’s skeptical doubts regarding the rational credentials ofinferences from the observed to the unobserved.

IIThe “official” deduction is contained in Section V of the published Introduction,which is given the heading: “The principle of the formal purposiveness of nature is atranscendental principle of the power of judgment” (KU 5: 181). As a prelude to theactual deduction, Kant tries to show that this principle is genuinely transcendental andthat it is required because the task that it performs is not already accomplished by thetranscendental principles of the understanding established in the Transcendental Ana-lytic of the first Critique.In the first Introduction, where Kant likewise insisted on the transcendental nature

of the principle, the alleged problem was that, because of its connection with empirical

6 Kant does, however, refer to the distinction between judgments of perception and judgments ofexperience in his Logic, with the latter characterized as “an empirical judgment through which I get a conceptof an object” (JL 9: 114).

7 It is noteworthy that in his Logic Kant characterizes induction and analogy as merely “logical presump-tions,” since they lack the true necessity possessed by inferences of reason (JL 9: 133).

182 PART I I I

concept formation, it seemed to be merely logical and tautological rather than syntheticand transcendental. As previously noted, Kant there argues for its transcendental statusby trying to show that it is the condition of the possibility of applying logic to nature.By contrast, in the published Introduction this latter claim is dropped, together withany reference to a worry that it might be merely a tautologous, logical principle.Instead, Kant insists that the principle cannot be merely empirical or psychological,since it makes a normative claim about how we ought to judge rather than simplydescribing how we do in fact judge (KU 5: 182). And, in support of this, Kant appealsto the previously mentioned maxims of judgment, which serve as a priori bases of theinvestigation of nature (KU 5: 182).By claiming that the principle is transcendental Kant commits himself to providing it

with some kind of deduction or justification. But one might still think that a separatededuction is redundant because the basic task was already accomplished in the firstCritique. In order to deflect this objection and to underscore the unique nature offormal purposiveness as a principle of the reflective power of judgment, Kant takesgreat pains to argue that the need for such a deduction is not obviated by theTranscendental Deduction of the first Critique.The basic point is that the transcendental laws laid down in the Analytic of the first

Critique do not themselves guarantee the existence of a cognizable order at theempirical level. Since these laws concern merely the “formal” conditions underwhich objects can be cognized together in a single spatiotemporal framework (theunity of experience), they are compatible with any number of different empiricalorderings. Simply put, they underdetermine the particulars falling under them. Thus,even though these laws ensure the existence of some order in nature, it need not be onediscernible in appearances by the human mind.8 For example, the most important ofthese transcendental laws (the principle of causality) states that for any event, b, theremust be some antecedent condition, a, such that given a, b necessarily follows. Thislicenses the search for causes, but it hardly ensures that it will be possible to find them,that is, to distinguish between a merely accidental succession a–b and one that isgenuinely causal.9 For as far as this transcendental law is concerned, experiencemight present few (if any) discernible law-like regularities capable of supportinginduction. And if this were the case, we could neither discover empirical laws interms of which particular phenomena can be predicted or explained nor connect theselaws in over-arching theories. As Kant puts it in a passage that constitutes part of theelaboration of the deduction:

8 Admittedly, there is a tension in Kant’s thought at this point, since, as Guyer notes, he appears to waverbetween the view that there might be no laws at all and that there might be laws that are not discoverable bythe human mind. See Guyer (1990a), 36 and (1990b), 233–4. I am assuming, however, that the latter reflectsKant’s considered opinion (or at least what he ought to have maintained) since the Second Analogy of itselfentails that there must be causal laws of some sort (albeit not necessarily ones that can be recognized as such).

9 I discuss this issue in more detail in connection with what I term the “weak” interpretation of theSecond Analogy in Allison (1996a), 80–91.


For it may certainly be thought that, in spite of all of the uniformity of things in nature inaccordance with the universal laws, without which the form of an experiential cognition ingeneral would not obtain at all, the specific diversity of the empirical laws of nature together withtheir effects could nevertheless be so great that it would be impossible for our understanding todiscover in them an order that we can grasp, to divide its products into genera and species inorder to use the principles for the explanation and the understanding of one for the explanationand comprehension of the other as well, and to make an interconnected experience out ofmaterial that is us for so confused (strictly speaking, only infinitely manifold and not fitted for ourpower of comprehension). (KU 5: 185)10

This passage raises a specter that is both reminiscent of, and significantly different from, themore famous specter that Kant raises in connectionwith the Transcendental Deduction inthe first Critique. In introducing the problematic of that deduction, Kant suggested that(for all that had been shown so far), “Appearances might very well be so constituted thatthe understanding should not find them to be in accordance with the conditions of itsunity” (KrV A90/B123). This may be termed “transcendental chaos” (disorder at thetranscendental level). Clearly, one of the major concerns of that deduction is to exorcizethis specter, which Kant attempts to do by showing that the possibility that appearancesare “so constituted” is ruled out on the grounds of its incompatibility with the conditionsof the unity of apperception and possible experience.By contrast, I call the present specter “empirical chaos” (disorder at the empirical

level), by which I understand a scenario in which something like Hume’s “uniformityprinciple” does not hold.11 Or, more precisely, it is one in which the uniformity thatnature necessarily exhibits in virtue of its conformity to the transcendental lawsimposed by the understanding does not translate into an empirically accessible unifor-mity, understood as one which could support induction and analogy.Since the operative assumption is that this possibility is left open by the Transcen-

dental Deduction of the first Critique, even though the latter succeeded in its appointedtask of establishing the necessary conformity to law of experience at the transcendentallevel, the specter cannot be exorcized by appealing to the unity of apperception and thetranscendental laws derived therefrom. Indeed, the problem arises precisely because thepossibility of empirical chaos or lack of sufficient uniformity is not precluded by theselaws, which ensure, for example, that nothing happens without a cause, but not thatthese causes are discoverable on the basis of empirical regularities. In both IntroductionsKant expresses this point by noting that since an empirically cognizable order of natureis contingent with respect to these transcendental conditions, it cannot be deduced as aconsequence thereof.

10 See also EE 20: 209.11 In the TreatiseHume formulates the principles as holding “that instances, of which we have had no experience,

must resemble those, of which have had experience, and that the course of nature continues always uniformly the same”Hume (2000), 62. I discuss this issue in detail in Allison (2008), 112–60.

184 PART I I I

Moreover, as Kant suggests in the Dialectic of the Teleological Power of Judgment,this contingency may be seen as a consequence of an even more fundamental one thatis endemic to our discursive understanding, namely, the contingency that the “partic-ular as such” has with respect to the universal supplied by the understanding (KU 5:404). For since, as discursive, our understanding proceeds from universals (concepts andlaws) to the particulars that are to be subsumed under them, and since these particulars,as sensibly given, are not themselves products of the act of understanding, it follows thatthere is an unavoidable element of contingency in the fit between universal andparticular. And if this is the case, the same contingency must also apply to the powerof judgment in its reflective activity of seeking universals under which to subsumesensible particulars. Nor, once again, can it be objected that the assertion of anineliminable contingency of fit between universal and particular conflicts with theresults of the Transcendental Deduction of the firstCritique. For that deduction was notconcerned with the “particular as such,” but merely with it qua spatiotemporal entityor event; and taken under that description it remains fully subject to the categories. Butinasmuch as the question of empirical lawfulness concerns precisely the “particular assuch,” such lawfulness is contingent with respect to the universal and the specter ofempirical chaos remains in place.It follows from this that, if this latter specter is to be dealt with, a distinct transcen-

dental principle is required. And it likewise follows, given Kant’s definition of purpo-siveness as the “lawfulness of the contingent as such” (EE 20: 217), that it must take theform of a principle of purposiveness.12 This principle cannot be used, however, to denythe very possibility of empirical chaos in anything like the manner in which thetranscendental unity of apperception denies the possibility of the transcendental varie-ty, that is, by somehow proving that nature, in its empirical diversity, necessarilyconforms to our cognitive needs. For any such objective deduction, even one subjectto the standard “critical” limitation to objects of possible experience or phenomena, isprecluded by virtue of the above-mentioned ineliminable contingency of fit betweenthe (empirically) universal and the particular.This does not, however, rule out the possibility of a subjective deduction, which would

leave the results of the Transcendental Analytic of the firstCritique in place but go beyondthem. The goal of such a deduction would not be to remove the specter by showing it tobe incompatible with the transcendental conditions of experience (since that is impossi-ble), but merely to render it idle. Moreover, this is precisely what Kant’s actual deductionattempts to accomplish by establishing the subjective necessity of presupposing thepurposiveness of nature in the process of empirical inquiry. In other words, the claim isnot that nature is purposive, i.e., that we have some sort of a priori guarantee that it isordered in a manner commensurate with our cognitive capacities and needs; nor is it eventhat we must believe it to be purposive in this sense (which is basically Hume’s position).

12 See also KU 5: 404.


The claim is rather that we are rationally constrained to approach nature as if it were soordered. Or, in Kant’s own terms, at the basis of all reflection on nature (the search forempirical laws) lies the a priori principle that “in accordance with these [empirical] laws acognizable order of nature is possible” (KU 5: 185).Since Kant’s reasoning here is akin to that underlying the well-known claim in the

Groundwork that we can act only under the idea of freedom, it may prove useful toexamine the latter briefly.13 There the “specter” is that despite the resolution of theThird Antinomy, which showed only that transcendental freedom is compatible withcausality according to laws of nature, our apparent practical rationality and agencymight be ultimately tropistic; that even though we take ourselves to be rational self-determiners, we are really moved by underlying causes, e.g. instinct.Kant’s ultimate response to this problem in the Critique of Practical Reason is that our

consciousness of standing under the moral law (the “fact of reason”) assures us of ourfreedom from the practical point of view.14 Kant does not take this route in theGroundwork, however, arguing instead that freedom is a necessary presupposition ofreason insofar as it regards itself as practical. And here the point is not that we mustbelieve ourselves to be free in order to believe that we are agents rather than automata;it is rather that we must act as if we were free, which is just what it means to act underthe idea of freedom. In other words, the idea of freedom has an essentially normativeforce. To act under this idea is to place oneself in the “space of [practical] reasons,” andtherefore to take oneself as subject to rational norms (of both a moral and prudentialsort) rather than merely to causal conditions. Thus, even though it remains alive as ametaphysical possibility, from the practical point of view (that of agency) the specterthat we might be merely automata is completely idle.The suggestion, then, is that the presupposition of the logical or formal purposive-

ness of nature be understood in essentially the same way, that is, as having normative orprescriptive force. In investigating nature we ought to treat it as if it were purposivebecause this is just what is involved in “applying logic” to it. For reasons that shouldnow be clear, there is simply no other procedure possible for the power of judgment inits reflection on nature, at least not if the goal is to form empirical concepts that can becombined in something like judgments of experience. Accordingly, one might say thatthe principle of purposiveness characterizes the “space of judgment,” since it definesthe framework in which alone rational reflection on nature is possible. Moreover, thisserves to explain Kant’s emphasis on the a priori nature of the principle. There isnothing optional about approaching nature in this way, just as there is nothing optionalabout presupposing freedom insofar as we take ourselves as rational agents. But this nomore proves that nature really is purposive than the latter proves that we really are freein some metaphysical sense.

13 I discuss this issue in Allison (1997a) and (forthcoming), both which are included in this volume.14 I analyze Kant’s deduction of freedom in the second Critique in Allison (1990), 243–9 and (forthcom-

ing), which is included in this volume.

186 PART I I I

IIIDoes Kant’s deduction, so construed, constitute a viable response to Humean skepti-cism regarding the rational basis of induction broadly conceived? Indeed, can we evenregard it as a deduction? Setting aside the historical issue of whether Kant actuallyintended his account as an answer to Hume, the answer to both questions appears atfirst glance to be negative. In fact, one might think that by appealing to a merelysubjective necessity to presuppose logical or formal purposiveness, Kant is reallyconceding Hume’s point regarding the uniformity principle rather than answeringhim. Moreover, if it is taken as an attempted answer to Hume, Kant would seem to beguilty of the very error of which he accused Hume’s Scottish common sense critics,namely, of taking for granted what Hume doubted and of demonstrating what henever thought of doubting (Pro 4: 258).Nevertheless, I think that such a dismissal of Kant’s argument and its anti-Humean

thrust would be mistaken. For as in the case of causality and the other a priori conceptsand principles that are at stake in the first Critique, Kant saw clearly that the basicquestion between himself and Hume was one of normativity or right, not simplyindispensability or pragmatic necessity, which Hume certainly did not deny in the caseof the uniformity principle. Hume’s question concerns the grounds for our belief in thisprinciple; the belief itself and its indispensability are never called into question. Andafter arguing that it is neither a demonstrable truth (since its negation is conceivable)nor a generalization from experience (since it is a condition of the possibility of anysuch generalization), Hume concludes that this belief can be understood only as theproduct of custom or habit.15

For Kant, by contrast, the necessity, though merely subjective, is nonethelessrational rather than causal or psychological. Moreover, it is not a matter of belief. Aswe have already seen, what is required is not the belief that nature is purposive, whichwould be inseparable from the belief that it has an intelligent author, but rather that weapproach it as if it were. The principle thus has prescriptive force; it dictates how we arerationally constrained to approach nature, if our concern is to “apply logic” to it, that is,to bring the data of experience under empirical concepts and laws. And if this is correct,Kant can claim to have grounded the right to make such a presupposition. But to dothis is to have resolved the quid juris, which is the proper concern of any Kantiantranscendental deduction.In insisting on the genuineness of this deduction, however, I do not wish to

minimize its differences from that of the first Critique. There it was a matter ofunderstanding legislating to nature, of laying out the formal constraints on any thingthat could become an object of empirical cognition (what Kant termed the “formalconditions of empirical truth”) (KrV A191/B236). Accordingly, the principles thatspecify these constraints or conditions are in this sense constitutive of possible experi-

15 See Hume (1999), 119–24 and (2000), 67–73 and passim.


ence. Here, by contrast, the claim is not that nature or objects of possible experiencemust conform to certain conditions stemming from the nature of our cognitivefaculties, but merely that we must investigate nature on the basis of the assumptionthat it accords with the conditions of our cognition of it; viz., that it constitutes a“logical system.” In this respect, the principle of purposiveness functions as an impera-tive that judgment imposes on itself, since it expresses the condition of the possibility ofthe successful application of this faculty.Kant indicates the true nature and function of this principle when he claims that

through it, “the power of judgment prescribes a law, not to nature (as autonomy) butto itself [my emphasis] (as heautonomy), a law for the specification of nature” (KU 5:185–6). Thus, even though the principle concerns nature as the object of investigation,its prescriptive force is directed back to the power of judgment itself.16 The term“heautonomy” is introduced by Kant in order to emphasize the purely reflexive, self-referential nature of this principle. To claim that the power of judgment is “heauto-nomous” in its reflection is just to say that it is both source and referent of its ownnormativity.17 And this is what distinguishes judgment’s a priori principle from those ofthe understanding, which legislates transcendental laws to nature, and of (practical)reason, which prescribes the objectively necessary laws of a free will.

16 At this point my analysis is very close to that of Juliet Floyd, who likewise sees Kant’s appeal toheautonomy as at the heart of his answer to Hume. See Floyd (1998), 206–14.

17 The term “heautonomy” derives from attaching the Greek definite article “he” to the pronoun “auto,”which stands for either “self” or “itself.” For a useful discussion of this issue see Floyd (1998), 205.

188 PART I I I

Essay Twelve

The Critique of Judgment as a“True Apology” for Leibniz

At the end of his response to Eberhard, who had notoriously proclaimed that there isnothing true in the [first] Critique that is not already to be found in Leibniz,1 Kantremarks in a palpably ironic fashion that “the Critique of Pure Reason might well be thetrue apology for Leibniz” (UE 8: 250). But despite its ironic intent, Kant’s “apology” isneither without serious philosophical import nor limited to the first Critique. Its focus ison what Kant takes to be the three “peculiarities” of Leibnizian metaphysics: theprinciple of sufficient reason; the doctrine of monads; and the conception of a pre-established harmony (UE 8: 247). In sharp contrast to his usual polemical procedure,however, Kant suggests that Leibniz actually understood these principles in an essen-tially critical manner; though his failure to express this clearly gave rise to the mis-understandings of epigones such as Eberhard.Although each of these principles involves topics taken up in the first Critique, one

aspect of the last relates directly to the problematic of the third. This is the so-called“harmony” between sensibility and understanding, which Kant insists is not onlynecessary for the very possibility of experience and its conformity to a priori laws,but also, indeed, “especially (as theCritique of the Power of Judgmentwill intimate) for thepossibility of an experience of nature under its manifold particular and merely empiricallaws, of which the understanding teaches us nothing a priori, as if nature weredeliberately ordered for our comprehension” (UE 8: 250).Since harmony, so construed, is equivalent to the principle of the logical or formal

purposiveness of nature for which Kant argued in both Introductions to the thirdCritique, he is in effect crediting Leibniz with anticipating one of the central ideas ofthat work. Accordingly, my goal in this essay is to show that this suggestion is to betaken seriously. More specifically, I shall argue that Kant’s exposition and deduction ofthis principle may be read as the culmination of his attempt to reconcile the Leibnizianrealism of universals with Locke’s view that all concepts are products of the “Work-manship of the Understanding.”2

1 Eberhard (1968), 289. 2 Locke (1975), 415.

IThe portion of the imaginary debate with Locke, which Leibniz constructs in his NewEssays, that concerns us here centers on Locke’s claim that “general and universal belongnot to the real existence of things; but are the inventions and creatures of theunderstanding, made for its own use, and concern only signs, whether words orideas.”3 That this claim is equally applicable to words and ideas is an immediateconsequence of Locke’s underlying semantic theory, according to which “Words intheir primary or immediate Signification, stand for nothing, but the Ideas in the mind of him thatuses them.”4 Both our interest and for the most part Leibniz’s, however, is limited to itsapplication to ideas, particularly complex ideas of substances.Like all complex ideas for Locke, those of substances are due to the “workmanship

of the understanding.” Moreover, far from denigrating such workmanship, Lockeemphasizes its absolute indispensability for both language and cognition. As he putsit in a manner that reflects his semantic theory, it holds of both words and ideas thatparticulars by themselves would be useless, since knowledge, “though founded inparticular Things, enlarges itself by general Views; to which, Things reduced intosorts, under general Names, are properly subservient.”5 In other words, without“general views,” the understanding would have nothing to work with except a setof ideas of distinct, isolated particulars, which it could neither connect into a cognitivewhole nor communicate linguistically. But general views require general terms, andthese, Locke insists, “become general by being made the signs of general Ideas.”6

Correlatively, the latter become themselves general “by separating from them thecircumstances of Time and Place, and any other Ideas that may determine them to thisor that particular Existence.”7

Among the products of this abstractive procedure are the ideas of substance-types orsorts, which, using the traditional terminology that he usually disdains, Locke terms“essences.” The distinctive feature of Locke’s position, however, is the sharp contrasthe draws between the “real essences,” which are the internal structure of things, andthe “nominal essences,” which are formed by the understanding.8 Although heacknowledges that the latter are based on observed resemblances, Locke denies thatthey correspond (or at least can be taken to correspond) to the real essences of theparticulars constituting a sort. In short, as a “modern,” Locke rejects the notion ofnatural kinds and the whole allegedly disreputable Aristotelian-Scholastic scheme ofexplanation, classification, and definition by genus and species that goes with it.Instead, he appeals to Boyle’s corpuscularian theory of matter as expressing the truthabout the (to us unknowable) inner nature or micro-structure of bodies. Since thismicro-structure is inaccessible to us, our classificatory and explanatory schemes must bebased entirely on constructed nominal essences. But rather than drawing any skeptical

3 Locke (1975), 414. 4 Ibid., 405. 5 Ibid., 410.6 Ibid., 410–11. 7 Ibid., 411. 8 Ibid., 417–18.

190 PART I I I

conclusions from this, Locke maintains that these merely nominal essences suffice forthe largely pragmatic ends of science.9

Although Locke’s rejection of natural kinds was clearly inspired by the corpuscular-ian theory of matter, according to which all bodies, whatever their qualitative differ-ences, are composed of particles of the same fundamental type, it is not entailed by thattheory.10 For one might conceivably hold that the kinds or species formed by theunderstanding on the basis of observed qualitative similarities, correspond to realdistinctions between the number, size, shape, and arrangement of the particles at themicro-level. Such a move was ruled out for Locke, however, by his denial that thehuman understanding is capable of attaining the requisite knowledge of the micro-structure of things. Moreover, Locke advances at least three arguments against theexistence of natural kinds that are largely independent of this theory. These are thearguments from the existence of monsters or borderline cases; from the man-madenature of the abstract ideas identified with the essences of things; and from the sheerabundance of observable similarities with which nature provides the understanding.11

Of these, the last is by far the most important, since it not only acknowledges, buttrades on, the very point on which defenders of natural kinds insist, namely, theexistence of significant observable similarities. Against this, Locke maintains that thesheer abundance of these suggests that we construct our ideas of sorts on the basis ofobservable features that we take to be significant rather than according to any plan laiddown by nature. Thus, the issue for Leibniz and later for Kant will concern largely theconnection between these observable features and the inner nature of things, whichcan also be understood as the question of the grounds for distinguishing betweenmerely accidental similarities or differences and those that are explanatorily or classifi-catorily significant.

IIEarly on in his discussion of the topic Leibniz makes two noteworthy points. First, hesupplements Locke’s account of the indispensability of general terms by putting in themouth of Philalethes (Locke’s spokesman) the comment that particular words by them-selves would be of no use “for judging the future by the past or one individual byanother.”12 In introducing these functions for general terms (and their correspondingideas), Leibniz anticipates both Hume’s fundamental question concerning the possibilityof empirical knowledge (the grounds of the inference from the observed to the unob-served) and Kant’s own treatment of induction and analogy as two species of empiricallybased inference (JL 9: 132–3). Second, Leibniz defends the notions of genera and speciesagainst Locke’s dismissive treatment of them. “The art of ranking things in genera andspecies,” he remarks, “is of no small importance and very much assists our judgment as

9 For an excellent account of Locke’s views on these matters, see Jolley (1999).10 Ibid., 144–6. 11 Ibid., 144–50. 12 Leibniz (1981), 288.

THE CRIT IQUE OF JUDGMENT AS A “TRUE APOLOGY” FOR LEIBNIZ 191

well as our memory.”And he adds that they help not merely in the retention of things (inthememory), “but also to find them.”13 In short, in addition to serving asmnemonic aids,the supposedly disreputable ideas of genera and species have essential roles to play in bothjudgment and the logic of discovery.Leibniz’s central focus, however, is on the implications of Locke’s distinction be-

tween real and nominal essences. To begin with, he rejects Locke’s contention that theessence of each species is nothing but an abstract idea and, therefore, that “‘general anduniversal, belong not to the . . . existence of things; but are the [workmanship] of theunderstanding.’”14 Against this, Leibniz counters that “generality consists in the resem-blance of singular things to one another, and this resemblance is a reality.”15

Insofar as these resemblances are part of the surface rather than the deep structure ofthings, Locke could easily acknowledge this without endangering his main pointconcerning the inaccessibility and therefore irrelevance of the latter for taxonomicand explanatory purposes. Moreover, since Locke himself insists on the great variety ofresemblances from which one must choose in order to frame ideas of sorts, it does notthreaten his thesis about the conventional nature of these ideas and the divisions basedupon them either. Thus, Leibniz’s opening salvo leaves the main lines of Locke’sposition, including his rejection of natural kinds, intact.Leibniz’s more telling arguments turn on the distinction between essential and acci-

dental properties. In response to Locke’s claim that essence pertains only to sorts, Leibnizretorts that there is also something essential to individuals, apart from which they wouldnot be of a given nature. For example, it is essential for substances to act, minds to think,and bodies to have extension andmotion.Moreover, in light of this, Leibniz distinguishesbetween two types of sort or species, one of which is natural in the sense that a member ofit cannot cease (at least according to laws of nature) being a member, and the other isaccidental to the individuals who constitute its membership. Human beings constitute aspecies of the first type and the healthy and handsome of the second.16

Leibniz further suggests that the whole dispute concerning natural kinds stems froman ambiguity in the term “species,” which he notes can be understood either mathe-matically or physically.17 Construed in the former way, a thing’s species encompasseseverything that pertains to it. Given this conception of species, it follows that theslightest difference suffices to assign things to different species and that any change thata thing undergoes entails a change of species. By contrast, physical (or real) species donot require such a rigid criterion of species identification and thus allow for thepreservation of specific identity throughout change.With this the issue shifts to the question of the outer criteria for the determination of

membership in a physical species; and it is here that the battle with Locke is fully joined.Leibniz’s basic principle is that “every outer appearance is grounded in the innerconstitution,” or, alternatively, “whatever we truthfully distinguish and compare is

13 Leibniz (1981), 291–2. 14 Ibid., 292.15 Ibid. 16 Ibid., 305. 17 Ibid., 308.

192 PART I I I

also distinguished or made alike by nature.”18 Nevertheless, agreeing with Lockeconcerning the huge variety of outer resemblances to choose from, Leibniz acknowl-edges the difficulty of determining in a given case whether one is in fact comparing anddistinguishing “truthfully” or merely on the basis of superficial and accidental features.Although he does not express himself in these terms, Leibniz’s treatment of this

problem amounts to construing the principle that outer resemblances and differencesare grounded in the inner constitution of things as regulative in the Kantian sense. It isin this spirit that he refers to the “care and experience” required “to mark out generaand species in a manner which comes fairly close to nature.”19 Similarly, he refers to thenatural order as something to which we can progressively come nearer through (in thecase of organic beings) a deeper study of the reproductive processes of species.20 And,appealing to the example of gold, he admits that definitions based on observablefeatures are purely nominal, indeed, “only conjectural” and at times even “merelyprovisional,” which entails both that they are subject to correction and that there is anorm with respect to which such corrections are made.21 Finally, near the end of hisdiscussion of the topic, Leibniz speaks of the necessity of a “presumption” of an“essential and unchangeable nature” and asserts that “things that differ only throughaccidental changes, such as water and ice . . . are of a single species.”22

In all of these passages Leibniz effectively treats the idea of an inner division of natureinto natural kinds reflected in observable features both as a norm guiding the search forexplanation and as a goal of classification that is to be approached asymptotically, butnever fully attained. And while the epistemological orientation of the New Essays mayhave prevented him from dwelling on the point, it seems clear that Leibniz derived thewarrant for such a “presumption” from the basic principles of his monadology. In otherwords, even though the principle functions merely regulatively with respect to theempirical investigation of nature, its justification remains metaphysical.

IIIAs Beatrice Longuenesse has reminded us, Kant agrees fully with Locke’s thesis that ourideas of sorts are due to the “workmanship of the understanding,” if not with theprecise manner of this workmanship.23 In fact, his anti-innatism, though quite differentin spirit than Locke’s, applies to all concepts, including the categories. For Kant’sconsidered view is that all concepts with regard to their form, i.e., universality, areproducts of the “logical acts” of comparison, reflection, and abstraction (JL 9: 94–5). Atthe same time, however, Kant rejects Locke’s conventionalism and within the frame-work of his conception of the reflective power of judgment argues for a position that isclose to Leibniz’s, albeit without the latter’s metaphysical commitments.

18 Ibid., 309. 19 Ibid. 20 Ibid., 310. 21 Ibid.22 Ibid., 325. 23 Longuenesse (1998), 119.


Kant initially characterizes the basic problem as that of the “application of logic tonature” (EE 20: 212n). By “logic” Kant means our discursive, conceptual abilitiesrather than formal logic; and his claim is that such application requires the conformityof nature in its empirical manifoldness to the conditions of the successful exercise ofthese capacities. Kant terms such conformity nature’s “logical” or “formal purposive-ness,” which may be expressed in Leibnizian language as the idea of a pre-establishedharmony between the order of nature and our cognitive capacities.The situation is complicated, however, by the fact that Kant characterizes this

principle of purposiveness in different ways in different portions of the Introductionsand appears to assign quite different functions to it, all of which are somehow related tothe project of “applying logic to nature.” Thus, he suggests that it is necessary for theformation of empirical concepts, the classification of “natural forms” into genera andspecies, the unification of empirical laws into a system, the formulation of empiricallaws in the first place, and the attribution of necessity to such laws.24

Nevertheless, it is possible to find some coherence in this variety of formulations, ifwe keep in mind that the essential function of the reflective power of judgment is tofind universals for given particulars and that this quest can take the form of seekingeither empirical concepts under which particulars can be subsumed for the sake ofclassification or empirical laws in terms of which their behavior can be explained.Moreover, as Hannah Ginsborg has pointed out, these two types of universal arethemselves closely connected, as are a taxonomic classification of “natural forms” interms of genera and species and a systematic organization of empirical laws. For onething, without assuming something like natural kinds we could not even begin to lookfor empirical laws or hope to distinguish such laws from contingent regularities, whichis basically Leibniz’s point against Locke. For another, determinate empirical conceptspresuppose known causal laws, since the inner properties in terms of which weconceptualize and classify things must include causal properties. Finally, the necessityand therefore the nomological character of relatively specific laws, such as that of thesolubility of gold in aqua regia, are a function of their derivability from higher level laws,such as those that hold at the molecular and atomic levels.25

Kant’s most fundamental concern, however, is with empirical concepts, since theyprovide the foundations of empirical knowledge and even the application of thecategories in determinative judgments. Thus, in what I take to be the most helpfulof Kant’s many formulations of the principle of purposiveness, it states that “for allthings in nature empirically determinate concepts can be found” (EE 20: 211). Thebasic idea is that a systematic organization of nature (approaching the ideal of a “logicalsystem”) is necessary for the formation of empirical concepts because these concepts areformed by the “logical acts” of comparison and reflection, which, in turn, require that

24 See essay 11 note 2. 25 Ginsborg (1990), 190.

194 PART I I I

the data be comparable. And in light of this Kant concludes in a note that, as acondition of the possibility of its own logical activity, the reflective power of judgment,

Must . . . assume . . . that nature in its boundless multiplicity has hit upon a division of itself intogenera and species that makes it possible for our power of judgment to find consensus in thecomparison of natural forms and to arrive at empirical concepts, and their interconnection witheach other, through ascent to more general but still empirical concepts; i.e., the power ofjudgment presupposes a system of nature which is also in accordance with empirical laws anddoes so a priori, consequently by means of a transcendental principle. (EE 20: 212n)

Apart from the attribution of transcendental status to the principle, the most notewor-thy feature of this passage is the suggestion of the inseparability of the conditions of theformation of empirical concepts and of the coherence or systematic unification of suchconcepts into genera and species. Clearly, some degree of coherence is necessary if theconcepts obtained through comparison and reflection are to be connectable with oneanother in judgment, that is, if they are to function as concepts at all; but it might verywell be doubted that their unification into genera and species is likewise required.Nevertheless, I take Kant’s point to be that a hierarchical ordering in terms of genera

and species is required by the very nature of a concept and is therefore a necessarycondition of the concepts themselves. Consider, for example, the concept “gold”understood as a yellow metal, soluble in aqua regia. It is composed (in part) of thesedistinct concepts, which constitute its intension, and it stands to each of them in therelation of species to genus (or logical form to matter). Thus, “gold” designates aspecies of yellow objects, of metal, of things soluble in aqua regia, etc.; while at the sametime the concept also functions as a genus under which different types of gold (or ofthings composed of gold) are to be distinguished as species. Moreover, this is notsomething unique to the concept of gold, but is a feature of every empirical concept. Inshort, every such concept (except for that of the highest genus26) is itself both a speciesof the concepts contained in it and a genus for the concepts falling under it, so that thevery possibility of concepts as general representations presupposes a system of conceptssubordinate to one another in terms of the relation of genera and species.If this is correct, it provides a Kantian justification for Leibniz’s defense of the

usefulness of genera and species in the face of Locke’s dismissive treatment of themas remnants of a discredited scholasticism. But since it appears to leave open thepossibility that any number of such classificatory systems might be produced by the“workmanship of the understanding,” it does not suffice to refute Locke’s conven-tionalism and to establish the necessity for natural kinds affirmed by Leibniz. Accord-ingly, it is worth exploring a second note in the first Introduction in which Kant seemsto affirm precisely this. In this frequently discussed passage he writes:

26 See essay 11 note 5.


Could Linnaeus have hoped to outline a system of nature if he had had to worry that if he found astone that he called granite, this might differ in its internal constitution from every other stonewhich nevertheless looked just like it, and all he could hope to find were always individualthings, as it were isolated for the understanding, and never a class of them that could be broughtunder concepts of genus and species[?] (EE 20: 216n)

This goes beyond the preceding note by making explicit the requirement that aclassificatory system reflect an underlying order of nature. Thus, whereas any numberof such systems might be possible, the assumption is that there is one (and only one)that, as it were, “carves nature at its joints.” Consequently, the goal of a systematizersuch as Linnaeus is to provide the system that reflects this order (or at least comes asclose as possible to doing so). Moreover, since the classification of phenomena has to bebased on observed uniformities and differences, the operative assumption must be thatouter similarities and differences correspond to inner or intrinsic ones. But this isequivalent to Leibniz’s principle that “every outer appearance is grounded in theinner constitution,” or, “that whatever we truthfully distinguish and compare [onthe basis of outer appearances] is also distinguished or made alike by nature.”Thus, within the confines of the first Introduction, Kant may be said to have

provided a defense of the chief elements of Leibniz’s critique of Locke on the basisof the principle of the logical purposiveness of nature. And the latter may itself be seenas the critical replacement for the Leibnizian doctrine of the pre-established harmonyin one of its guises. But in order for this defense to amount to a “true apology,” Kantmust provide a transcendental grounding for this principle. This is the task of itstranscendental deduction to which I now turn.

IVThe “official” deduction of the principle of purposiveness is contained in Section V ofthe published Introduction. As a prelude to this deduction, Kant tries to show that theprinciple is genuinely transcendental and that it is required because the task that itperforms is not already accomplished by the transcendental principles of the under-standing established in the first Critique.The reason why Kant insists that the principle is transcendental is that it makes a

normative claim about how we ought to judge rather than simply describing how we do,in fact, judge. And by insisting on this Kant commits himself to providing it with adeduction. Even granting this, however, one might still argue that a separate deductionwould be redundant, since the basic task was already accomplished in the first Critique.In order to deflect any such objection and to underscore the unique nature of formalpurposiveness as a principle of the power of judgment, Kant takes great pains to arguethat the need for such a deduction is not obviated by the results of the first Critique.The basic point is that the transcendental laws laid down in the Analytic of the first

Critique do not themselves guarantee the existence of a cognizable order at the

196 PART I I I

empirical level. Since these laws concern merely the “formal” conditions under whichobjects can be cognized together in a single spatiotemporal framework (the unity ofexperience), they underdetermine the particulars falling under them and are thereforecompatible with any number of different empirical orderings. Thus, even though theselaws ensure the existence of some order in nature, it need not be one discernible inappearances by the human mind.27 As Kant puts it in a passage that constitutes part ofthe elaboration of the deduction:

For it may certainly be thought that, in spite of all of the uniformity of things in nature inaccordance with the universal laws, without which the form of an experiential cognition ingeneral would not obtain at all, the specific diversity of the empirical laws of nature together withtheir effects could nevertheless be so great that it would be impossible for our understanding todiscover in them an order that we can grasp, to divide its products into genera and species inorder to use the principles for the explanation and the understanding of one for the explanationand comprehension of the other as well, and to make an interconnected experience out ofmaterial that is us for so confused (strictly speaking, only infinitely manifold and not fitted for ourpower of comprehension). (KU 5: 185)28

This passage raises a specter that is both reminiscent of, and significantly different from,the more famous specter raised in connection with the Transcendental Deduction inthe first Critique. In introducing the problematic of that deduction, Kant suggested that(for all that had been shown so far), “Appearances might very well be so constitutedthat the understanding should not find them to be in accordance with the conditions ofits unity” (KrV A90/B123). This may be termed “transcendental chaos” (disorder atthe transcendental level). Clearly, one of the major concerns of that deduction is toexorcize this specter, which Kant attempts to do by showing that the possibility thatappearances are “so constituted” is ruled out on the grounds of its incompatibility withthe conditions of the unity of apperception or possible experience.By contrast, I call the present specter “empirical chaos” (disorder at the empirical

level), by which I understand a state of affairs in which there are either few (if any)observable similarities between things and events, or in which those that are found donot correspond to similarities in the inner nature of things. In such a scenario, as Leibnizclearly recognized, there would be no basis for “for judging the future by the past orone individual by another.”Since the operative assumption is that this possibility is left open by the Transcen-

dental Deduction of the first Critique, even though the latter succeeded in its appointedtask of establishing the necessary conformity to law of appearances at the transcendentallevel, the specter cannot be exorcized by appealing to the unity of apperception and thetranscendental laws derived therefrom. Indeed, the problem arises precisely because thepossibility of empirical chaos is not precluded by these laws, which ensure, for example,that nothing happens without a cause, but not that these causes are discoverable on the

27 See essay 11 note 8. 28 See also EE 20: 209.


basis of empirical regularities. In both Introductions Kant expresses this point by notingthat, since an empirically cognizable order of nature is contingent with respect to thesetranscendental conditions it cannot be deduced as a consequence thereof.It follows from this both that a distinct transcendental principle is required to deal

with the latter specter and that it must take the form of a principle of purposiveness.29

This principle cannot be used, however, to deny the very possibility of empirical chaosin anything like the manner in which the transcendental unity of apperception entailsthe denial of the possibility of the transcendental variety, that is, by somehow provingthat nature, in its empirical diversity, necessarily conforms to our cognitive needs. Forany such objective deduction, even one subject to the standard “critical” limitation toobjects of possible experience or phenomena, is precluded by virtue of the ineliminablecontingency of fit between the (empirically) universal and the particular.30

Nevertheless, this does not rule out the possibility of a subjective deduction, whichwould leave the results of the Analytic of the first Critique in place but go beyond them.31

The goal of such a deduction would not be to remove the specter by showing it to beincompatible with the transcendental conditions of experience (since that is impossible),but merely to render it idle. Moreover, this is precisely what Kant’s actual deductionattempts to accomplish by establishing the subjective necessity of presupposing thepurposiveness of nature in the process of empirical inquiry. In other words, the claim isnot that nature is purposive, i.e., that we have some sort of a priori guarantee that it isordered in a manner commensurate with our cognitive capacities and needs; nor is it eventhat we must believe it to be purposive in this sense. It is rather that we are rationallyconstrained to approach nature as if it were so ordered. For only by doing so is it possible tosubsume the data of experience under empirical concepts that are serviceable for theexplanation and prediction of phenomena. Again in Leibnizian terms, what is necessary isthe “presumption” of an “essential and unchangeable nature” and that “things that differonly through accidental changes, such as water and ice . . . are of a single species.”A model for such a subjective deduction, which takes the form of licensing a

“presumption” by showing its rational necessity is provided by Kant’s well-knowndiscussion of freedom in the Groundwork, where he argues that freedom is a necessarypresupposition of reason insofar as it takes itself to be practical. Here the point is notthat we must believe ourselves to be free in order to believe that we are agents ratherthan automata; it is rather that we must act as if we were free, which is just what itmeans to act under the idea of freedom.32 In other words, the idea of freedom has an

29 The latter follows given Kant’s definition of purposiveness as the “lawfulness of the contingent as such”(EE 20: 217). See also KU 5: 404.

30 For Kant’s discussion of such contingency, see also KU 5: 404–8.31 The thesis that Kant’s new transcendental principle involves an abandonment of original “critical”

principles has been affirmed by Tuschling (1990) and (1992). I criticize Tuschling’s analysis in Allison (2000c),which is included in this volume.

32 For my discussion of this issue, see Allison (1997a) and (forthcoming), which are included in thisvolume.

198 PART I I I

essentially normative force. To act under this idea is to place oneself in the “space of[practical] reasons,” and therefore to take oneself as subject to rational norms (of both amoral and prudential sort) rather than merely to causal conditions. Thus, even thoughit remains alive as a metaphysical possibility, from the practical point of view (that ofagency) the specter that we might be merely automata is completely idle.The suggestion, then, is that the “presumption” of the logical or formal purposive-

ness of nature be understood in essentially the same way, that is, as having normative orprescriptive force. In investigating nature we ought to treat it as if it were purposivebecause this is just what is involved in “applying logic” to it. For reasons that shouldnow be clear, there is simply no other procedure possible for judgment in its reflectionon nature, at least not if the goal is to attain empirical knowledge. Accordingly, onemight say that the principle of purposiveness characterizes the “space of judgment,”since it defines the framework in which alone rational reflection on nature is possible.Moreover, this serves to explain Kant’s emphasis on the a priori nature of the principle.Pace Locke, and in essential agreement with Leibniz, there is nothing optional aboutapproaching nature in this way, just as there is nothing optional about presupposingfreedom insofar as we take ourselves as rational agents.

VUp to this point I have tried to show that through his exposition and deduction of theprinciple of the purposiveness of nature Kant arrived at results in the two Introductionsto the third Critique that are in all essentials equivalent to those affirmed by Leibnizagainst Locke in Book III of theNew Essays. But this result raises a fresh and potentiallytroubling question regarding the distinctiveness of the Kantian position. Might Eber-hard have been right after all in his contention that there is nothing true in the Critiquethat is not already to be found in Leibniz? What makes this question particularlypressing is the recognition that already for Leibniz the principle that “every outerappearance is grounded in the inner constitution” is viewed effectively as a regulativeidea on the basis of which we are rationally constrained to approach nature.Nevertheless, I believe that, despite their substantial agreement, there remains a deep

difference between Leibniz and Kant on this issue, indeed, one that justifies calling theformer a “dogmatist” and the latter a “critical” thinker. This difference does not,however, lie in the nature and status of their respective principles, but rather in thegrounding they provide for them. As I have already indicated, even though Leibnizdoes not deal explicitly with this issue of grounding in his polemic with Locke, it isclear that he took it to be provided by the basic principles of his monadologicalmetaphysics. Thus, in the last analysis, what justifies the presumption of the outer–inner correlation or of natural kinds for Leibniz is the doctrine of the pre-establishedharmony, metaphysically construed as holding between the sensible and intelligibleworlds.


For Kant, by contrast, the grounding of the principle of purposiveness is purelyimmanent, as is befitting its subjective status and connection with reflective judgment.Thus, Kant claims that through this principle “judgment prescribes not to nature(which would be autonomy) but to itself [my emphasis] (which is heautonomy), alaw for the specification of nature” (KU 5: 185–6). The term “heautonomy,” isintroduced in order to emphasize the reflexive, self-referential nature of this princi-ple.33 To claim that judgment is heautonomous in its reflection is just to say that it isboth source and referent of its own normativity. Consequently, even though theprinciple concerns nature as the object of investigation, its prescriptive force is directedback to judgment itself. And this, I believe, both constitutes the main differencebetween the Kantian and Leibnizian positions on the issues discussed here and providesthe warrant for viewing the third Critique as containing a “true apology” for Leibniz.

33 See essay 11 note 17.

200 PART I I I

Essay Thirteen

Kant’s Antinomy ofTeleological Judgment

Like many other Kantian texts, the antinomy of the teleological power of judgment inthe third Critique is deeply puzzling. At the heart of the puzzle is Kant’s contention that“all appearance [Anschein] of an antinomy” between the mechanistic and teleologicalprinciples stems from the confusion of a principle of reflective with one of thedeterminative power of judgment (KU 5: 389). Not surprisingly, this has given riseto the widespread view that the resolution of the antinomy consists entirely, or at leastprimarily, in the demotion of these principles from constitutive principles of determi-native judgment to merely regulative principles of reflective judgment.1

Although this reading is suggested by the text, it raises a number of difficulties thatmake it ultimately untenable. First, assuming, as is usually done, that the principle ofmechanism of the third Critique is equivalent to the causal principle of the first, itsuggests that Kant underwent a remarkable change of mind regarding the status of thisprinciple. Indeed, since it involves demoting the causal principle to merely regulativestatus, this change might be thought to threaten the coherence of the “critical”philosophy.2 Second, given Kant’s views on the requirements for an antinomy, itremains mysterious on this reading why he should have thought that there was eventhe appearance of an antinomy of the power of judgment (an antinomy betweenconstitutive principles would not pertain to the power of judgment). Finally, theassumption that merely appealing to the regulative status of the apparently conflictingprinciples is sufficient to resolve the antinomy leaves us with a monumental exegeticaltask. As has been noted, the passages, including the one cited above, that support thisinterpretation are located at the very beginning of Kant’s discussion in sectionscharacterized as containing the “representation” [Vorstellung] of the antinomy (}70)and “preparation” [Vorbereitung] for its resolution (}71). Thus, apart from whatever

1 A list of commentators who have taken this line up to 1970 is given by McFarland (1970), 120–1.Among more recent commentators, the most elaborate defense of this view is provided by Butts (1984), esp.272–3.

2 Although Kant claims in the first Critique that the Analogies, in contrast to the Mathematical Principles,are “regulative” (A179/13222), this has a completely different sense than the claim that the transcendentalideas are merely regulative, since it is compatible with their objective validity as necessary conditions of thepossibility of experience.

other difficulties it may involve, this reading appears to make the major portion of theDialectic (}72–}78), which contains some of the most fascinating and obscure discus-sions in the entire Kantian corpus, irrelevant to its proposed task.These problems set the agenda for the present essay, the concern of which is largely

exegetical. Its goal is to provide a reading of the text that places the assignment of regulativestatus of the principles within the overall structure of the solution to the antinomy, findsroom for a genuine antinomy of the reflective power of judgment (between merelyregulative principles), and accounts for the role in the argument played by the notions ofan intuitive intellect and a supersensible ground. Before turning to these issues, however, itwill be necessary to examine briefly Kant’s conception of “mechanism,” with particularattention to the manner in which it is conceived in the third Critique.

IOne of the first surprises confronting a reader of the Dialectic of the TeleologicalPower of Judgment is the claim that the principle of mechanism has merely regulativestatus (KU 5: 387). Equally surprising is the suggestion that the constitutive ordeterminative version of the principle, “All production of material things is possibleon mere mechanical laws,” is not demonstrable by reason (KU 5: 387). Does not theSecond Analogy show, or at least purport to show, that all occurrences in thephenomenal world are causally determined according to universal laws and does notKant frequently equate this system of universal law with the “mechanism of nature”?How, then, can he now claim that this is merely a regulative principle and that itcannot be established by reason?The answer is that Kant is doing nothing of the sort and that the attribution to him of

a change in the status of the Second Analogy is due to a failure to grasp the conceptionof mechanism to which he is now appealing.3 To begin with, prior to the writing ofthe third Critique, Kant used the term “mechanism” and its cognates in at least twodistinct senses. First, there is the narrow or technical sense, sometimes termed materialmechanism, which refers to causality by means of interaction between moving particlesin space (matter in motion).4 Mechanism in this sense is closely related to the science ofmechanics, which holds, as its fundamental principle, that all change must have anexternal cause.5 As such, it is contrasted with psychological explanation, in which thecauses are non-material and “internal.” The key point, however, is that both of thesemodes of causation and explanation fall within the scope of mechanism in the extendedsense. The latter, which Kant characterizes in the second Critique as “mechanism ofnature,” is also described as “all necessity of events in time according to natural law”

3 I am here largely following the analyses of Ewing (1924), 229–30, and McLaughlin (1989).4 See, for example, MAN 4: 477, 536. For “material mechanism” see KrV A691/B719 and R 5995 18:

418–19.5 MAN 4: 543. As Kant there makes clear, this is equivalent to the principle of inertia.

202 PART I I I

(KpV 5: 97).6 Only mechanism in this sense is equivalent to the transcendentalprinciple of causality and, as such, it is contrasted with transcendental freedom, notwith psychological causation.Although there are important similarities, the argument of the antinomy turns

largely on an appeal to two senses of mechanism that cannot simply be identifiedwith those noted above. Mechanism, in the main sense in which it is used here, refersto the explanation of wholes solely in terms of the causal interaction of their compo-nent parts.7 As Kant puts it in his most explicit account of the subject:

Now if we consider a material whole, as far as its form is concerned, as a product of the parts andof their forces and their capacity to combine by themselves (including as parts other materials thatthey add to themselves), we represent a mechanical kind of generation. (KU 5: 408)

Mechanism, so described, is closely related to the notion of a material mechanism. Infact, it consists in the application of this conception of mechanism to the specific task ofexplaining natural wholes. On occasion, Kant terms this physical-mechanical explana-tion.8 But whereas in his earlier discussions of mechanism in the narrow sense thecontrast is with psychological causation (the underlying assumption being that bothbelong to the mechanism of nature in the broad sense), the contrast now is with ateleological mode of explanation that appeals to the idea of the whole as the ground orcondition of the parts. We shall soon consider the significance of this latter mode ofexplanation (which Kant sometimes refers to as the “technic of nature”) for theestimation of living organisms. For present purposes, however, the essential point issimply that mechanism in this sense is a regulative principle, since there is nothing inthe Second Analogy that requires all causal explanation to be of this sort.9

Against this it might be argued that Kant indicates that the principle of mechanism isprovided to judgment by themere understanding a priori [ihr der blosse Verstand a priori an dieHand giebt] (KU 5: 386) and that he further insists that “if it is not made the basis for researchthen there can be no proper cognition of nature” (KU 5: 387).10 Together, these claimssuggest that Kant still linksmechanism closely to, if he does not identify it with, the principleof causality and that he must therefore grant it more than a merely regulative status.Such a conclusion would, however, be mistaken. As Kant makes clear in his account

of the peculiarity of the human understanding, the reason why we are constrained tounderstand nature in mechanistic terms has nothing directly to do with the constitutiverole in experience of the principle of causality. It is rather that because of the discursivenature of our understanding, which requires us to move from the “analytic universal” toparticulars (from concepts to intuitions), we are incapable of understanding a “real

6 See also R 5978 and 5995 18: 413 and 418–19.7 This is noted by Ewing and McLaughlin. See note 3 above.8 See, for example, GTP 8:179. Elsewhere Kant refers to physical-mechanical causes (EE 20: 235–6).9 As McLaughlin puts it, “The concept of causality involves a succession in time but not an incasement in

space” (1989), 364.10 See also Kant, EE 20: 235.

KANT’S ANTINOMY OF TELEOLOGICAL JUDGMENT 203

whole in nature” other than as “the effect of the concurrent moving forces of the parts”(KU 5: 407). Mechanism in this sense certainly presupposes the Second Analogy, but itis neither entailed by nor equivalent to it. Accordingly, granting it regulative status doesnot entail any change in status of the causal principle.Finally, this connection of mechanism in the narrow sense with the discursivity of

the human understanding enables us to understand the extended sense of mechanismto which Kant also appeals in the third Critique. Mechanism in this sense encompassesany mode of causality that operates non-purposively.11 Kant introduces this conceptionin connection with his appeal to the problematic idea of an intuitive (non-discursive)intellect. The basic idea is that such an intellect could cognize “mechanistically” whatwe, because of our discursive intellect, can only conceive as the product of anintelligent causality, that is, in terms of a purpose. Since the concept of causality andthe corresponding principle, like all the pure concepts and principles, are merely rulesfor a discursive understanding, they obviously cannot be associated with a non-discursive intellect. Accordingly, this extended sense of mechanism cannot be equatedwith the “mechanism of nature” of the first two Critiques.12

IIKant begins the Dialectic of the Teleological Power of Judgment by considering theconditions under which this power could yield an antinomy. First we are told that in itsdeterminative capacity the power of judgment cannot produce an antinomy because itdoes not have any principles of its own capable of generating one. Thus, if there is to bean antinomy of the power of judgment, it must pertain to it in its reflective capacity, inwhich this power serves as a principle to itself.13 But since the function of the reflectivepower of judgment is to provide maxims or subjective principles of reflection thatenable judgment to find universals (concepts and laws) for given particulars (therebymaking possible the comprehension of nature in terms of empirical laws), it follows thatthere can only be an antinomy of this power if there is a conflict between equally

11 See, for example, Kant, KU 5: 406, 408–9, EE 20: 219.12 This analysis is to be contrasted with that of Marc-Wogau, who distinguishes three senses of mechanism

in the third Critique (1938), 220. The first and widest of these consists in the conception of the determinationof a particular through the concept of causality, entirely apart from the question of the nature of theunderstanding involved. By characterizing the conception in this way, Marc-Wogau underscores its connec-tion with an intuitive or non-discursive intellect. He fails to note, however, that, ex hypothesi, such an intellectwould not make use of the concept of causality (or any other concept) and therefore that this would notcount as a “mechanism” in any but a Pickwickean, purely negative sense. In addition, he fails to include onhis list what I have characterized as the main sense of mechanism in the third Critique. In my judgment, bothmistakes are the result of an understanding of mechanism primarily in terms of the causal concept of theSecond Analogy.

13 Kant’s procedure here parallels his procedure in the Dialectic of the Aesthetic Power of Judgment,where he argues that an antinomy could arise only at the metalevel and that it must concern the principles of acritique of taste rather than taste itself (see KU 5: 337–9). For my analysis of this antinomy and Kant’sresolution of it, see Allison (2001a), 236–54.

204 PART I I I

necessary maxims, both of which have their ground in our cognitive capacities. Thelatter condition is necessary in order to ensure that the illusion generated by the conflict(or appearance thereof) is “natural” and “unavoidable,” therefore requiring a transcen-dental critique to “expose” and “resolve” it.Accordingly, the task is to locate two equally necessary maxims, both of which are

required for the unification of nature in accordance with empirical laws. Withoutfurther argument, Kant affirms that there are, indeed, two such maxims; that they donot seem to be compatible with one another [nicht wohl neben einander bestehen zu konnenden Anschein haben]; and that this apparent incompatibility throws the power ofjudgment into a state of confusion regarding its principles of reflection. These are:“All generation of material things and their forms must be estimated as possible inaccordance with merely mechanical laws” and “Some products of material naturecannot be judged as possible according to merely mechanical laws (judging themrequires an entirely different law of causality, namely that of final causes)” (KU 5: 387).Although the point is usually glossed over, so formulated these maxims do not

merely seem to be incompatible with one another, they are in fact incompatible.14 Thefirst maintains that all production of material things must be estimated mechanisticallyand the second that some cannot be. Since this is a direct contradiction, it is surprisingto find Kant suggesting that a contradiction arises only when these maxims areconverted into the corresponding objective (constitutive) principles of the determina-tive power of judgment: “All generation of material things is possible in accordancewith merely mechanical laws” and “Some generation of such things is not possible inaccordance with merely mechanical laws” (KU 5: 387). In that event, Kant maintains,there would be an antinomy, although one pertaining to reason, not to the power ofjudgment. Conversely, he continues, the two above-mentioned maxims of the powerof judgment do not contradict each other because to claim that I must estimate allproduction of material beings as possible only in accordance with mechanical laws isnot to claim that they are possible in terms of such laws alone. This suggests that thecontradiction is to be removed by converting ontological claims about conditionsunder which things are possible into epistemic claims about the conditions underwhich they may be conceived or estimated as possible. As we have already seen,however, this merely relocates rather than removes the contradiction.In addition to the puzzle about why Kant apparently thought that the contradiction is

removed by converting the propositions from constitutive to regulative status, we havethe question of what has happened to the sought-for antinomy of the reflective power ofjudgment. Initially we were told that such an antinomy would require the possibility ofspecifying two incompatible but equally necessary maxims, both of which have theirground in our cognitive capacities (in this case the reflective power of judgment). Now,however, we apparently learn that precisely because the principles in question are merely

14 This was noted by Hegel (1969), 739. More recently, it has been reaffirmed by McLaughlin (1989),361, who also refers to Hegel as the first and (as far as he knows) only other one to grasp this point clearly.


maxims of this power they donot conflict and there is no antinomy.15Kant thus appears atone stroke both to have ignored the antinomy that he does in fact describe and to havedenied the possibility of the antinomy that he endeavors to present.The first part of the puzzle can be dealt with fairly easily. According to Kant’s

analysis, the import of the claim that the principles in question are mere maxims ofreflection is that they are to be regarded as guidelines or directives rather than assertionsof either ontological or epistemic possibility.16 Thus, the thesis, properly construed,says only that “I should always reflect upon them [events in material nature and allforms as their products] in accordance with the principle of the mere mechanism ofnature” (KU 5: 387). Since this requires only that we investigate nature as far as possiblein accordance with the principle of mechanism, it leaves ample room for the appeal toan alternative principle, on the occasions when this seems to be required. Correlatively,the second assumes that the search for mechanistic explanation be carried as far aspossible, but affirms the necessity of appealing to the teleological principles for theunderstanding of a range of biological phenomena (natural purposes). This does notconflict with the maxim of mechanism because, as Kant insists:

[I]t is not thereby said that those forms [biological phenomena] would not be possible inaccordance with the mechanism of nature. It is only asserted that human reason, in the pursuitof this reflection and in this manner, can never discover the least basis for what is specific in anatural end. (KU 5: 388)

This simple solution to the first part of the puzzle serves, however, to exacerbate thedifficulties with the second. Since the guidelines or recommendations, so described,obviously do not conflict, it becomes difficult to see how even the appearance of anantinomy could arise for the power of judgment. In short, we need an account of howthe reflective power of judgment could fall victim to “illusion” and therefore succumbto a “natural” and “inevitable” dialectic.The heart of Kant’s response to this problem lies in his contention that the antinomy

arises from the confusion of a principle of the reflective with one of the determinativepower of judgment. Otherwise expressed, Kant’s claim is that an antinomy, or at leastthe appearance of one, is generated by the failure to recognize that these principles havemerely regulative status as maxims of reflection.17 Of itself, however, this seemsinsufficient to produce anything like an antinomy. Given Kant’s general conception

15 Marc-Wogau (1938), 225, n. 11, distinguishes helpfully between the apparent but resolvable conflictbetween the maxims of the reflective power of judgment and the real and unresolvable conflict between theconstitutive principles, which would not be an antinomy of the power of judgment.

16 This point is clearly expressed by Butts (1984), 272–3, who emphasizes that regulative principles cannotlogically contradict one another because they lack propositional form. Although I follow Butts on this point,it should be clear by now that I do not accept his contention that this is sufficient to resolve the antinomy.

17 Up to this point I am in agreement with Marc-Wogau, who claims that the antinomy arises becauseeach maxim involves a claim of objective validity (1938), 230. He does not, however, attempt to link thiswith Kant’s general theory of antinomies according to which such a claim must be shown to be “natural” and“inevitable” apart from a transcendental critique.

206 PART I I I

of an antinomy, this would require both that the confusion be “natural” and, therefore,undetectable without a transcendental critique, and that it rest upon the tacit assump-tion of some principle that generates two equally compelling but apparently incom-patible alternatives.With regard to the first point, the essential factor is that, absent a transcendental

critique, regulative maxims, such as the two at issue, seem to involve some kind ofontological commitment.18 That is to say, if I am committed to the search formechanistic explanation as a single and sufficient research strategy, I am not onlypresupposing that all phenomena in nature can be accounted for in mechanisticterms, I am also committed to the belief that they were in fact produced mechanisti-cally, or at least in a manner compatible with their being explicated in these terms.Correlatively, if I hold that certain classes of phenomena (organic beings) require adifferent mode of estimation, it is because I am committed to a belief about how suchphenomena are possible. More specifically, I am committed to the view that thepurposiveness of organic beings is comprehensible only as the product of an “architec-tonic understanding” or, equivalently, an “(intelligent) world-cause acting in accor-dance with ends” (KU 5: 388–9). In both cases, the slide from (or confusion between)the methodological and ontological levels appears to be “natural” in the sense that it isbased on assumptions that one would normally make, unless instructed otherwise by atranscendental critique.19

It should also be clear that this slide generates a conflict at the ontological level(between two views regarding possibility) that does not occur when the principles areconstrued purely methodologically. Moreover, as is requisite for an antinomial dispute,both parties share a common assumption, which they interpret in contradictory ways.In the present instance, the assumption is that purposiveness entails intelligent causalityor design.20 Rejecting on a priori grounds any explanatory role for such a conceptionof causality in the investigation of nature, the (dogmatic) mechanist naturally adopts areductive strategy in dealing with apparent purposiveness. Correlatively, operating

18 In the original version of this essay I neglected to include the qualification “absent a transcendentalcritique” at this point, which gave the erroneous impression that I was claiming that regulative principlesappear to involve an ontological commitment. The point that I was trying to make and have often made insubsequent writings, however, is that such a critique is necessary in order to abandon the standpoint oftranscendental realism, which might be regarded as the natural “default” standpoint of philosophicalreflection, and that from that standpoint it is impossible seriously to entertain the notion that a principlecould be both merely regulative and yet of genuine normative force. In short, for a transcendental realist therecan be no genuine normative force without a corresponding ontological commitment. I discuss this issuefrom various perspectives in several of the essays included in this volume. I am grateful to my commentator,Klaus Kaehler, for calling this error. See Kaehler (1991), 47.

19 Kant claims that if we were to regard material things as things in themselves the dogmatic teleologist’santi-mechanistic claim would follow. See KU 5: 408–9. Since regarding material (and immaterial) things asthings in themselves is the defining feature of transcendental realism for Kant, I take him to be claiming thatthe dogmatic teleologist is committed to transcendental realism and that this prevents the latter fromrecognizing the merely regulative status of the principle of purposiveness. See also the preceding note.

20 This is noted by Zumbach (1984), 12.


with the same assumption, the (dogmatic) teleologist takes the same apparent purpo-siveness as evidence of a distinct type of causality. Thus the antinomial conflict arises.If this diagnosis is correct, a successful resolution of the antinomial conflict would at

least require showing both that neither form of dogmatism is capable of accounting forthe apparent purposiveness exhibited by organic phenomena and that it can beaccounted for in a manner that avoids any ontological commitment. This would justifythe claim that the two principles in question have merely regulative status, which in the“preparation” is simply asserted rather than demonstrated.

IIIKant’s attempt to establish these points begins with the seemingly paradoxical claimthat, “No one has doubted the correctness of the fundamental principle that certainthings in nature (organized beings) and their possibility must be judged in accordancewith the concept of final causes” (KU 5: 389). Since the rejection of the concept offinal causes lies at the very heart of the scientific revolution of the seventeenth centuryand, with the notable exception of Leibniz, is a virtual commonplace in the philosophyof the time, one naturally wonders what Kant could have had in mind. Nor is thispuzzlement completely removed by the claim that the concept of final causality isuniversally recognized to be necessary at least as a “guideline” [Leitfaden] and thereforethat the only question is whether it has merely subjective validity as a maxim ofjudgment or is an “objective principle of nature” (KU 5: 390). Certainly manyphilosophers—Descartes, Hobbes, and Spinoza, for example—explicitly denied anyepistemic significance to this concept or principle.Kant soon makes clear, however, that he has precisely such views in mind; indeed,

Spinoza is one of the philosophers whom he explicitly discusses in this context.21

Accordingly, the most reasonable course is to construe Kant’s claim to be that no onehad denied the apparent significance of the concept of final causes for the investigationof nature, which is a more plausible thesis, since even Spinoza found it necessary tooffer an explanation of the confused belief in final causes.22 So construed, the issue iswhether some things in nature (living organisms) merely seem purposive or whetherthis is an objective feature of them. As one might expect, common to both alternativesis the assumption that real purposiveness entails designedness. Thus one position deniessuch purposiveness because it rejects designedness, while the other appeals to someversion of designedness to explicate this very purposiveness. The former Kant labels the“idealism” of purposiveness and the latter the “realism” thereof. Since Kant uses bothterms in a pejorative sense, it is clear that we are to construe them as analogues ofempirical idealism and transcendental realism, both of which Kant attacks in the first

21 For my analysis of Kant’s view on Spinoza, including an account of his Spinoza critique in the thirdCritique, see Allison (1980).

22 See Spinoza (1982), 57–62. (Appendix to part one of the Ethics.)

208 PART I I I

Critique, rather than of the transcendental idealism and empirical realism that heaffirms.23

The intent of the argument is to show that neither form of dogmatism is able toaccount for the phenomenon of purposiveness, which in some form or other isuniversally acknowledged. The idealism of purposiveness (whether construed withGreek mechanism and its modern day versions as the “casuality” [Casualität] ofpurposiveness or with Spinoza as the “fatality” [Fatalität] thereof) fails to do the jobbecause it cannot provide a viable alternative to the notion of an intelligent cause. Toappeal to pure chance, like the Epicureans, is to abandon the attempt to render thephenomenon intelligible rather than to provide an explanation. Moreover, Spinoza’sattempt to derive it from a first cause that acts by virtue of the necessity of its naturefares only slightly better.24 Correlatively, the realism of purposiveness (which can takethe form either of theism or hylozoism) likewise fails because it cannot establish the(real) possibility of the supersensible principle to which it appeals.Since, ex hypothesi, we must conceive of living organisms as purposive (natural

purposes) and since we have seen that no dogmatic position (whether it be idealist orrealist) can account for the possibility of this conception, Kant concludes that only thecritical path remains. As Kant here describes this path, it consists in the consideration ofa concept “in relation to our cognitive faculties hence in relation to the subjectiveconditions for thinking it, without undertaking to decide anything about its object”(KU 5: 395). In the present case, the result is that the necessity of considering thepurposive arrangement as the product of design (an intelligent cause) is purely a subjectiveone, due to the peculiar nature of our cognitive capacities. By this move, Kant at onestroke both accounts for the key assumption underlying the whole debate and shows that,properly construed, it does not carry with it any ontological commitment.If this is to be at all convincing, however, Kant must indicate this mysterious peculiar

feature of our cognitive capacity and show just how it makes the appeal to designednesssubjectively (but merely subjectively) necessary for the estimation of living organisms.Moreover, despite the overall obscurity of his account, which is due largely to hisspeculations regarding the problematic notion of an intuitive intellect, Kant’s answer tothe first question is relatively clear. The relevant feature of our cognitive capacities isthe discursivity of our understanding. In short, the same discursivity that underlies thenecessity of estimating natural wholes mechanistically also accounts for the necessity ofregarding living organisms in light of the idea of an intelligent cause.The claim that our intellect is discursive is equivalent to the claim that human cognition

depends on the subsumption of sensibly intuited particulars under concepts.25 This is thepicture of cognition that underlies the fundamental distinction in the first Critiquebetween sensibility and understanding, as well the corresponding insistence that cognition

23 On Kant’s reference to idealism and realism in this context see Philonenko (1988), esp. 32.24 For the details of this see Allison (1980), 214–18.25 In later writings I refer to this as the “discursivity thesis.”


requires the unification of both factors. In the present case, however, Kant is concernedwith the conditions of the possibility of the subsumption of given particulars under theuniversals of the understanding. The basic problem,which is central to the thirdCritique asa whole, is that the universal principles underdetermine the particulars falling under them.Thus, on the one hand, cognition requires a harmony orfit between the universals and theparticulars to be subsumed under them, while, on the other hand, when one goes beyondthe transcendental conditions of experience to which particulars necessarily conform, thisfit turns out to be a contingent matter.Kant deals with this general problem in both versions of the Introduction, providing

what amounts to a transcendental deduction of the “formal purposiveness” of nature,that is, of the (subjective) necessity for the power of judgment in its reflective capacityto presuppose that nature is specified in its empirical laws in a manner consonant withthe requirements of the understanding.26 In the present context, he appeals to the samecontingency in order to argue for the subjective necessity of a very different conceptionof purposiveness (“real” or “material” intrinsic purposiveness).27 Since an essentialfeature of this argument has already been touched upon in the discussion of mecha-nism, we can spell out the main points fairly briefly.To begin with, as discursive, our understanding must proceed from what Kant calls

the “analytic universal” to particulars, that is, from abstract, generic concepts to theconcrete objects of empirical intuition (KU 5: 407). Once again, it follows from thisthat the understanding cannot, of itself, determine what is specific in these particulars,which means that we have the familiar contingency of fit problem. The new conten-tion is that this contingency determines the way in which our understanding isconstrained to think of wholes in nature.According to Kant, there are two ways in which a discursive understanding can

comprehend the possibility of such wholes. The first is the previously discussedmechanistic way, which is the only one that yields genuine scientific explanation. Aswe have seen, this involves regarding natural wholes as composed of and conditionedby their parts. Since any whole mechanistically conceived is a mere aggregate, theobvious limitation of mechanism is its inability to account for wholes that are not mereaggregates, that is, those in which the whole is thought as prior to and conditioning itsparts and their internal arrangement. Such wholes, Kant maintains, can be conceived aspossible only by considering the representation of the whole as containing the groundof the possibility of its parts, that is, by considering the whole as the product of anintelligent cause or design (KU 5: 408). This is the second way of conceiving of wholesand it is based on the model of intentional production or art. Moreover, since Kantclaims to have shown in the Analytic of Teleological Judgment that because of their

26 See Kant, KU 5: 181–6; EE 20: 211–16. I discuss this deduction in Allison (2001b) and (2003), whichare included in this volume.

27 For analyses of the different senses of purposiveness in the third Critique, see Marc-Wogau (1938),69–72, and Tonelli (1957–8), 154–66.

210 PART I I I

epigenetic features, living organisms must be regarded as just such wholes, it follows,given the peculiarity of our understanding, that we can only represent such beings toourselves as products of design.28

Assuming the argument of the Analytic of the Teleological Power of Judgment andthe basic outlines of Kant’s theory of the human understanding, this suffices to establishthe first part of the desired result, namely, that our representation of living organisms aspurposive reflects a peculiarity of our understanding. It is not sufficient, however, toshow that this is due merely to a peculiarity of our understanding, since for all that wehave seen so far, it remains possible that every conceivable understanding (includingGod’s) must conceive living organisms in precisely the same manner. Moreover, if thiswere the case it would follow (according to Kant) that the principle of purposiveness isobjectively valid as a constitutive principle. Accordingly, it is essential to Kant’s projectto show that this peculiarity does not attach to every conceivable understanding.Since this peculiarity was shown to be a consequence of the discursiveness of our

understanding, Kant’s strategy is to conduct a thought experiment in which weconceive of the possibility of a non-discursive understanding or intuitive intellect,which would be an understanding “in the most general sense of the term” (KU 5: 406).Our conception of such an intellect is completely negative and, as such, parasitic onthat of the familiar discursive variety with which it is contrasted. The salient point,however, is that such an intellect would not move from the “analytic universal” to theparticular and therefore would not experience the contingency of fit that is endemic toour way of cognizing. Instead, it would operate by means of “the syntheticallyuniversal (of the intuition of a whole as such)” and proceed from this whole to itsparts (KU 5: 407). Consequently, it would not need to appeal to the idea of anintelligent cause in order to represent to itself the parts as dependent on the whole,which is to say that it would “conceive” of natural products in light of the extendedand purely negative sense of mechanism noted earlier.Kant takes the bare conceivability of such an intellect to show both that our

discursive understanding is not the only (logically) possible form of understandingand that our peculiar manner of estimating living organisms reflects a merely subjectivenecessity, which, as such, cannot lay claim to objectivity, even with respect toappearances (KU 5: 408). Since in the first Critique Kant in effect regards space, time,and the categories as grounded in the peculiarity of our understanding (and sensibility),while still affirming their objective reality with respect to the phenomenal realm, thisdisclaimer of all objectivity might appear strange. It should be kept in mind, however,that purposiveness (like mechanism) is a principle of the reflective power of judgment,which functions regulatively in light of the governing ideal of formal purposiveness orsystematically. Accordingly, here as elsewhere in the third Critique it is a matter of the

28 For a useful discussion of the issues involved here in light of contemporary views in biology and thephilosophy of biology, see Zumbach (1984), 1–29.


power of judgment legislating to itself, not to nature, that is to say, of the “heauton-omy” rather than the “autonomy” of this power.29

IVThis connection with formal purposiveness or systematicity is also the main reason whyKant insists that the resolution of the antinomy requires not merely showing that themechanistic and teleological principles are, qua maxims, logically compatible but alsothat they are unifiable [vereinbar] in a single, higher order principle. As Kant sees it, ifthis were not the case, “they could not subsist alongside one another in the consider-ation of nature” (KU 5: 412). In other words, without such unifiability, at least inprinciple, the regulative idea that empirical laws cohere in a system would have to beabandoned in the biological domain, since mechanistic and teleological modes ofestimation could not be combined in a single research program.Moreover, the systematicity requirement cannot be met simply by eliminating

mechanism completely from biology. Unlike products of art, biological phenomenaare parts of nature (natural products) and, as such, must be susceptible to mechanisticexplanation (KU 5: 413–14). Since mechanism cannot be eliminated, while teleologi-cal reflection is required if one is even to begin to conceptualize biological phenomena(grasp them as organized), the only alternative is to subordinate the mechanistic to theteleological principle. This requires that the two principles be unifiable and Kantconcludes that our warrant to assume this is provided by the governing transcendentalprinciple of purposiveness (KU 5: 414).The question, then, is how this unifiability is to be conceived. Since both must be

preserved, it cannot be by means of a reductive strategy, which would seek to explainorganic phenomena entirely in mechanistic terms.30 Despite the indispensability of theprinciple of mechanism, Kant remarks that going to the extreme of trying to explaineverything in such terms would leads reason to lose itself in fanciful postulations that

29 For the distinction between heautonomy and autonomy see Kant, KU 5: 185 and EE 20: 225.30 This corresponds to what Zumbach terms “explanatory reductionism,”which in his reading is the view

that Kant explicitly rejects. He distinguishes this from both an “ontological reductionism,” which maintainsthat biological processes are describable and have causes that are describable in physiochemical terms, and a“methodological reductionism,” which requires us to investigate nature as if it were explicable in purelymechanistic terms. To each of these there corresponds a form of anti-reductionism. According to Zumbach,Kant’s explanatory anti-reductionism does not commit him to either the ontological or the methodologicalvarieties. In fact, he holds both that Kant is not an ontological anti-reductionist, since this would amount tothe kind of dogmatic view he criticizes, and that he is “to some degree” even a methodological reductionist,since he maintains that we should always reflect upon vital phenomena as far as possible in accordance withthe principle of the mere mechanism of nature (KU 5: 387). See Zumbach (1984), 87–100. I believe thatZumbach is correct in denying that Kant is an ontological anti-reductionist; but even granting the distinctionbetween a methodological and an explanatory reductionism, I find the attribution to Kant of even a weakform of the former somewhat puzzling. Kant’s claim, after all, is that we ought to proceed in our investigationof nature as far as possible (my emphasis) in accordance with the principle of mechanism, which alreadysuggests the possible need for another principle.

212 PART I I I

are no better grounded than those suggested by a teleological approach that simplyignores the mechanism of nature (KU 5: 411).31 Nor can unification be achieved bytreating the two principles as conjointly constitutive, that is, as both involved in theexplanation of the production of a given object. As Kant indicates by appealing to theclassical example of a maggot, each of these principles, taken constitutively, is incom-patible with the other, since they would yield conflicting accounts of how the genesisof an organism is possible (KU 5: 411).From this Kant concludes that the only way in which we can regard these two

principles as unifiable is to think of them as jointly derivable from some higher orderprinciple that underlies but is distinct from both of them. But such a principle would,by definition, have to be something supersensible. As Kant points out, however, theonly conception of the supersensible that is available for theoretical purposes is the“undetermined concept of a ground that makes the judging of nature in accordancewith empirical laws possible” (KU 5: 412). Although as supersensible and completelyindeterminate, such a conception cannot be used to explain how the teleological andmechanistic principles can be unified, it does allow room for the conceivability of theirunification, which is all that is necessary.Accordingly, the resolution of this antinomy, like that of the Antinomy of Taste in

the Dialectic of the Aesthetic Power of Judgment and the Third Antinomy in theCritique of Pure Reason, culminates in an appeal to a supersensible ground. At least in thepresent case, however, this ground seems to be little more than the principle ofpurposiveness itself. Once again, the key is the contingency of fit. To claim that nature,in its empirical lawfulness, is contingent with respect to the transcendental conditionsof experience is just to say that such lawfulness is not a function of these conditions,which means that it cannot be regarded as a “contribution of the subject.” But sincethis lawfulness is necessarily presupposed by the subject as a condition of the possibilityof empirical knowledge, it likewise cannot be regarded as a merely empirical matterthat might break down at any moment. Consequently, the only way in which we canthink of it is to attribute it to some supersensible ground, which, as in all Kant’s talkabout the noumenal (at least from the theoretical point of view), functions as aplaceholder for our ignorance.It is worth noting, however, that a supersensible ground would be accessible only to

an intuitive intellect, which suggests that the conception of such an intellect functionsat two points and in two quite different ways in Kant’s overall argument. Its initialfunction, which has already been discussed, is completely negative. By providing analternative to our discursive understanding, it enables Kant to drive a conceptual wedgebetween the peculiarities (or conditions) of our understanding and those of anyunderstanding, which underlies Kant’s denial of ontological import to whatever canbe shown to be due to the former. Now, by contrast, in its role as correlate to a

31 Kant makes a similar claim in GTP 8: 178–80.


supersensible ground, it functions positively to license the thought of the unifiability ofmechanism and teleology in a higher common ground.32

Finally, this puts us in a position to survey the complex relationship between thetranscendental principle of the formal purposiveness or systematicity of nature and theteleological principle. As Kant insists, the presupposition of formal purposiveness doesnot of itself justify the quite distinct assumption that nature exhibits a material or realpurposiveness, either in some of its products taken individually or in their interrelations(intrinsic and extrinsic purposiveness) (KU 5: 193; EE 20: 233–34). Nevertheless, itdoes require us to investigate organic beings in light of the model of purposivenessbecause only in this way can we even begin to conceptualize them.33 In addition, wehave seen that the same transcendental principle both requires us to assume that themechanistic and teleological principles are unifiable and provides whatever contentthat can be given to the indeterminate idea of the supersensible ground of nature inwhich they conceivably cohere. Admittedly, this still leaves us with many problems,including the connections between this antinomy and the Antinomy of Taste on theone hand and the Third Antinomy of the Critique of Pure Reason on the other. But thetreatment of these topics must be reserved for other occasions.

32 This twofold function of the intuitive intellect is noted by Düsing (1968), 68.33 The main point here is that the idea of purposiveness, as the principle of the reflective power of

judgment, is required even to acquire empirical concepts of organic beings. See particularly EE 20: 235.I discuss this thesis in Allison (2000c), (2001b), and (2003), which are included in this volume.

214 PART I I I

PART IV

Essay Fourteen

The Gulf between Natureand Freedom and Nature’sGuarantee of Perpetual Peace

One of the more overlooked features of Kant’s philosophy is its recognition of twodistinct freedom–nature problems. The first is the familiar problem of the compatibilityof the causality of freedom with the causality of nature, which is dealt with inthe resolution of the Third Antinomy in the first Critique. The second is posed in theCritique of the Power of Judgment. In the second section of the published Introduction tothe work, Kant refers to the “immense gulf” [grosse Kluft], which separates thesupersensible from appearances and which therefore seems to stand in the way oftheir influencing one another (KU 5: 175). This is seen as a problem because action inaccordance with “laws of freedom” and the final purpose [Endzweck] prescribed bythese laws (the highest good) ought to be actualized in the world of sense. Thus, insteadof a rigid separation between the realms of freedom and nature, which was thoughtnecessary in order to allow space for freedom, Kant now insists on the necessity of amediating concept (the purposiveness of nature), which would make possible thetransition [Übergang] from the concept of nature to the concept of freedom (KU 5:195–6).In Kant and the Experience of Freedom, Paul Guyer calls attention to this presumed gulf

and suggests that it must be seen as a consequence of significant changes in Kant’s moraltheory subsequent to the second Critique.1 According to Guyer, Kant came to realizethat, despite the unlimited freedom of the noumenal self (its “power to remake thephenomenal world of appearance and its natural laws of causality”), it could onlysucceed in developing a virtuous disposition “by working with, not against, the feelingsof the natural and embodied human agent.” And to this end, both aesthetic andteleological judgment have significant roles “by offering both sensible representationsof key aspects of morality and opportunities for the cultivation of the moral feelings.”2

Although I differ from Guyer on a number of points, particularly his understandingof freedom, I shall not take issue with him on these matters here. Instead, I shall offer analternative account of the gulf, one which attempts to show its relation to moral ends

1 Guyer (1993), 27–33. 2 Ibid., 33.

rather than moral feelings and which focuses on its connections with the reflectivepower of judgment and its principle: the purposiveness of nature. And, as befitting thisoccasion, I shall be particularly concerned with its connection with the ideal ofperpetual peace. In so doing, I am not denying the moral significance of aestheticexperience, but merely affirming that such experience must itself be understood as partof a larger story, with which the third Critique as a whole is concerned.3

The essay is divided into four parts. The first argues that the gulf is a genuine one forKant and that it is the concern of the reflective power of judgment to fill it. The secondshows how Kant’s teleological account of history (the meeting place of natural andmoral teleology) is crucial to this project. The third considers nature’s supposedguarantee of perpetual peace as an expression of Kant’s moral teleology. Finally, thefourth considers some of the difficulties connected with the conception of a guarantee.

IIn order to show both the genuineness of the second freedom–nature problem and itsconnection with the reflective power of judgment, it is useful to consider it in relationto the much discussed epistemological problem of systematicity, which Kant uses tointroduce the conception of the reflective power of judgment and its principle ofpurposiveness. The central point with regard to the latter is that the transcendentalprinciples of the understanding established in the first Critique only guarantee thelawfulness of nature at a very general level. Not only is it the case that particular,empirical laws cannot be derived from these higher level laws, but the latter leave itcompletely undetermined whether, in fact, the order of nature is such that it can bediscovered by the human mind. As Kant puts it in the first Introduction:

For the multiplicity and diversity of empirical laws could be so great that it might be possible forus to connect perceptions to some extent in accordance with particular laws discovered onvarious occasions into one experience, but never to bring these empirical laws themselves to theunity of kinship under a common principle, if, namely, as is quite possible in itself (at least as far asthe understanding can make out a priori), the multiplicity and diversity of these laws, along withthe natural forms corresponding to them, being infinitely great, were to present to us a rawchaotic aggregate and not the least trace of a system, even though we must presuppose such asystem in accordance with transcendental laws. (EE 20: 209)

This scenario, which is possible in the sense that it is not excluded by the transcendentalprinciples of the Analytic of the firstCritique, raises a specter analogous to the one posedin the Transcendental Deduction of that work, when Kant considers the possibilitythat, “Appearances might very well be so constituted that the understanding should notfind them to be in accordance with the conditions of its unity” (KrV A90/B123). The

3 I discuss Kant’s account of the moral significance of beauty in Allison (1997b), (1998a), and (2001a),219–35.

218 PART IV

first Critique’s specter may be characterized as “transcendental chaos” (disorder at thetranscendental level), and it is exorcised by the transcendental unity of apperception.The third Critique’s specter, by contrast, may be termed “empirical chaos” (disorder atthe empirical level), and it is supposedly removed by the reflective power of judgment’sprinciple of purposiveness. The precise nature of this hypothetical empirical chaosremains a controversial matter, but it presumably would include a state of affairs inwhich there were few (if any) observable regularities and those that did obtain do notcorrespond to the intrinsic characters of things. In short, our empirical concepts andclassifications, to the extent to which they were possible at all, would merely reflectsurface features of things, and would not correspond to their inner natures or essentialproperties.4

Although at first glance the nature–freedom problem might seem to have little incommon with this epistemological worry, further consideration suggests some inter-esting parallels. To begin with, it shares with its epistemological counterpart thestructural feature of being analyzable at both the transcendental and empirical levels.At the transcendental level, the question is whether free agency is compatible with thecausality of nature (the transcendental principle of causality) and this question isanswered positively by means of the resolution of the Third Antinomy. At theempirical level, the question becomes whether the ends, dictated by the “laws offreedom” (moral laws) are realizable in the sensible world, that is, in the empirical orderof nature. Thus, whereas the empirical level epistemological worry is whether there is ahumanly cognizable order of nature at all, the parallel practical worry is whether thisorder is such as to allow for the realization of moral demands. Given all that we couldknow a priori, this order might be such as to frustrate the promotion of moral goals, theattainment of anything approaching a kingdom of ends among human beings. In otherwords, the question is whether what is required by (moral) theory is achievable inpractice (in the “real world”).Inasmuch as the “laws” in question are preeminently (though not exclusively)

psychological or anthropological, they do not count as laws in the strict sense forKant.5 Nevertheless, though lacking in genuine nomological necessity, such “laws”presumably depict essential and invariant features of human nature in its “empiricalcharacter.” Foremost amongst these is the propensity to discord or “unsociable socia-bility” [ungesellige Geselligkeit], which, on the one hand, leads human beings to seek thesociety of others, and, on the other hand, generates conflicts that make the socialcondition untenable.6 Since this appears to constitute a significant impediment to theattainment of moral goals, particularly the goal of perpetual peace, it seems that Kant is

4 I discuss the specter of empirical chaos and Kant’s response in Allison (2001b) and (2003), both of whichare contained in this volume.

5 See Kant, MAN 4: 471; and A 7: 161.6 The systematic significance of this conception for Kantian ethics has been explored by Wood (1991b)

and (1999), 283–300.

NATURE, FREEDOM, AND NATURE’S GUARANTEE OF PERPETUAL PEACE 219

correct in insisting in the published Introduction to the third Critique on the necessityof being able to think of nature “in such a way that the lawfulness in its form is at leastin agreement with the possibility of the ends that are to be realized in it in accordancewith the laws of freedom” (KU 5: 176). If we could not regard nature in this way, ourmoral life would be as chaotic as our theoretical life would be without the assumptionof a cognizable order of nature. In both cases the problem is the threat posed byparticularity to universal and necessary ends. Moreover, in both cases the solution liesin the assumption (as a principle of the reflective power of judgment) of the purpo-siveness of nature.This conclusion cannot be evaded on the grounds that the attainment of ends is

irrelevant to Kantian morality. In this respect, the account of the good will in theGroundwork is misleading. Although it is true that the goodness of a good will is not afunction of what it accomplishes in the world, it remains the case that such a will strivesto realize certain goals.7 These include a lawful [rechtlich] condition, that is, a civilsociety under a republican constitution, a condition of perpetual peace between states,and the two ends that are also duties of the Doctrine of Virtue: namely, one’s ownperfection and the happiness of others. More generally, there is the requirement towork for the advancement of the highest good on earth, which is best seen as atotalizing concept encompassing all universally valid ends rather than as a distinct end.8

IIThe turn from Kant’s moral philosophy proper to his moral teleology is marked by ashift of focus from that of the individual to the species. As is clear from Kant’s historicalwritings, nature is thought of as purposive with respect to the race as a whole, not theindividual. Indeed, in the third Critique Kant is quite explicit that the highest good orfinal purpose [Endzweck] of creation, namely, the existence of rational beings undermoral laws and the consequent apportionment of happiness to virtue, is to be taken as acollective goal. To be sure, it is still individuals who will be rewarded or punishedstrictly according to their merits, but Kant views this as conceivable only insofar as therace itself attains the appropriate level of moral development.9

The teleological story is further complicated by Kant’s introduction of a distinctionbetween two senses in which the race can be regarded as an end or purpose. To beginwith, there is the conception of the human race as the ultimate purpose [letzte Zweck] ofnature, that is, the single species with respect to which nature as a whole may bedeemed purposive. Although boldly stated such a conception might seem excessivelyPanglossian, Kant brings it into accord with the basic principles of the third Critique by

7 See Kant, KU 5: 471n. For a discussion of this topic see Düsing (1968), 104–5.8 On the highest good as a totalizing concept see RGV 6: 5. For a discussion of this point see Düsing

(1971), 32–3.9 Kant, KU 5: 447–52.

220 PART IV

qualifying it in two essential respects. First, he deflates it from a dogmatic or determi-native principle to a maxim of the reflective power of judgment. So regarded, itstipulates that (for certain purposes) we may conceive of nature in this way, but thatwe cannot attribute any objective validity to this conception. Second, Kant denies thatthe purposiveness of nature is to be thought of as its arrangement for human happiness.On the contrary, he insists that if nature is to be regarded as purposive with respect tothe human race, it can only be in virtue of what is unique to our species (at least as far aswe can know), namely, the capacity to set and act according to ends. And since Kantidentifies culture with what produces and enhances this uniquely human aptitude forsetting ends, he contends that culture (of the human race) is the only viable candidatefor the ultimate purpose of nature (KU 5: 431).As Kant points out, however, not any form of culture qualifies for this ultimate

purpose, but only one which prepares humanity for its higher (moral) vocation.Accordingly, it is necessary to distinguish between two types of culture: the cultureof skill [Geschicklichkeit] and the culture of discipline [Zucht]. The former is the capacityto attain the ends that humanity collectively sets for itself (including physical comforts).It is here that political institutions become central as the means to obviate the threat ofviolence that stems from the inequality and oppression, which Kant, like Rousseau,views as inseparable from the development of culture in this sense. From this point ofview, purposiveness is manifested primarily in what nature does to prompt the humanrace to develop republican forms of government, which make possible the rule of lawwithin states and international confederations, which help to preserve the peacebetween states.The culture of discipline is characterized as the “liberation of the will from the

despotism of desires” (KU 5: 432). Although Kant depicts this form of culture inexplicitly negative terms, it is clear that he has in mind the development of an interestin ends that are independent of sensuous desire, specifically, aesthetic and intellectualends. While agreeing with Rousseau that of itself the development of such interests andpropensities makes humanity merely civilized [gesittet] rather than moral [sittlich], Kantalso contends (against Rousseau) that this development helps to prepare the race “for asovereignty in which reason alone shall have power . . . and allows us feel an aptitudewithin us for higher ends, which lie hidden in us” (KU 5: 433–4). Accordingly, it isonly if nature is also regarded as promoting this latter form of culture that it can bethought to have humanity as its ultimate purpose.Even though it may be the ultimate purpose of nature, culture, including the culture

of discipline, is not itself an unconditioned good. Indeed, its status depends on theexistence of a higher, unconditioned purpose that it subserves. The latter is the finalpurpose of creation, which Kant characterizes as the existence of rational beings “undermoral laws” (KU 5: 448–9). Since this is not something that nature could of itself bringabout, it cannot be viewed as a purpose of nature in the strict sense. Nevertheless, Kantmaintains that, insofar as nature is viewed as a spur to the development of culture inboth forms, it can be considered as indirectly facilitating the realization of this final


purpose. In order to conceive of nature as purposive in this way, however, it isnecessary to regard it as dependent on a moral author, that is, God in the propersense. And with this we make the transition from physical to moral teleology (andtheology), which is Kant’s main concern in the final sections of the third Critique.Although we cannot here pursue the details of Kant’s account of the final purpose

and the moral proof for the existence of God with which it is inseparably connected, itmust be noted that he introduces what appear to be two distinct conceptions of thefinal purpose, which are only subsequently connected. The first is the previouslymentioned existence of rational beings or the human race under moral laws, whichKant is careful to distinguish from the conformity to such laws (KU 5: 448n). The claim isthat if there is to be a final purpose of creation, this is the only conceivable candidate.What is noteworthy about this conception is that, as the final purpose of creation(which, as such, can only be conceived by means of an appeal to a moral author) it isneither our purpose nor something that we can regard as a duty to promote.By contrast, the second conception of the final purpose, which Kant identifies with the

highest good and introduces in connectionwith themoral proof, is happiness in harmonyor proportion with morality. This is the final purpose for us, the highest good in the worldthat we can attain through the use of our freedom (KU 5: 450) and, as such, it is one thatwe have a duty to promote. Only in }88 does Kant bring these two conceptions togetherand connect them with the nature of the reflective power of judgment. The main pointthere is that only insofar as we recognize a final purpose in the first sense (of creation) canthere be one for us and therefore a duty to strive for its realization.Elsewhere Kant goes into more detail regarding this final purpose for us, characterizing

it as an ethical community [ethische Gemeinwesen] (RGV 6: 94), which corresponds to thekingdom of ends of the Groundwork realized on earth, that is, a secularized version of thekingdom of God. Such a community is distinguished from the juridical state in that itinvolves a union of human beings under non-coercive laws (laws of virtue) rather thancoercive juridical laws. As such, it cannot be brought about through juridical legislation orpolitical revolution, but requires a collective change in the moral disposition.Kant claims that this idea has objective practical reality, which means that we have a

duty to work for its realization, indeed, “a duty sui generis, not of human beings towardhuman beings, but of the human race towards itself” (RGV6: 97). The explanation of thisunique duty lies in the need for moral regeneration, an overcoming of the “sovereignty”of the evil principle, which is the central theme of the work. As Kant views this matter,such regeneration cannot be fully attained by the isolated individual, since the basic sourceof corruption is to be found in society, our relations with other human beings (RGV 6:93–4). In short, Kant asserts that no one can be truly virtuous unless all are virtuous, or atleast until one lives in a society composed primarily of virtuous individuals.10

10 Wood has argued for a “communitarian” reading of Kant in which virtue is a “collective and even asocial end” (1991b), 332. I discuss this matter in Allison (2009a), which is included in this volume.

222 PART IV

Although Kant apparently envisaged this ethical community as a future historicalcondition, its connection with his philosophy of history remains somewhat obscure. Inthe essays devoted to the latter, Kant tends to link purposiveness in history primarilywith the development of republican form of government and the confederation ofstates required for perpetual peace, that is, with the perfection of culture and thereforethe ultimate purpose of nature rather than the final purpose of creation. And since Kantnever wrote a systematic, comprehensive treatise on history, we have no definitiveaccount of the historical connection between these two purposes and their realizations.Nevertheless, hints of what Kant may have had in mind are contained in his

historical writings. Perhaps the most intriguing and widely noted of these is thestatement that the union of states, which would lead to perpetual peace, constitutesmerely “the halfway mark in the development of mankind” (IAG 8: 26). Assuming, asseems reasonable, that the still distant halfway point in history corresponds to theperfection of culture,11 it may be surmised that the task of the second half is to progressfrom this to moral perfection, or at least to the victory over the evil principle marked bythe attainment of an ethical community. In the terms of Allen Wood, the first half maybe termed the “epoch of nature” and the second half the “epoch of freedom.”12

IIIKant frames the argument of Perpetual Peace as the response of the idealist politicaltheorist to the cynical political realist (practical politician), who disparages utopianschemes of all sorts. Thus, his opponent is the politician who subscribes to the motto:“That may be correct in theory, but it is of no use in practice,” which is the title of the1793 essay in which Kant discusses the theory–practice problem in a broader context.What is distinctive about Kant’s response is that he attempts to meet the realist on his

own ground, suggesting that the very aspects of human nature to which the latterappeals in ridiculing the practicality of the ideal of perpetual peace are the means bywhich we can expect it to be eventually brought about.Although in many ways it constitutes the philosophical center of the work, this

argument is located in a “supplement” [Zusatz], “On the Guarantee of PerpetualPeace,” which follows upon Kant’s presentation of the preliminary and definitivearticles of a putative treaty that would supposedly lead to a state of lasting peaceamong nations. Kant begins by stating that the guarantor is “nothing less than thegreat artist [Kunstlerin] nature . . . from whose mechanical course purposiveness shinesforth visibly, letting concord arise by means of the discord between human beings evenagainst their will” (EF 8: 360). Then, after noting that this process may be termed“providence” insofar as we discern in it the wisdom of a higher cause predeterminingand directing the course of nature to the objective final purpose of the human race, he

11 At one point Kant equates perpetual peace with the perfection of culture (MAN 8: 121).12 Wood (1991b), 343.


continues: “We do not, strictly speaking, cognize in these artifices of nature or even somuch as infer from them but instead (as in all relations of the form of things to ends ingeneral) only can and must add it in thought, in order to make for ourselves a concept oftheir possibility by analogy with actions of human art” (EF 8: 362). What is added inthought is the guarantee that “what man ought to do in accordance with laws offreedom, but does not do . . . he will do, without prejudice to this freedom, even by aconstraint of nature” (EF 8: 365). And again, “[N]ature wills irresistibly that rightshould eventually gain supremacy.What we here neglect to do eventually comes aboutof its own accord” (EF 8: 367).Despite its semi-popular tone, Kant’s position here is essentially that of the third

Critique. First, there is an explicit appeal to the conception of nature as art, or the techne ofnature, which is perhaps the central conception of the third Critique as a whole.13

Second, purposiveness is linked specifically with the form of things, as it is consistently inthe Critique of the Power of Judgment. Third, as in the third Critique, Kant here contendsthat purposiveness is read into nature rather than being either directly observed orinferred from it. In short, it is a contribution of the subject, a presupposition ofreflection, which is deemed necessary in order to conceive of nature as amenable tothe realization of a final purpose. Once again, the operative assumption is that suchamenability cannot be viewed as the contingent result of the operation of merelymechanical causes and that (given the nature of our conceptual apparatus) the onlyalternative is to conceive it in terms of the analogy with human intentional activity, thatis, as the product of art [Kunst], though not, of course, of fine art [schone Kunst].The teleological account that Kant provides on the basis of the conception of nature as

art falls into two parts. The first considers “what nature does for its own endwith respect tothe human race as a class of animals” and then asks: “What nature does for this purposewith reference to the end that the human being’s own reasonmakes a duty for him” (EF 8:365)? Not surprisingly, war and the propensity thereto turn out to be central to both.Kant characterizes what nature has done with respect to the human race considered

as a class of animals as “preparatory arrangements” and identifies three contributions:(1) By various devices, the most remarkable being the availability of wood in barrenclimes, nature has made it possible for humans to live in every part of the earth. (2) Bymeans of war, which drives people apart, she has ensured that even the most inhospi-table regions are inhabited. (3) By the same means, she has forced human beings toadopt more or less lawful relations with each other, that is, she has made some form oflegal order necessary (EF 8: 363).Presumably, the latter two “arrangements” pertain to culture, or at least the culture of

skill.14 Thus, echoing Rousseau, Kant connects the story of the human race’s populationof the whole earth and socialization with the domestication of animals, the development

13 This is argued persuasively by Kuypers (1972).14 Although Kant does not refer to culture in the text of Perpetual Peace, he does mention it in this context

in his Vorarbeiten to the work (VEF 23: 171).

224 PART IV

of agriculture, and the incipient technology and commerce that goes along with thisdevelopment.15 More importantly, human nature is seen as so constituted that warappears to be something noble, to which humans are impelled by the love of glory,quite apart from any selfish motives. This last point suggests that war might also be viewedas playing a role with respect to the culture of discipline; but lest he be thought to beglorifying war, Kant ends this segment of his account by citing the Greek saying that,“War is bad in that it makes more evil people than it takes away” (EF 8: 365).All of this, however, is merely propaedeutic to Kant’s main concern, which is

nature’s contribution to the human race’s collective moral end, particularly withrespect to public right. And here the message is clear: nature will “come to the aid ofthe general will grounded in reason, revered but impotent in practice” (EF 8: 366) byforcing self-seeking individuals to submit themselves to the rule of law. Moreover, thestate that finally emerges from such a process (perhaps after many futile experiments)will bear more of a resemblance to a republic, where freedom is preserved as well aslimited by a system of checks and balances, than to a Hobbesian despotism. It is in thiscontext that we must understand Kant’s famous claim that the problem of organizing astate “is soluble even for [für] a nation of devils (if only they have understanding)” (EF 8:366).16 Kant is here in effect maintaining that the creation of a political order havingthe external form of justice can be viewed strictly as a problem of technically practicalreason, that is to say, a game theoretical problem. As Kant puts it, the attainment ofsuch a condition must be possible, since it requires “only the mechanism of nature, andwhat the task requires us to know is how this can be put to use in human beings inorder so to arrange the conflict of their unpeacable [unfriedlichen] dispositions within apeople that they themselves have to constrain one another to coercive law and so tobring about a condition of peace in which laws have force” (EF 8: 366).Such a “simulacrum of morality” is the most that nature, through the mechanism of

the passions and with the aid of technically practical reason, can force human beings toproduce and to this extent it forces us to be free.17 It cannot make us morally good,

15 See Rousseau (1964). It is clear from a number of texts that Kant was deeply influenced by this essay ofRousseau.

16 The German text reads: “Das Problem der Staatserrichtung ist, so hart wie es auch klingt, selbst fur ein Volk vonTeufeln (wenn sie nur Verstand haben) auflosbar . . . ” In his comments on the original version of this essay, BerndLudwig correctly pointed out that I had erroneously translated “für” by “as,” rather than as “for” therebysuggesting, incorrectly in his view, that the problem of a political order embodying principles of justice couldbe solved by a race (or nation) of devils rather than simply for it. Although I gratefully accept his emendationof the translation, I reject the implications which he derives from it. On my reading, it is crucial that thepassage be taken in both senses, which is perfectly compatible with the text as it stands. In other words, toclaim that the problem is soluble for a hypothetical nation of devils is to claim both that the solution can beapplied to and that that it can be arrived at by such a nation. The former is possible because it does notpresuppose any moral motivation on the part of the subjects; the latter because it requires only an exercise oftechnically practical reason. In affirming only the former, Ludwig ignores the crucial parenthetical clause:“wenn sie nur Verstand haben.” If it were merely a matter of being subject to the laws, this condition would notbe necessary.

17 In the VEF 23: 171, Kant states that nature can compel us to adopt an analogue of moral laws.


since, as a function of inner disposition, this is entirely the work of freedom. Never-theless, by forcing us into legal arrangements at both the intra and international levels, italso serves as an objective end of reason. And since, as Kant also tells us, “a good moraleducation [Bildung] of a people is to be expected from a good constitution” (EF 8: 366),it presumably contributes to the promotion of the highest good.

IVThe basic lesson to be learned from the preceding is that far from being a mere popularor occasional piece, Perpetual Peace stands in essential connection with the third Critiqueand its systematic concern with an Übergang between nature and freedom. Evengranting this, however, there are certain problems peculiar to this work concerningnature’s alleged guarantee. First, the claim that perpetual peace is guaranteed seems to bemore than is required to support a duty to seek such peace; more than Kant elsewheresuggests is required; and more than the account of the third Critique would warrant.Second, following Hegel, it might be argued that rather than supporting a duty claim,such a guarantee undermines any obligation, since we cannot have a duty to work for agoal that we know will come about anyway, independently of our effort.18 Third, thisguarantee of an actual attainment appears to contradict Kant’s usual position thatperpetual peace and a perfectly just civil society are themselves ideals to be approxi-mated but never fully attained. Finally, it seems difficult to reconcile Kant’s claim in thefirst supplement regarding the capabilities of a race of intelligent devils with hisinsistence in the first appendix on the necessity of a moral politician (as distinct fromthe political moralist) and, more generally, on a moral foundation for politics. In theremainder of this essay, I shall focus on these four problems in the hope that this may (atleast indirectly) shed some light on the larger questions concerning the coherence ofKant’s moral teleology.To begin with, it is undeniable that a guarantee of success is more than is required to

secure an obligation. Ought implies can, not shall. Moreover, Kant readily admits asmuch elsewhere. For example, he claims that all that is required to preserve theobligation to work for the moral improvement of the human race is that the futilityof such a project not be wholly certain (TP 8: 309) or, equivalently, that its accom-plishment is not demonstrably impossible (TP 8: 310). Nor, it must be admitted, isthere specific reference to a guarantee in other texts dealing with the general issue,including }83 of the third Critique, where nature’s ultimate purpose with respect tohumanity is specified.A closer consideration of the guarantee, however, suggests that Kant’s claim is not as

audacious as it might appear and that the differences between the account in PerpetualPeace and other texts are more rhetorical than substantive. The so-called guarantee has

18 This is one of the “whole nest of thoughtless contradictions,” which Hegel, using the Kantianexpression, claims to find in Kant’s doctrine of the postulates. See Hegel (1977), 365–83.

226 PART IV

two grounds, only one of which is teleological. The first, non-teleological, ground isthe reflection that the problem of finding a just political order must be soluble, since itdoes not require any moral change but is essentially a matter of technical rationality.This is the point that the reference to the nation of intelligent devils (the counterpart toRousseau’s angels) is intended to underscore. But the fact that the problem is solublefor human reason does not of itself guarantee that it will be solved. Thus, the needfor the second and properly teleological thesis that our nature, that is, our unsocialsociability, will compel the human race to find the solution. In short, the guaranteeconsists in pointing out that both capacity and motivation are present, which issufficient to show that, contrary to the claims of the cynical political moralist, theend is attainable. Moreover, I do not think that Kant either intended or needed toclaim more than this.What is both distinctive and problematical about this otherwise rather unremarkable

line of thought is the claim that this compulsion is to be regarded as part of a “secretplan” of nature or providence rather than a brute, contingent fact. But in consideringthis seemingly whimsical claim it is crucial to keep in mind that it is a matter ofobligatory ends. Put simply, if it is a duty to realize a given end, we cannot regard itmerely as a contingent fact that the enabling conditions of its realization obtain. For thiswould engender the contradiction that it is necessary to pursue an end that might notbe attainable. Thus, we must regard the end as attainable (as really, not merely logicallypossible), and since the attainability depends on nature, it follows that we must think ofnature as providing the enabling conditions.Similar considerations also enable us to deal with the Hegelian objection that the

putative guarantee is actually incompatible with the duty, since we cannot be obligatedto do what will occur anyway without our efforts. Admittedly, some of Kant’s claimsregarding nature, including those cited earlier in connection with the guarantee,suggest the fatalistic picture that fuels this line of objection. As is clear from the context,however, the nature in question is human nature and the required result is not thoughtto obtain independently of all human effort, but merely independently of any morallymotivated effort. Not only is the duty motive inoperative, but the ends themselvesneed not be recognized as morally necessary by those who actualize them. In Kantianterms, this means that the ends are realized through the “mechanism of the humaninclinations” (EF 8: 368) rather than through the exercise of pure practical reason. Butfar from removing, this redefines the obligation, which is to replace what is contin-gently produced by nature with the self-conscious product of pure practical reason.Nature hands the torch, as it were, to autonomous human reason, which then carrieson the task guided by the idea of right.Moreover, in light of the above, the apparent incompatibility between the claim of a

guarantee and Kant’s insistence elsewhere that a perfectly just society and perpetualpeace are ideals to be approximated but never fully attained can be seen as a differenceof emphasis rather than of doctrine. Since Kant (like Fichte after him) was a philoso-pher of infinite tasks, this clearly represents his considered view. Indeed, how else can


one take Kant’s famous remark that “out of such crooked wood as the human being ismade, nothing entirely straight can be fabricated” (IAG 8: 23) But this is compatiblewith the guarantee, if it is taken to refer to progress toward the goal rather than itscomplete attainment. What is actually achievable are republican constitutions andinternational arrangements such as a league of nations; but these are themselves merelyapproximations of the not fully attainable ideals of perfect justice and peace. Accord-ingly, there is no need to assume any real contradiction between Perpetual Peace andKant’s other writings on this score.Finally, we can see that there is no contradiction between Kant’s claim about the

capabilities of a putative nation of intelligent devils and his insistence on the necessity ofa moral politician and a moral foundation for politics.19 These claims simply reflect twodifferent aspects of Kant’s doctrine of juridical duties. On the one hand, they are duties,which, as such, express unconditional demands of reason rather than hypothetical rulesof prudence or expedience. On the other hand, unlike duties of virtue, these demandsabstract from the question of motivation. They require compliance but not intention.Given this conception, it is perfectly conceivable that a nation of intelligent devilscould arrive at forms of political organization that correspond to what duty requireswithout even recognizing such correspondence, much less being motivated by it.By contrast, the moral politician, although perhaps not motivated by the thought of

duty alone (principium executionis), nevertheless appeals to the idea of right as theprinciple of legislation (principium diiudicationis). Thus, with reference to Kant’s distinc-tion between acting merely in accordance with and from duty, one might say that thedifference between the intelligent devil and the moral politician is that the former actsmerely in accordance with the idea of legality (where this accordance is completelyaccidental and unintended), while the latter legislates from (or in light of) this idea.Moreover, if this is so, we can see how the moral politician is able to overcome thesheer contingency of the unintended agreement with the law attainable (at least inprinciple) by intelligent devils. And, even though it falls far short of the ethicalcommunity, this surely counts as progress.20

19 For a critique of Kant along these lines, see Weyand (1964), 150. Yovel (1980), 188–90, offers acompletely amoral reading of Kantian political thought, thereby ignoring Kant’s account of the moralpolitician.

20 I wish to thank Allen Wood for his helpful comments on an earlier version of this essay.

228 PART IV

Essay Fifteen

Kant’s Conception of Aufklärung

Kant’s views on enlightenment are best known through his essay, “An answer to thequestion: What is enlightenment?,” which was published in the Berlinische Monatsschriftin 1784 in response to the question posed in the same journal the previous year by theBerlin theologian Johann Friedrich Zollner.1 This is, however, merely the first and bestknown of a series of reflections on the subject contained in the Kantian corpus. Inaddition to passing references in a number of writings, Kant offers general character-izations of the nature of enlightenment in “What does it mean to orient oneself inthinking?” (1786) and the Critique of the Power of Judgment (1790), and all of these haveto be understood against the backdrop of Kant’s ongoing concern with related topics asdocumented in his Anthropology from a Pragmatic Point of View (1798). Thus, rather thanconfining myself to the famous enlightenment essay, I shall consider these texts as wellin an attempt to provide an overview of the Kantian conception of enlightenment.Unfortunately, this broad focus and limited space makes it impossible to consider

here the important political dimensions of Kant’s enlightenment essay.2 Similarly,I must forgo a detailed discussion of the relation between Kant’s essay and the essaypublished that year on the same topic by Moses Mendelssohn.3 While not completelyneglecting the latter topic, my major concern is to show that Kant had a complex andnuanced conception of enlightenment, one which is closely connected to some of hisdeepest philosophical commitments, and is as distinct from the views of his contem-poraries, including Mendelssohn’s, as the “critical philosophy” is from the rationalismof Leibniz, Wolff, and Baumgarten.

IEnlightenment, for most of Kant’s contemporaries, was seen as an intellectual condi-tion marked by the possession of knowledge, in contrast to an ignorance that was

1 Zollner’s question was: “Ist es rathsam, das Ehebündniss nicht ferner durch die Religion zu sanciren?”Berlinischer Monatschrift 2 (1783), 516. Reprinted by Albrecht and Hinske (1973). For a discussion of this seethe introduction by Albrecht and Hinske (1973), xl–xlii; Altmann (1973), 654ff.; and Schmidt (1996), 1–6.

2 For discussions of the political context and implications of Kant’s essay see Beyerhaus (1921), 1–15;Schulz (1974), 60–80; and Albrecht and Hinske (1973), xlii–xlv.

3 Mendelssohn (1973), 444–51. For discussions of its relations to Kant’s essay, see Altmann (1973), 661–6;Schulz (1974); Schmidt (1989), 269–91 and 30, (1992), 77–101.

viewed as the breeding ground for superstition. In this spirit, Mendelssohn explicitlydistinguished between enlightenment and culture, regarding each as a partial expres-sion of an encompassing notion of education [Bildung], which includes everythingrelated to the vocation of man [Bestimmung des Menschen]. Enlightenment concerns thetheoretical side of this education and culture the practical.4 It consists in the rationalknowledge of basic truths regarding the human vocation. Insofar as these truths serve toground a rational belief in divine providence, Mendelssohnian enlightenment has beenaptly described by Alexander Altmann as a “consoling enlightenment” [trostvol Aufklärung].5

The great gulf separating Kant’s conception from any such view is already clear fromthe famous opening lines of his essay:

Enlightenment is the human being’s emergence from his self-incurred tutelage [Unmündigkeit].6

Tutelage is the inability to make use of one’s understanding without direction from another. Thistutelage is self-incurred when its cause lies not in lack of reason but in lack of resolution andcourage to use it without direction from another. Sapere aude! Have courage to make use of yourown understanding! is thus the motto of the Enlightenment.7

Kant here makes four main points: (1) Enlightenment is to be understood negatively asthe liberation or escape from a condition rather than positively as the possession orcomprehension of a body of knowledge. (2) The condition from which one is to beliberated is not simply ignorance (as in traditional views) but Unmündigkeit (usuallytranslated as “tutelage,” “immaturity,” or “incompetence”), understood as an incapac-ity to use one’s understanding without the direction of another. (3) This condition maybe self-incurred [selbstverschuldet], that is, something for which the individual is in someway responsible, rather than the result of a natural incapacity or of being deceived byothers (e.g., priests and despots). (4) Consequently, what is required to escape thiscondition is not intellectual development but an act of will, specifically a resolution tothink for oneself.Clearly, the central notion here isUnmündigkeit, which refers both to a natural minority

of age and a “legal or civil immaturity.”The latter encompasses not only children,who are“naturally immature,” but also women of any age, since they are not deemed fit torepresent themselves in legal proceedings and must therefore be represented by guardians[Vormünder].8 Kant obviously has this legal sense in mind in the enlightenment essay; butin his most detailed treatment of Unmündigkeit in the Anthropology he discusses it inthe context of a consideration of mental deficiencies.9 Significantly, in contrast to othermental deficiencies, such as lack of wit or judgment, Unmündigkeit concerns not the

4 Mendelssohn (1973), 444–51.5 Altmann (1982), 276–85.6 Following Lewis White Beck, I am here translating “Unmündigkeit” as “tutelage” rather than “minori-

ty,”which is the term that Gregor uses. But given the multiplicity of the senses of this term (discussed below),I shall henceforth leave it untranslated.

7 BA 8: 35.8 See Schmidt (1996), 89.9 A 7: 208–9.

230 PART IV

understanding itself, but its exercise. Thus one could have a perfectly sound understandingand still be unmündig, if one is not capable of using it properly. In an amusing andinstructive example, illustrative of a form of Unmündigkeit that is supposedly compatiblewith a sound understanding, Kant cites an anecdote regarding a scholar, buried in hisbooks,who, in response to a servant’s warning that the housewas on fire, answered: “Youknow, things of that sort are my wife’s affair.”10

This anecdote is instructive because it provides an instance of an Unmündigkeit that isnot merely compatible with a sound understanding but also self-incurred, a claim forwhich Kant was criticized by Hamann and more recently by Rudiger Bittner. In aletter of December 1784 to Christian Jacob Kraus devoted more or less to a discussionof Kant’s essay, Hamann remarks that, though he can tolerate the concept of enlight-enment being elucidated “more aesthetically than dialectically” through the analogywith immaturity and guardianship, he cannot accept the “accursed adjective ‘self-incurred,’” which he characterizes as the “proton psuedos.”11 Similarly, Bittner dismissesthe thesis that an inability to use one’s own understanding might be one’s own fault.12

But pace Hamann and Bittner, there is nothing problematic in the suggestion thatone might voluntarily put oneself in a condition that results in a significant lessening ofone’s capacity to think for oneself. The familiar contemporary phenomenon of cults is acase in point. Similarly, Kant is correct in claiming in that “[T]o make oneself immature[sich selbst unmündig zu machen], degrading as it may be, is very comfortable.”13 Indeed,though I cannot develop the theme here, one might read Kant’s account of how thepeople relish theirUnmündigkeit, gladly surrendering the right to think for themselves totheir spiritual guardians (the clergy), as an anticipation of Dostoevsky’s analysis of theflight from freedom in the “Legend of the Grand Inquisitor.”The reverse side of this relish for the comforts of Unmündigkeit, one is almost

tempted to say for the consolations of unenlightenment, is a fear of the burdens ofenlightenment. This fear goes together with a certain laziness and is inculcated in the“great unthinking masses” [gedankenlosen grossen Haufens] by their guardians, who areable to use it to their advantage.14 Moreover, Kant notes that this state of affairs makesit extremely difficult for an individual to work himself out of the condition ofUnmündigkeit, once it has become virtually a second nature. A more realistic possibilityis that the public (by which Kant means the educated, reading public, and not the greatunthinking masses15) may enlighten itself. In fact, though he insists that it willnecessarily be a slow and gradual process, Kant claims that such enlightenment isvirtually certain to follow, if freedom of expression be granted.16

10 Ibid., 210. See also R 15: 528–9, 229–31; and AF 25: 541.11 Hamann (1965), 289. An annotated English translation of the letter is contained in Schmidt (1996),

145–53.12 Bittner (1996), 346. 13 A 7: 209. 14 AA 8: 36.15 See Davis (1992), esp. 171–4. 16 BA 8: 36.

KANT’S CONCEPTION OF AUFKLARUNG 231

The main reason why the change must be a gradual one is that it concerns a way ofthinking or Denkungsart rather than the content of what is thought.17 Specifically, it isthe change from a Denkungsart that leaves the determination of truth to others (theguardians) to one that refuses to accept truth on the authority of others, but insteadinsists on thinking for oneself, on subjecting all claims of truth to the test of one’s ownreason. So construed, enlightenment transcends the theory–practice separation em-phasized by Mendelssohn and becomes a species of autonomy. Moreover, in sharpcontrast to the latter’s religious-teleological view, Kant characterizes every man’svocation [Beruf] simply as thinking for himself [selbst zu denken].18

IIKant’s next substantive discussion of the nature of enlightenment is in a note appendedto the close of the essay, “What does it mean to orient oneself in thinking?” This essay,which contains Kant’s response to the famous “Pantheismusstreit” betweenMendelssohnand Jacobi, appeared in 1786, some two years after the enlightenment essay. The disputebegan as a somewhat parochial controversy regarding Lessing’s Spinozism, but soonbroadened into a full-scale debate concerning the relative standings of faith and reason.Kant’s concern is to offer his brand of critical rationalism as an alternative to both theirrationalism of Jacobi and the dogmatic rationalism of Mendelssohn. Clearly, theformer is Kant’s main target, but the latter does not escape unscathed. In fact, Kantsuggests that Mendelssohn’s procedure of “arguing dogmatically with pure reason in thefield of the supersensible is the direct path to philosophical enthusiasm, and that only acritique of this same faculty of reason can fundamentally remedy this ill.”19 Thus, Kantnot only rejects Mendelssohn’s program for enlightenment, but in effect treats thelatter’s dogmatic rationalism as a species of anti-enlightenment. Although the essay itselfis of considerable importance, I shall here discuss only the note, which I cite in full:

Thinking for oneself means seeking the supreme touchstone of truth in oneself, (i.e. in one’s ownreason); and the maxim of always thinking for oneself is enlightenment. Now there is less to thisthen people imagine when they place enlightenment in the acquisition of information [Kennt-nisse], for it is rather a negative principle in the use of one’s faculty of cognition, and often he whois richest in information is least enlightened in the use he makes of it. To make use of one’s ownreason means no more than to ask oneself, whenever one is supposed to assume something,whether one could find it feasible to make the ground or the rule on which one assumes it into auniversal principle of the use of reason. This test is one that everyone can apply to himself; andwith this examination he will see superstition and enthusiasm disappear, even if he falls far shortof having the information to refute them on objective grounds. For he is using merely the maximof reason’s self-preservation. Thus it is quite easy to ground enlightenment in the individual subjectsthrough their education.20

17 BA 8: 36. 18 Ibid. 19 WHD 8: 138n. 20 Ibid., 8: 146.

232 PART IV

Since a maxim is something that one adopts or chooses for oneself, by characterizingenlightenment as a maxim of thinking for oneself, Kant makes explicit its connectionwith a resolution or act of will. Correlatively, the failure to adopt a maxim is something forwhich one may be held responsible. Kant is also explicit in rejecting the alternative(Mendelssohnian) view, which identifies enlightenment with the possession of knowledge.This rejection is grounded in the denial of the possibility of the attainment of the kind ofknowledge inwhich it supposedly consists. Instead, enlightenment is once again understoodin fundamentally negative terms and, as such, is virtually identified with criticism.Kant goes significantly beyond his earlier characterization, however, in providing an

account of just what is involved in thinking for oneself or using one’s own reason.Basically, one is to ask oneself whether the ground of one’s assumption can be regardedas suitable for all cognizers. This amounts to an appeal to a cognitive version of theprinciple of the universality of reasons, that is, the idea that if something justifies mybelief, it must also justify the belief of any other rational being under similar conditions.The suggestion is that this principle, of itself, without any need for a high degree ofknowledge, is sufficient to avoid the twin evils of fanaticism (Jacobi and possibly evenMendelssohn) and superstition.At first glance at least, the final major claim in the note appears to conflict with Kant’s

earlier account. There, it will be recalled, Kant distinguished between the attainment ofenlightenment by an individual and by the public. The former was judged to be verydifficult but the latter quite possible, assuming the existence of sufficient freedom ofexpression.Here, by contrast, the claim is that it is the enlightenment of the individual thatis easy, while that of an entire age is a very difficult and prolonged task.Nevertheless, a closer consideration of the two texts suggests that the difference is

more apparent than real, reflecting the distinct emphases of the two essays rather thanany fundamental change of mind on Kant’s part. The emphasis in “What is Enlighten-ment?” is on the difficulty of a single individual leaving a state of Unmündigkeit andbecoming enlightened. The claim is that once one is in this condition it becomesvirtually impossible to work one’s way out of it.By contrast, in “What does it mean to orient oneself in thinking?” Kant assumes an

individual who is not yet corrupted, and argues that for such a being enlightenment iseasily attained, given a proper education. Moreover, the public of the earlier essayshould not be identified with that of the later one. As already noted, by “the public”[das Publicum] Kant understands a certain segment of society (specifically, those who areliterate and have attained a certain level of education) rather than all of society. But theenlightenment of an entire age would require the liberation of all members of society,presumably including the “great unthinking masses,” of whom Kant is so contemptu-ous. Accordingly, on this score as well there is no real conflict between the two texts.On the contrary, in the earlier essay Kant explicitly distinguished between an age ofenlightenment and an enlightened age and denied that his was the latter.21

21 AA 8: 40.


IIIKant’s brief discussion of the nature of enlightenment in the third Critique occurswithin the context of his account of taste as a form of sensus communis (}40). This iscontrasted with the common human understanding [gemeinen Menschenverstandes],which is said to be guided by three normative principles or maxims: (1) to think foroneself; (2) to think from the standpoint of everyone else; (3) to think alwaysconsistently.22 In the Anthropology and the Jäsche Logic Kant cites the same threemaxims, characterizing them positively in the former work as rules for the pursuit ofwisdom, and negatively in the latter as universal rules and conditions for avoidingerror.23

Not surprisingly, it is the first of these maxims, also described as the maxim both ofan unprejudiced reason and of one that is never passive, that is directly connected withenlightenment. This connection is mediated by the identification of a passive, i.e.heteronomous, reason, with prejudice, and the characterization of superstition asprejudice in the preeminent sense [in sensu eminenti]. The latter, in turn, is said toconsist in thinking of nature as not subject to the rules which the understanding laysdown as the ground of nature.24 In other words, superstition is understood as the failureto recognize the governance of nature by the transcendental laws of the understanding.Presumably, Kant does not mean by this that superstition consists in the failure torecognize that these laws have their seat in the human understanding, since that wouldmake everyone who does not acknowledge the results of the Transcendental Analyticsuperstitious. The point is rather that it consists in the failure to recognize thenormativity of such laws with regard to explanation and the consequent openness tonon-natural modes of explanation, i.e. miracles.25 Nevertheless, since Kant does holdthat the (transcendental) laws of nature are, in fact, the very principles of the under-standing, it follows that this openness results from a failure to use one’s understanding,which is perhaps why the maxim of thinking for oneself is later identified with themaxim of the understanding.26

Given this, Kant defines enlightenment as “liberation [Befreiung] from superstition,”while noting in passing that it can also be regarded as liberation from prejudicegenerally.27 Although the targeting of superstition as the enemy is a commonplace ofenlightenment thought, Kant’s analysis of it reflects his underlying conception of a self-incurredUnmündigkeit. The claim is that superstition creates a kind of mental blindness,which the afflicted person regards as virtually obligatory, and which reveals in anespecially clear way the need to be guided by others and therefore the presence of apassive or heteronomous reason.28 In other words, while prejudice in general already

22 KU 5: 294. 23 A 7: 200 and 228–9; and JL 9: 57. 24 KU 5: 294.25 This is clear from the similar characterization of superstition that Kant offers in The Conflict of the

Faculties, where it is described as “the tendency to put greater trust in what is supposed to be non-natural thanin what can be explained by laws of nature, whether in physical or in moral matters” (SF 7: 65n).

26 KU 5: 295. 27 Ibid., 294. 28 Ibid., 294–5.

234 PART IV

constitutes a condition of passive or heteronomous reason, superstition provides anillusory justification for remaining in that state. Indeed, as the connection withobligation suggests, it might be described as a kind of principled Unmündigkeit.But whether it be understood in connection with prejudice in general or supersti-

tion in particular, enlightenment remains for Kant an essentially negative intellectualvirtue. This limitation of the scope of enlightenment is already implicit in the juxtapo-sition of the maxim of enlightenment, thinking for oneself, with the other two maximsof thinking from a universal standpoint and of consistent thinking. It becomes fullyexplicit, however, in a note attached to the discussion, in which, while commentingagain on the difficulty of becoming enlightened, particularly for the public, Kant insiststhat “genuine enlightenment” consists in a merely negative Denkungsart.29

Enlightenment, so construed, can hardly be equated with the full perfection of thehuman intellect, as it was for Mendelssohn and many of Kant’s other contemporaries.Nevertheless, it remains the supreme negative intellectual virtue, since it is the conditiosine qua non of any authentic use of the understanding or reason. For without freedomfrom prejudice, how could one even begin to adopt a universal standpoint? Andthough a passive or heteronomous reason might think consistently, it would amountto nothing more than a consistent prejudice. Thus, it is no accident that Kant presents itas the first, though not the only, maxim of the common human understanding.

IVFinally, though it is impossible to pursue the issue in any depth, it must at least be notedthat Kant’s insistence on the necessity for all three maxims contains the basis for aresponse to critics such as Gadamer, who see in enlightenment a merely abstractnegativity, “a prejudice against prejudice itself, which denies tradition its power.”30

Such criticisms apply to Kant’s conception, only if the maxim of enlightenment isviewed in isolation. In particular, the maxim to think from the standpoint of everyoneelse makes it clear that thinking for oneself does not mean simply ignoring the opinionsand values of others (which could include “tradition” in Gadamer’s sense), but ratherinvolves an endeavor to see things from the point of view of the other as well.Similarly, the maxim of consistent thinking requires not merely logical consistency,but also the integration of the standpoints of self and other into a coherent whole,which is far removed from an abstract negativity. Indeed, as Kant himself suggests in theAnthropology, in explicit allusion to the enlightenment essay, the adoption of all threemaxims is required for “the most important revolution within the human being . . . ‘hisexit from his self-incurred Unmündigkeit.’”31 And not withstanding the contemporarycritics of the “enlightenment project,” such a revolution still seems a worthy goal.

29 Ibid., 294n. 30 Gadamer (1999), 270. 31 A 7: 229.


Essay Sixteen

Teleology and History in Kant:The Critical Foundations ofKant’s Philosophy of History

Although the title of Kant’s essay, “The Idea of Universal History with a CosmopolitanAim” indicates its central theme, it reveals little or nothing about its underlyingmethodology and its connection with the emerging “critical” philosophy. Indeed, asfar as the title is concerned, the only hint of a connection with the latter is provided bythe inclusion of the term “idea.” This is a technical term for Kant referring to conceptsof reason, which, as distinct from concepts of the understanding, whose legitimate useis restricted to possible experience, involve the thought of an absolute totality orcompleteness that can never be met with in a possible experience and is therefore“transcendent” with respect to the latter. In the first Critique, Kant appealed to thePlatonic republic and a constitution that provides for “the greatest human freedomaccording to laws that permit the freedom of each to exist together with that ofothers” as examples of such ideas (KrV A316/B372–3); but his focus was on the“transcendental ideas” (the soul, the world, and God), which arise from extendingcertain concepts of the understanding to the “unconditioned,” thereby producing thethought of a complete systematic unity. While illusory, in the sense that no real objectcorresponds to them, these ideas nonetheless play an essential regulative role in guidingthe understanding in its endemic search for unity in experience.

“Idea” as it appears in the title of our essay, is clearly not to be understood accordingto the model of the transcendental ideas, since it refers to human history rather than toany illusory transcendent entity. It is closer to the two political ideas noted above, but itdiffers from them in that they are practical ideas, which function as norms or ideal types,whereas the idea of a universal history is theoretical, characterizing a way in which aphilosopher might conceive this history in the endeavor to attain a synoptic compre-hension of it. What elevates this conception to the status of an idea is that it involves thethought of completeness or absolute totality in two senses: first, it is concerned withhumankind as a whole rather than a particular segment thereof (this is what makes it a“universal history”); second, it conceives this history as a totality, encompassing allgenerations and embodying an underlying telos.

The title further indicates that the pattern or purpose that underlies and regulatesphilosophical reflection on history is a political one, namely, a cosmopolitan state ofaffairs, by which Kant understands not a world state, which would be the culminationof tyranny and the end of all freedom, but a confederation of free states or league ofnations, which would provide the condition under which humankind’s greatestscourge, war and the constant threat thereof, could be permanently abated. As such,this essay contains the first statement of a view that Kant was to work out more fully insubsequent years, culminating in Perpetual Peace (1795).The main focus of this essay, however, is on the underlying methodology and

connection with the “critical” philosophy of Kant’s philosophy of history rather thanon the particular view of history that it contains. Inasmuch as Kant’s approach tohistory is explicitly teleological, this requires a consideration of the central themes oftheCritique of the Teleological Power of Judgment, which is the second part of theCritique ofthe Power of Judgment, commonly referred to as the third Critique. Even though the latterwork was published six years after “Idea of a Universal History,” it provides the lensthrough which the earlier work must be examined. But before turning to this, it isnecessary to consider briefly the prefatory portion of the essay, where Kant defines theproblem for which teleological reflection provides the solution.The above is the subject matter of the first part of this essay, which is divided into

seven parts. The second provides an introduction to the two central conceptions of thethird Critique as a whole: the purposiveness of nature and the reflective power ofjudgment. The third deals with Kant’s philosophy of biology. Although not directlygermane to his account of history, Kant’s controversial thesis that the conception of anintrinsic purposiveness is an ineliminable condition of our understanding of organismsprovides the framework in which Kant’s whole approach to teleology must be viewed.The fourth examines Kant’s attempt to extend purposiveness from particular organismsto the relation between organic beings (including human beings) and the order ofnature as a whole. The fifth considers Kant’s claim that the conception of nature as ateleological system requires the assumption that something serves as the ultimate end ofthis system and that this can only be humankind.1 The sixth analyzes the connectionbetween the conception of humankind as the ultimate end of nature and Kant’steleological account of history. Finally, by appealing to the distinction between anultimate end of nature [letzter Zweck] and a final end of creation [Endzweck] and theaccount of an “ethical community” that Kant introduces in Religion within the Bound-aries of Mere Reason, the seventh part explores the complex but related question of theconnections between Kant’s politically oriented historical teleology, his moral theory,and his trans-political yet social view of the ultimate goal of history.

1 In this essay I generally use the term “humankind” to render “der Mensch” and its plural form (“dieMenschen”) rather than the literal translation “the human being” or “human beings,” since in both this essayand the third Critique Kant is usually referring to the species as a whole.

TELEOLOGY AND HISTORY IN KANT 237

IKant begins the prefatory portion of “Idea of a Universal History” by posing thefreedom–nature problem with respect to human history. In the resolution of the ThirdAntinomy in the firstCritique, Kant had argued merely that if one adopts the standpointof transcendental idealism, natural causality and freedom (in the transcendental sense)need not contradict each other. Now, abstracting from the transcendental question offreedom, Kant notes that as appearances, that is, as empirically accessible events, humanactions are law-governed in the same manner as any other natural occurrences. And, asillustration of this thesis, Kant points to statistical tables, which reveal a predictable rateof marriages, births, and deaths, given a sufficiently large sampling. The significance ofthis for Kant consists in the fact that, despite the assumption of human freedom, itopens up the possibility of depicting a pattern, indeed a progress, in human affairsthrough the development of humankind’s “original predispositions” (IAG 8: 17).2

Nevertheless, though necessary, such statistical regularity is hardly sufficient to indicatea pattern, much less a progressive development in human affairs writ large. There remaintwo obstacles to the project of finding such a pattern: first, human beings, unlike animals,do not behave instinctively, which introduces a certain unpredictability that is not foundin the animal kingdom; second, despite their presumably rational nature, human beings,considered as a whole, do not act upon any agreed plan. On the contrary, Kant lamentsthat a survey of human affairs indicates that, while wisdom may occasionally manifestitself in particular cases, “everything in the large is woven together out of folly, childishvanity, often also out of childish malice; so that in the end one does not know what tomake of our species, with its smug imaginings about its excellences” (IAG 8: 18).At this point, Kant reflects that there remains only one possible way for the

philosopher to uncover a rational plan in human history, namely, to investigatewhether one may discover an aim of nature (rather than a consciously adopted aim ofhuman beings) “in this nonsensical course of things human” (IAG 8: 18). In otherwords, at issue is whether it is possible to attribute a purpose to nature with respect tohumankind, which the latter has not consciously chosen for itself. And since thepurpose that Kant has in mind is a political one, namely, the greatest possible freedomof each under law that is compatible with the freedom of all, this comes down to thequestion of whether it is possible to consider nature as forcing us to be free. Althoughthe precise expression is not used by Kant, this procedure has been aptly termed the“cunning of nature,” which alludes to Hegel’s famous “cunning of reason” [List derVernunft], which plays a similar role in the latter’s philosophy of history.3

2 A predisposition [Anlage] is a feature of the nature of an organism that accounts for its developing incertain determinate ways. For a useful discussion of this conception and its relevance to both Kant’sphilosophy of history and his moral theory, see Wood (1999), esp. 118–22, 210–12.

3 Extensive use of this expression has been made by Yovel (1980), esp. 125–7. Yovel notes (140n) that theexpression was also used by Weil (1963). It should also be noted that Kant himself uses virtually the sameexpression, referring to the “Kunstanstalten der Natur” [artifices of nature] (EF 8: 362; 332).

238 PART IV

This brings to the fore the problem of teleology, which Kant introduces in the firstproposition of the essay. It states simply that, “All natural predispositions of a creatureare determined sometime to develop themselves completely and purposively” (IAG 8:18). And in justification of this thesis Kant further remarks:

With all animals, external as well as internal or analytical observation confirms this. An organ that is notto be used, an arrangement that does not attain its end, is a contradiction in the teleological doctrine ofnature. For ifwe depart from that principle, thenwe no longer have a lawful nature, but a purposelesslyplaying nature; and desolate chance takes the place of the guideline of reason. (IAG 8: 18)

The inclusion of the qualifier “sometime” [einmal] in the statement of the propositionforeshadows Kant’s subsequent claim that, in the case of humankind, the completedevelopment of the predispositions that involve the use of reason will require anindeterminately lengthy historical process because reason cannot develop fully withinthe lifetime of any individual, but only gradually in the species as a whole. Indeed, thisappears to be the main reason why humankind for Kant has a history.For present purposes, however, the main point is that the teleological picture of

nature that Kant sketches here is far removed from the conception of nature provided inthe firstCritique. From the point of view of that work, nature is conceived as the totalityof appearances standing under laws, where the laws are of the mechanistic variety,ultimately grounded in the Principles of the Pure Understanding. In short, the nature ofthe first Critique (at least that of the Transcendental Analytic) is essentially a Newtoniannature that appears to have no room for anything like a “teleological doctrine ofnature.” Accordingly, if an appeal to teleology is to be legitimated and made the basisfor an account of human history, Kant must go beyond what he said in the firstCritique.Moreover, the fact that he fails to do so in this essay and instead simply offers ateleological account in a seemingly dogmatic manner has led some to believe that thisessay as well as some of Kant’s other seemingly whimsical forays into the philosophy ofhistory are either regressions to a pre-critical standpoint or purely occasional pieces thatneed not be taken seriously.4 As already noted, however, it is not that Kant failed toprovide a critical foundation for his speculations about the purposiveness of nature andits hidden purposes regarding humankind; it is rather that he only did so retrospectivelyin the thirdCritique. Thus, I shall turn to this work, especially its second part, theCritiqueof the Teleological Power of Judgment, in an endeavor to analyze this grounding.

IIAs a first step, however, it is necessary to say a word about the two fundamental andclosely related conceptions that underlie the Critique of the Power of Judgment as a whole:

4 For a discussion of this issue and defense of the philosophical significance and systematic place of Kant’swritings on history see Fackenheim (1956–7), 381–98; Beck (1963), vi–xxiii; Weyand (1964), 7–21; andDespland (1973), 1–14.


the reflective power of judgment and the purposiveness [Zweckmässigkeit] of nature. Inthe two versions of his Introduction to this work, Kant points out that judgment,together with the understanding and reason, is one of the three “higher” cognitivefaculties (the “lower” faculty being sensibility), but up to this point it is the only onethat had not received a Critique.5 The reason for this is that, unlike the other two highercognitive faculties, judgment does not appear to have any a priori principle of its own,which would either require or warrant a separate Critique. On the contrary, judgment,as characterized in the first Critique, is a faculty of subsumption, whose function is todetermine whether given particulars fall under a rule. And since these rules are allprovided by the understanding and since (on pain of an infinite regress) there can be norules for subsuming under rules, there can be no distinct principle for judgment, whichprecludes providing it with a critique.6

Although Kant retains the conception of judgment as a faculty of subsumption in thethird Critique, he creates the conceptual space for assigning it an a priori principle, bydistinguishing between two roles that it might play with respect to subsumption:determination and reflection. The difference turns on the question of whether therule, which includes concepts, laws, and principles, is given or whether what is given ismerely some particular content that is to be subsumed under a sought for rule. In theformer case, the function of judgment is determinative; in the latter it is reflective.Inasmuch as the task of the latter power is to seek the rule under which givenparticulars may be subsumed, it, unlike the former, does require an a priori principleof its own, which Kant identifies with the purposiveness of nature.For Kant, to term something purposive [zweckmässig] is to say that it is designed or

produced according to the idea of some plan or end [Zweck]. Correlatively, purposive-ness [Zweckmässigkeit] is the quality something has of appearing purposive, that is,appearing to be designed. There are, however, two quite distinct ways in which thismay be understood. It can mean either as if designed for the sake of our cognitivecapacities, or as if designed with respect to something’s own inner possibility. Thisunderlies the division of the third Critique into a critique of the aesthetic and of theteleological powers of judgment.Although our concern here is almost entirely with the latter, it may be useful to say a

word about the former, since it will enable us to appreciate why Kant’s definitiveaccount of teleology ended up in a Critique of the Power of Judgment. Even the mostcursory discussion of the former, however, is immediately confronted with the problemthat it itself comes in two radically distinct forms, which Kant suggests belong togetherwithout ever really explaining how they do. The first, which is treated only in the twoversions of the Introduction, is called “logical” or “formal purposiveness” and refers to

5 Contrary to what the title suggests, Kant states that in the first Critique it was the a priori status ofprinciples of the understanding that was established and in the second Critique that of reason (understood aspractical reason).

6 See Kant, KrV A132–34/B171–74.

240 PART IV

nature’s conformity at the empirical level to the requirements of the understanding foran orderly experience that goes beyond the orderliness supposedly guaranteed by theTranscendental Analytic in the first Critique.7 Simply put, nature is regarded as doing usan epistemological favor by making possible both its taxonomical ordering in terms of acoherent set of concepts and its nomological ordering in terms of a set of empirical lawsthat allow for the construction of overarching theories. Kant does not claim that natureactually does us such a favor, or even that we must believe that it does. The point israther that wemust proceed in our investigation of nature on the assumption that it doesbecause this is the only way in which the reflective power of judgment can coherentlyproceed. Accordingly, the prescriptive force of this principle is directed back at thereflective power of judgment itself, dictating how one ought to judge, as opposed tospecifying the nature of what is judged about.By contrast, inasmuch as they are non-cognitive, aesthetic judgments ormore precisely

judgments of taste, which Kant also assigns to the reflective power of judgment, rely onnature “favoring” us in a quite different way. He explicates this by introducing theconception of a “subjective purposiveness,” which refers to the “form” of an object asit is given in sensory intuition and which manifests itself by occasioning a harmony of thefaculties of the understanding and imagination in “mere reflection,” that is, in a reflectiondirected toward enjoyment in the contemplation of an object rather than cognition. Froma systematic standpoint, this seems to have been the most important function of thereflective of power of judgment for Kant, since he acknowledges that it was his belatedrecognition that judgments of taste rest upon an a priori principle that led him towrite thethird Critique, which was initially conceived as a critique of taste.8

In the case of teleological judgments, purposiveness does not involve the idea ofnature being designed for us, but serves instead as a condition of our understanding thepossibility of a natural object. The inclusion of this in a Critique of the Power of Judgmentstems from the fact that this understanding, which consists in finding a unifyingprinciple, is likewise the work of the reflective power of judgment and, as such, amatter of judgment legislating to itself rather than to nature. Once again, however, thesituation is further complicated by the fact that Kant distinguishes between two formsof such purposiveness: intrinsic and relative or extrinsic.9 The first of these concernsreflection upon particular organic beings and constitutes the essence of Kant’s philoso-phy of biology; the second concerns the relations between these beings and nature as awhole viewed as a teleological system. Since the latter leads to the application of

7 Kant uses the term “logical purposiveness” in the first Introduction (EE 20: 217–20) and “formalpurposiveness” in the published version (KU 5: 181–6). I discuss this conception of purposiveness and itsepistemological function in Allison (2001b) and (2003), which are included in this volume.

8 See Kant’s letter to Reinhold of December 28–31, 1787, where he first announces his plan to publisha “Critique of Taste” (Br 10: 513–15). For my discussion of this, see Allison (2001a), 3–6.

9 Kant appears to treat “relative” and “external” purposiveness as synonyms, using them interchangeably incontrast to inner purposiveness. For texts bearing on this distinction, see EE 20: 249–50; KU: 226–7, 366–9,377–81, 425–9. To avoid redundancy, I shall henceforth refer to such purposiveness as extrinsic.


teleology to history it is our main concern. But since that rests upon Kant’s account oforganisms, I shall attempt to provide a brief discussion of that topic.

IIIThe fundamental concept in Kant’s philosophy of biology is that of an end or purposeof nature [Naturzweck]. He tells us by way of a provisional characterization rather than astrict definition, that “a thing exists as a natural end if it is cause and effect of itself(although in a twofold sense)” (KU 5: 370; 243). Kant assigns a central role to thisseemingly paradoxical concept because the subject matter of biology is the livingorganism and he thought that the latter could be understood only in light of it. It isnot that Kant viewed this either as an empirical concept derived from the observationof organisms or as equivalent to the concept of the latter, which would make the claimthat an organism is an end or purpose of nature analytic. It is rather a construct onKant’s part, which he claimed to be required for the experience and comprehension ofan organism as such.10 In fact, it was the perceived need to introduce this concept intobiology that made a critique of the teleological power of judgment necessary in the firstplace for Kant. This critique has two major tasks: (1) to expound this concept anddemonstrate its indispensability for biology (the main task of the Analytic) and (2) toestablish its coherence and its reconcilability with the mechanism of nature (thefunction of the Dialectic). Due to constraints of space, I shall here concern myselfalmost entirely with the former.11

The basic problem is that the two conditions that must be met by anything that is tocount as an end of nature appear incompatible; for insofar as we regard something as anend, we assign it to an intelligent cause rather than to the mechanism of nature, whileinsofar as we regard it as a product of nature, we seemingly rule out any appeal to such acause. Moreover, to refer to such a thing as both cause and effect of itself seems merelyto restate the problem in other terms. Accordingly, my goals in this section are to try tounderstand why Kant thought that the introduction of this concept was required;illustrate the role that it supposedly plays in biological explanation; and examine thestatus that Kant assigns to it.To begin with, Kant is insistent that all genuine explanation in natural science is

mechanistic. Although there is some controversy about what this means, I take thebasic idea to be that the explanation of all physical wholes must be in terms of the causaland reciprocal interaction of their parts.12 But, according to Kant, living organismscannot be understood in this way, since the function and interaction of their parts canbe understood only in relation to the whole. In short, in the case of organic beings, the

10 See McLaughlin (1990), esp. 46.11 I analyze this antinomy in Allison (1991a), which is included in this volume.12 See McLaughlin (2003) and Allison (1991a).

242 PART IV

whole is (conceptually) prior to the parts. That is why Kant notoriously denied thatthere could ever be a Newton of a blade of grass.13

Kant also thought that the only way in which we can provide a causal explanation ofsomething wherein the whole may be regarded as the causal ground of its parts is if weattribute causality to the conception of the whole. As an example of this, Kant appeals toa geometrical figure (a regular hexagon) that is found sketched in the sand in apresumably uninhabited land. The point is that anyone happening upon such a figurewould not be able to conceive it as the product of purely natural forces, say theinteraction of sand, sea, and wind, or perhaps as the footprint of some animal. It notthat such a mode of production would be physically impossible in the sense that itwould violate known laws of nature; it is rather that the chance of it being broughtabout in such manner would be so remote as to defy explanation. As Kant puts it, “thecontingency of coinciding with such a concept [that of a regular hexagon], which ispossible only in reason, would seem to him [the observer of the figure] so infinitelygreat that it would just as good as if there were no natural law of nature, consequentlyno cause in nature acting merely mechanically . . . ” (KU 5: 370; 242).14 Accordingly,such a figure can be comprehended only by regarding it as the product of an intelligentcause, that is, as an end.It was fairly common in the eighteenth century, especially among theologically

inclined Newtonians, to regard nature as a work of art in this sense. Moreover, this ledto the widespread metaphor of nature as a machine, typically a clock. And since organicbeings appear to exhibit a systematic structure in which the various parts cooperate toform and preserve a living organism, it was likewise natural to consider organisms asmachines designed by a supernatural intelligence.Nevertheless, the distinctive feature of Kant’s position is that he not only rejected

the reductive attempt to explain the nature and behavior of organic beings in purelymechanistic terms, he also repudiated the alternative account of such beings asmachines and, therefore, as products of (divine) art or, as he put it, as “an analogueof art” (KU 5: 374; 246). This is because he thought that organisms have certainepigenetic properties that cannot be attributed to machines such as clocks. As he notes,one part of a clock produces the motion of another and was placed there for the sake ofthe other, but it is not the efficient cause of the production of the other. This does notapply to organisms, however, since in their case the parts have not simply motivepower, but a self-propagating formative power (KU 5: 374; 246). The basic idea is thatorganisms are not merely systems but self-regulating ones, in which the various partshave a capacity to replace and repair one another, which is something that Kant deniedcould be said of either a machine or a product of art.As such, an organismmay be viewed as cause and effect of itself, in a two-fold sense; and

this gives content to the concept of a natural end, which apart from the existence of

13 See Kant, KU 5: 400.14 At one point Kant defines purposiveness as “the lawfulness of the contingent as such” (EE 20: 217).


organisms would be problematic. Kant illustrates this by means of the example of a tree,which, in contrast to a machine, has a three-fold epigenetic capacity. First, a tree generatesanother tree and, therefore, is cause of itself qua species. Second, it generates itself quaindividual through growth by means of the assimilation of nutrients. And, finally, it has acapacity to replace certain parts, if they are damaged or destroyed (KU 5: 371–2).Kant claims that this applies to every organic being, which in virtue of these

capacities and, more generally, its organization, must be regarded as an end of natureand therefore assigned an intrinsic purposiveness, quite apart from any further ends itmight serve. Indeed, he goes further and suggests that this principle applies not only toevery organic being considered as a whole, but to every part of such a being.15 And thisleads Kant to the formulation of a principle, which also serves as the definition of anorganism: “An organized product of nature is that in which everything is an endand reciprocally a means as well. Nothing in it is in vain, purposeless, or to beascribed to a blind mechanism of nature” (KU 5: 376). The previously noted firstproposition of the “Idea for a Universal History,” namely, that “All natural predis-positions of a creature are determined sometime to develop themselves completely andpurposively,” may be seen as a corollary of this principle, since any predisposition thatdid not develop in this way would not serve the end of the organism and therefore, asKant says in his exposition of this proposition, would constitute “a contradiction in theteleological doctrine of nature.”In contrast to his seemingly dogmatic stance in the earlier essay, in the third Critique

Kant is careful to insist that the teleological principle pertains merely to the reflectiverather than to the determinative power of judgment and, as such, does not bring with itany ontological commitment. Again, it is a matter of judgment legislating to itself in itsendeavor to comprehend a certain type of natural object rather than to nature.Moreover, it is not that teleological replaces mechanistic explanation in biology,since the latter remains for Kant the only legitimate form of scientific explanation; itis rather that, for reasons already given, in the reflection on organic beings it mustalways be subordinated to the teleological principle.16

IVAs already noted, in the Critique of the Teleological Power of Judgment Kant distinguishesbetween intrinsic and extrinsic purposiveness. And while his main focus is on theformer in virtue of its centrality to biology, Kant does not entirely neglect the latter.Nevertheless, at least within the Analytic of the Teleological Power of Judgment, his

15 In the same context, Kant grants the possibility that many parts of an animal body could be conceived asconsequences of purely mechanistic laws, for example, hair, skin, and bones; but he insists that “the cause thatprovides the appropriate material, modifies it, forms it, and deposits it in its appropriate place must always bejudged teleologically, so that everything in it must be considered as organized, and everything is also, in acertain relation to the thing itself, an organ in turn” (KU 5: 377).

16 On this point see Kant, KU 5: 417–21.

244 PART IV

treatment of extrinsic purposiveness is both perfunctory and problematic.17 Thesituation reflects a tension in Kant’s account: on the one hand, he calls attention tothe sharp difference between the two conceptions of purposiveness; while, on theother hand, he suggests, without much in the way of argument, that the necessity ofassuming the former provides a warrant for introducing the latter.With respect to these two forms of purposiveness, the crucial distinction is between

judging something to be purposive in virtue of its internal form and judging theexistence of something to be purposive in virtue of subserving some end. And, withregard to the latter, if one is to avoid an indefinite regress of means and ends (a is ameans for b, b for c, etc.), it is necessary to assume something that is itself an end ofnature, a status which, Kant points out, can never be determined merely by theobservation of nature.18 Indeed, even the fact that something turns out to be necessaryfor human survival or flourishing, either in general or in particular locations, such as theseemingly inhospitable frigid zones, does not of itself suffice to establish such purpo-siveness, as long as it remains unclear why human beings had to exist at all (somethingwhich Kant ironically suggests is not so clear in the case of the New Hollanders or theFuegians19) or in the frigid regions.20

The issue is further complicated by the fact that assigning something (or somespecies) the privileged status of an end of nature presupposes that nature as a wholehas an end in the sense of a goal (“scopus”), which leads, in turn, to the supersensible.21

In other words, inasmuch as nature for Kant is not regarded as a self-conscious agent (inthe manner of Hegel’s Geist), to conceive of a purpose of nature (the sensible) is, at thesame time, to conceive of a purpose assigned to nature by something outside nature (thesuper-sensible). In fact, we shall see that for Kant ascribing an ultimate end [letzterZweck] to nature presupposes a final end [Endzweck] of creation, which takes us beyondanything that nature itself for all its “cunning” can bring about.Despite these complications, after emphasizing the radical difference between the

two species of purposiveness and the special difficulties associated with the latter, Kantinsists that the concept of a natural end, whose home base is the conception of organicbeings, “necessarily leads to the idea of the whole of nature as a system in accordancewith the rule of ends [Regel des Zwecke], to which idea all of the mechanism of nature inaccordance with principles of reason must now be subordinated (at least in order to testnatural appearances by this idea)” (KU 5: 379). Moreover, in this context, Kant furthersuggests that this leads to the maxims that “everything in the world is good for

17 See Kant, KU 5: 366–9 and 377–81.18 Ibid., 368–9.19 Ibid., 378.20 Kant here surmises that only the greatest incompatibility among human beings could have forced them

into such inhospitable regions (KU 5: 369). Elsewhere, however, he suggests that part of nature’s “plan” is touse war to drive human beings into even the most inhospitable regions of the earth and to this end may bethought to have provided certain purposive arrangements, such as Artic moss, which supplies food for thereindeer, which, in turn, afford sustenance and conveyance for the inhabitants of this region (EF 8: 363).

21 Kant, KU 5: 378.


something, that nothing is in vain,” and even that “by means of the example thatnature gives in its organic products, one is justified, indeed called upon to expectnothing in nature and its laws but what is purposive in the whole” (KU 5: 379). Inshort, it leads to an expansion of the domain of the teleological principle fromeverything within an organism to nature as a whole.Even though Kant issues the appropriate caveats regarding the status of this new

principle and the maxims it enjoins, insisting that it is a matter of the reflective ratherthan the determinative power of judgment and that the principle is merely regulativerather than constitutive, Kant’s interjection of necessity into his claim seems less thancompelling. Clearly, the necessity to which Kant refers cannot be either logical orcausal; but neither can it be of the same kind as the (subjective) necessity of using theconcept of an end in considering organisms, since that does not require one to think ofnature as a whole in these terms; and if it is not any of these forms of necessity, it isdifficult to see what other kind it could be.Accordingly, I think that the best way to make sense out of the passage is to ignore

the reference to necessity altogether and to attribute to Kant the weaker (though stillsubstantive) thesis that, once the unavoidability of regarding organisms as natural ends isrecognized, it becomes reasonable to apply the teleological maxims to nature as awhole. At the very least it is not unreasonable to do so, as it presumably would be if itcould be shown that organisms can be understood totally in mechanistic terms, sincethat would involve applying to the whole what has been shown to be not applicable tothe parts.This reading also reflects the non-dogmatic spirit of both Kant’s remark and his

amplification of it. As noted, he suggests parenthetically that the function of thisextension of the concept of an end to nature as a whole is to “test natural appearancesby this idea,”which can be seen as a matter of tentatively testing the teleological watersto see what emerges from the assumption that nature embodies a teleological system.22

Presumably, the possibility (though not the fruitfulness) of this extension is justified onthe grounds that a universal reductive mechanism has already been ruled out by theanalysis of organisms.In his amplification, Kant further suggests that when applied to things that appear

unpleasant and even counter-purposive to us, such as vermin, mosquitoes and otherstinging insects, it can lead to the recognition of positive purposes that they serve. Ingeneral, he remarks that such teleological consideration is at least “entertaining andsometimes also instructive.” (KU 5: 379). Although this is far from establishing thenecessity of a teleological approach to nature as a whole, it was evidently as far as Kantwas willing to go within the confines of the Analytic of the Teleological Power of Judgment,reserving his fuller treatment of the topic, including the application of teleology tohistory, for the Appendix.

22 For a similar approach in the first Critique see KrV A691/B719.

246 PART IV

VKant begins the pivotal }83 in the Appendix to Critique of the Teleological Power ofJudgment, which deals with the implications of the thesis that humankind is the ultimateend of nature, by referring back to }82, where it was initially affirmed. He writes:

In the preceding we have shown that we have sufficient cause to judge humankind not merely,like any organized being, as a natural end, but also as the ultimate end of nature here on earth, inrelation to which all other natural beings constitute a system of ends in accordance withfundamental principles of reason, not, to be sure, for the determining power of judgment, yetfor the reflecting power of judgment. (KU 5: 429)

Since the argument to which Kant here alludes provides the basis for his application ofteleology to human history, I shall endeavor in this section to give a brief account of it,before turning to its application in the next. The argument presupposes two proposi-tions that were supposedly established in the Analytic: (1) that we are permitted (at leastfor heuristic purposes) to consider nature as a whole as a teleological system; and (2)that this requires regarding something (or some species) as the ultimate end of nature.Given this, the central claim of }82 is that only humankind is a suitable candidate forthis role.The argument takes the form of a progressive iteration of the question: why (in the

sense of for what purpose) does something exist? Setting aside all of inorganic nature,since it can be understood mechanistically without any need to pose this question, Kantconsiders the vegetable kingdom, herbivores, and carnivores, which apparently consti-tute the three domains in which he divides (at least for the sake of this argument) non-human organic nature. Predictably, Kant suggests that the first may be thought to existfor the sake of the second, the second for the third, and all three for the sake ofhumankind, which is therefore the ultimate end of nature in the sense of being that forthe sake of which the others exist. And in support of this claim Kant notes that “he[humankind] is the only being on earth who forms a concept of ends for himself andwho by means of his reason can make a system of ends out of an aggregate ofpurposively formed things” (KU 5; 427).23

Inasmuch as it attributes a special status to humankind in virtue of its capacity to setends, the argument is reminiscent of the well-known argument for the so-called“Formula of Humanity” in the Groundwork, where Kant appeals to the same capacityto support the thesis that humanity (or rational agency) must be regarded as an end initself and therefore never treated merely as a means.24 In reality, however, thoughsharing a common premise regarding end-setting, these arguments operate atcompletely different levels and lead to quite distinct conclusions. The force of theGroundwork argument is entirely practical, issuing in a conclusion regarding howhumanity (in both oneself and others) ought to be treated. By contrast, the argument

23 See also Kant, KU 5: 431. 24 Kant, GMS 4: 428–9.


from the third Critique is theoretical (albeit one directed merely to the reflective powerof judgment) and it affirms the place of humankind in a teleological system of nature.As we shall see, the assignment of a special moral status to humankind only enters thepicture with the introduction of the quite distinct conception of a final end [Endzweck]of creation; and while Kant will argue that humankind can be regarded as the ultimateend of nature only if it is also the final end of creation, this connection is not operativeat this juncture in the argument.Nevertheless, Kant position here is radically anthropocentric and he was aware of the

objections that could be raised against such a position and attempted to deal with two ofthem. The first, which he associates with Linnaeus, affirms a more balanced view,according to which all members of the kingdom of nature, including human beings,may be regarded as ends in certain relations and as mere means in others. For example,while many aspects of nature are obviously beneficial to humankind and may thereforebe regarded as means, in virtue of activities such as hunting and the removal of some ofthe destructive powers of nature, humankind can also be viewed as a means formaintaining a certain ecological balance.25 Although Kant does not say what he findswrong with this conception, it seems clear that (1) by effectively denying that there isany ultimate end of nature, it would violate the condition under which it is possible toapply teleology to nature in the first place; and (2) in denying a privileged status tohumankind it ignores the significance of the fact that it is the only species of rationalbeings on earth and, as such, the only species capable of conceiving a system of ends.The second objection reflects the standpoint of a hard-headed mechanist. Basically, it

argues that it is unwarranted to assign a privileged status among natural beings tohumankind, since it is subject to the same destructive forces as everything else in nature;while even those features of the human habitat that seem to be arranged purposively forthe benefit of the species are products of the mechanism of nature. Speaking as the devil’sadvocate, Kant remarks that it is of no avail to attempt to avoid this conclusion by limitingthe scope of the mechanism of nature to the other species, since humankind is sodependent on these that if they are under the sway of this mechanism, it must be as well.26

In response, Kant counters that if this line of argument proves anything, it proves toomuch. This is because it not only shows that humankind is not the ultimate end ofnature and therefore that the aggregate of organized beings cannot be regarded as asystem of ends, but also that “even the products of nature that we previously held to benatural ends can have no other origin than that in the mechanism of nature” (KU 5:428). Kant’s point here seems to be that by reducing everything to the mechanism ofnature, the argument under consideration contradicts the analysis of organisms, whichis the foundation on which his account of natural ends is built. And inasmuch as he hadalready argued that the concept of a natural end makes it reasonable to regard nature asa whole as a teleological system, Kant apparently thought that this sufficed to dismiss

25 Kant, KU 5: 427. 26 Ibid., 427–8.

248 PART IV

the mechanistic critique. Although the argument seems somewhat dubious, we shouldkeep in mind the relative modesty of its claim. As we have already seen, rather thantaking himself as having provided a proof that humankind has the status of an ultimateend, he claims only to have shown that “we have sufficient cause” [my emphasis] toattribute such a status to humankind, at least with respect to the reflective power ofjudgment. Thus the question comes down to a matter of determining what counts as“sufficient.”

VISetting that aside, to claim that humankind is to be regarded as the ultimate end ofnature viewed as a teleological system invites the further question: what end is naturesupposed to effect with respect to humankind? Addressing this question is the task of}83. Kant assumes that this end is something that is to be found within human beingsthemselves, and he recognizes two generic candidates: it can either be one that issatisfied by the beneficence of nature itself, or some special aptitude that nature gives usthat is conducive to the pursuits of all sorts of ends. The first of these is happiness andthe second culture.27 After ruling out happiness on several grounds, the most compel-ling of which is that if this were nature’s aim with respect to humankind it did not do avery good job of it, Kant arrives by elimination at culture as the sought for end.Before discussing culture and its teleological function, however, it is necessary to say

a word about the connection between two key teleological concepts to which I havealready alluded and which are both applied to humankind by Kant: an ultimate end[letzter Zweck] of nature and a final end [Endzweck] of creation. Although a final end (inthe sense of the highest goal or purpose) is also an ultimate end, the converse need nothold. Indeed, it does not hold when the highest goal attainable by some means is notitself the highest goal conceivable for whatever it serves as means. Moreover, accordingto Kant, this is precisely the case with nature and humankind.Since, as we shall see in more detail in the next section, the final end for humankind

is moral and since morality is a product of freedom rather than nature, it follows thatnature’s ultimate end cannot be to make us moral. Nevertheless, this does not meanthat nature has no teleological role to play with regard to morality, merely that itcannot be a direct one. Kant’s point is perhaps best expressed by characterizing nature(in its teleological function with respect to humankind) as a “moral facilitator.” In otherwords, it helps us to help ourselves, often against our will. And to consider nature asfunctioning in this way is just what it means to consider it as a teleological system withhumankind as its ultimate end.As already noted, the vehicle through which nature promotes its end is culture;

though the situation is complicated by the fact that Kant distinguishes between two

27 Ibid., 429–30.


forms of culture: the culture of skill [Geschicklichkeit] and the culture of discipline[Zucht]. The former may be defined as the capacity to attain the ends that humankindsets for itself, whatever they may be. Kant characterizes the latter in negative terms asthe “liberation of the will from the despotism of desires” (KU 5: 431–2).Although for the reasons given above, the culture of discipline is not directly related

to morality in the sense of making us morally better, inasmuch as it helps to wean usfrom desires stemming from our animal nature, Kant assigns it a significant propaedeu-tic function. Specifically, he claims that it makes us receptive to higher ends than naturecan afford; and since these are moral ends this makes it a moral facilitator par excel-lence.28 Moreover, while it is impossible to pursue this topic here, it must at least benoted that this form of culture is intimately related to what Kant describes as the centralproblem of the third Critique as a whole, namely, providing a transition [Übergang] fromnature to freedom, in which the appreciation of beauty (both natural and artistic) playsa pivotal role.29 The general idea is that this liberation from sensuous desires creates acapacity for a disinterested delight in beauty, while the latter, in turn, helps to make usmore receptive to moral ideas.While the culture of discipline is of great significance for the third Critique as a

whole, it is the culture of skill that is most directly relevant to Kant’s teleologicalaccount of history. This is largely because it is the development of certain skills thatprovides, quite apart from the intent of the possessors and users of these skills, the spurto the material progress that leads eventually to the formation of civil societies withrepublican institutions and a confederation of states designed to preserve the peace. Inshort, this form of culture is the prime means by which the “cunning of nature”operates in history.It is in this spirit that Kant suggests that, combined with humankind’s natural

predispositions, cultural progress in this sense leads unavoidably to inequality, oppres-sion, an attachment to luxury, and all of the social and political evils that these bringwith them, including war, the greatest of all such evils. At the same, however, Kant alsoargues that all this “splendid misery” (KU 5: 432) will eventually force humankind todo what it would not of itself do willingly, namely, to work toward the developmentof those republican institutions and a confederation of states, which would eventuallyusher in a permanent state of peace. In this way, then, the culture of skill likewisefunctions as a moral facilitator, albeit in a somewhat more roundabout way than theculture of discipline.Since this is essentially the same view of history that Kant articulated in “Idea of a

Universal History” some six years earlier, I shall not say anything further about it here.For present purposes the major point is that }83 is the place in which Kant integratedhis account of history into the Critique of the Teleological Power of Judgment and thereforeinto the “critical” philosophy, which, as already noted, was very much a work in

28 I discuss the role of beauty as a moral facilitator in Allison (2001a), 229–35.29 For my account of the complex issue of the Übergang, see Allison (2001a), 195–218.

250 PART IV

progress in 1784, when he did not envision a third Critique, much less one dealing withteleology as part of its eventual content. At the same, however, it must also be notedthat Kant relegated his discussion to the Appendix rather than, as originally planned, tothe body of the work.30 Although this suggests a certain ambivalence on Kant’s part, itis not, I think, directed at the intrinsic importance of the material, since some of Kant’smost important thoughts about moral theology are likewise contained in this Appendixand are directly related to his account of humankind as the ultimate purpose of nature.Rather, Kant’s ambivalence concerns the place of this discussion within the systematicstructure of the third Critique, particularly his recognition of its parergonal status vis-a-vis the central argument of the Critique of the Teleological Power of Judgment.31

VIIThe preceding account of Kant’s application of teleology to humankind and its historyhas featured the following four theses. (1) If nature is to be regarded as a teleologicalsystem, it must be thought of as having an ultimate end, which can only be humankind.(2) Humankind may be considered as such an end only if it is also related to a final,unconditioned end, which must be moral. (3) Nature, by itself, cannot produce such anend, since that can only result from freedom; but it nevertheless can be thought of aspreparing the way for or facilitating the development of morality. (4) It does thisthrough culture, mainly the culture of skill, which, since it requires the development ofhumankind’s rational capacities, is a lengthy historical process, culminating in republi-cation institutions, which maximize freedom under law, and a confederation of statesguaranteeing perpetual peace.32 Considered from the point of view of the Kantianphilosophy of history, however, this invites the further question: does the progressfrom the ultimate end of nature to the final end of creation involve a historicaldimension or, as a work of freedom rather than nature, is it to be considered anatemporal “noumenal” process taking place in individuals, which, as such, stands totallyoutside history as well as nature?A glance at what says about the final end of creation in the later sections of the

Appendix of the Critique of the Power of Judgment, as well as the overall account of thesecondCritique, seems to suggest the latter view. Confining ourselves to the former text,Kant there introduces two distinct conceptions of the final end, which are eventuallybrought together. The first is humankind considered as noumenon. This is justified on

30 In the first introduction, Kant indicates that he planned to give equal weight to the discussions of innerand relative purposiveness, devoting a separate book to each, which will contain both an analytic and adialectic (see EE 20: 251).

31 Although Kant did not relegate his treatment of it to an appendix, there is a parallel story regarding thesublime, which Kant characterized as a “mere appendix to our aesthetic judging of the purposiveness ofnature” (KU 5: 246). For my discussion of this, see Allison (2001a), 306.

32 The long period of time required to attain this result is merely implicit in the third Critique account butmade fully explicit in IAG 8: 18–19.


the grounds that humankind, so considered, serves as a subject of morality (KU 5: 435).Somewhat later Kant characterizes the final end, understood in the same sense, as“humankind (each rational being in the world) under moral laws” (KU 5: 448).The other conception of a final end with which Kant operates is that of the highest

good (happiness in proportion to worthiness to be happy, that is, to morality). Strictlyspeaking, this is the highest good for us (human beings under moral laws) rather than ofcreation itself; indeed, for Kant it is one that we have a duty to promote. Kant bringsthem together in }88 by arguing that only insofar as we recognize a final end forcreation can there be one for us and therefore a duty to strive for its realization. Ineffect, this is the essence of Kant’s moral argument for the existence of God in the thirdCritique. Starting with the ubiquitous Kantian premise that ought implies can, it claimsthat we can only conceive of this end that we have a duty to fulfill as achievable on theassumption of the existence of a moral author of the world, for whom our attainmentof the highest good constitutes the purpose of creation.33

Nevertheless, from a Kantian point of view it also seems reasonable to expect thathistory will play a role in the full development of the predisposition to morality inhumankind. Otherwise, the ultimate end of nature would stand in no essentialconnection with the final end of creation, which would contradict Kant’s account inthe third Critique. To be sure, Kant is quite clear that the political and cosmopolitangoals at which the cunning of nature supposedly aims do not of themselves suffice toproduce genuine morality. For example, Kant claims famously that he considered thecreation of republican institutions possible even for a nation of devils, as long as they areintelligent (EF 8: 366). In the same work, however, Kant also remarks that “a morallygood condition of a people is to be expected only under a good constitution” (EF8: 366). Accordingly, Kant’s position seems to be that such institutions are necessary,though not sufficient, conditions of the full development of morality in a people,which gives them an essential propaedeutic function.Moreover, Kant hints at the idea of a further, trans-political aim of history already in

“Idea of a Universal History,” when he remarks that the confederation of states, itselfan unrealized ideal, would not mark the culmination of the historical process, butroughly its halfway point (IAG 8: 26).34 Although Kant does not tell us in this essaywhat he had in mind for this far distant second half of history, it seems reasonable tosurmise that the task that he assigns to it involves the attainment of the final end, which,as we have seen, consists in the realization (or least promotion) of the highest good.The puzzle is how this could be regarded as an historical process. Kant does not addressthis question in either the essay or the third Critique; but he does offer a possiblesolution to it in Religion within the Boundaries of Mere Reason, with the introduction ofhis conception of an ethical community, which incorporates both a social and ahistorical dimension into his doctrine of the highest good.

33 Kant, KU 5: 447–53. 34 The significance of this passage is underscored by Wood (1991b), 341.

252 PART IV

Since this is a complex topic, I can here do no more than baldly state the gist ofKant’s position as I see it.35 The underlying premise is that the development of avirtuous disposition, which is the portion of the highest good that is up to us, is notsomething that can be attained by individuals on their own, any more than it can bebrought about by nature. This, Kant suggests in a manner reminiscent of his earlieraccount of unsociable sociability, is because we become subject to malignant inclina-tions and the vices associated with them as soon as we are among human beings (RGV6: 94).36 Accordingly, an individual’s attainment of virtue is possible only as a memberof an association of human beings “merely under the laws of virtue” (RGV 6: 95).Such an association is an ethical community, as distinguished from a political commu-nity, which would be a society under principles of right or justice.37 The idea seems tobe modeled on that of the “realm of ends” of the Groundwork; but while Kant claimsthat, like the latter, this community is a mere ideal that can never be fully attained byimperfect beings such as ourselves (RGV 6: 100), he also insists that working toward itis a unique duty, “not of human beings toward human beings but of the human racetoward itself” (RGV 6: 97).Finally, Kant maintains that even approximating such a condition (which is the best

we can hope for) requires a lengthy process, wherein the focus is no longer on the statebut on the church, the goal being the gradual transformation of the latter from anecclesiastical to an ethical organization advocating a purely rational and moral religion.Since this is an historical process, presumably involving (at least ideally) humankind as awhole, it seems a likely candidate for the task that Kant assigns to the distant second halfof history in his self-proclaimed philosophical “chiliasm” (IAG 8: 26). Admittedly, itcannot be claimed conclusively that this expresses Kant’s intent in the 1784 essay, sincehis philosophy, particularly his moral philosophy, underwent a significant number oftwists and turns between then and 1793. Nevertheless, to adopt Kant’s expression, itprovides a guiding thread for the entertaining and perhaps useful project of interpretingKant’s philosophy of history as constituting a systematic whole.

35 The conception of an ethical community is also discussed in Allison (2009b), which is included in thisvolume.

36 This aspect of Kant’s thought has been most fully discussed by Wood (1999), 309–20.37 Kant also suggests that without the foundation of a political community, an ethical one could never be

brought into existence by human beings (RGV 6: 94), thus making the latter a necessary (though not asufficient) condition of the former.


Essay Seventeen

Reason, Revelation, andHistory in Lessing and Kant

The standard view during the Enlightenment was that the truth of a putative historicalrevelation was determined by two criteria. The first was its conformity with the dictatesof natural religion, understood as a set of truths concerning the existence and provi-dential governance of God, the immortality of the soul, and basic moral principles, allof which are readily accessible to unaided human reason. Any revelation that did notcontain these truths was dismissed as spurious. The second concerned those elements ofthe revelation, e.g., the messiahship of Jesus, the doctrine of the Trinity, and thevicarious satisfaction, which go beyond the truths of natural religion. Since theirclaim to being essential articles of faith depends on their revealed status, this criterionconcerned the facticity of the revelation claim itself. And since this concerned a matterof fact, albeit a highly anomalous one involving a divine intervention in history, thiswas thought to be subject to the standards of historical evidence. Accordingly, thequestion of whether Christianity contained the revealed word of God was believed toturn on the reliability of the historical evidence for the revelatory events narrated in theNew Testament.I shall here examine the critiques of this view provided by Lessing and Kant and the

alternative, historically based, models for understanding the relation between reasonand revelation that they put in its place. The discussion is divided into three parts. Thefirst analyzes the views of Lessing, with special focus on the “Education of the HumanRace” (henceforth referred to as Education). The second considers Kant’s views on therelation between reason and revelation in Religion within the Boundaries of Mere Reason(henceforth referred to as Religion). I conclude with some reflections regarding salientpoints of comparison between the two thinkers.

I(A) Lessing as theological polemicist: After a preliminary skirmish with neologists such asJ. A. Eberhard, whom Lessing accused of maintaining a superficial form of rationalism,which retained the historicity of the Christian revelation while emptying it of itsdistinctive content, Lessing immersed himself thoroughly in the theological disputesof the time with his publication, in his capacity as ducal librarian at Wolfenbüttel, of a

series of fragments from a manuscript of Hermann Samuel Reimarus.1 In theseselections, published by Lessing between 1774 and 1778, Reimarus launched, in thename of a purely natural religion of reason, a systematic assault on the conception of ahistorical revelation as a means of salvation, the claim of Christianity to contain such arevelation, and the authority of the Bible.2 This provided the perfect foil for Lessing,since it gave him the opportunity in his editorial “counter-assertions” to spell out hisprogram for the radical separation of the question of the truth of the Christian religionfrom the question of its historical foundations.Since Reimarus appealed to both of the aforementioned criteria in his critique of

Christianity’s claim to be based upon a divine revelation, Lessing’s counter-assertionsaddress both parts of this critique. The focal point of the first part of Reimarus’s critique isthe fragment entitled “That the Books of the Old Testament Were Not Written toReveal a Religion.” Appealing to natural religion as a norm, Reimarus argues that theOld Testament cannot be viewed as containing a divine revelation because it does notcontain the doctrine of immortality. Lessing accepts Reimarus’s premise but denies hisconclusion. Indeed, he goes beyond Reimarus, pointing out that prior to the Babyloniancaptivity the Israelites, with the exception of a few enlightened individuals, did not evenpossess the true concept of the unity of God. Nevertheless, Lessing insists that this has nobearing on the revealed character of the Old Testament. In support of this, Lessing arguesfirst that by this criterion the holy books of the Brahmans would have a superior claim tocontain a divine revelation than the Hebrew Scriptures, and second that the lack of thesedoctrines does not disqualify the Old Testament, since its contents were accommodatedto the limited understanding of the Israelites at the time.3 Lessing develops this theme inEducation; but before turning to this work, a word is in order regarding his response toReimarus’s critique of the historical foundations of Christianity.Reimarus explicitly raised the issue in two of the fragments: “The Passage of the

Israelites through the Red Sea,” which describes the absurdity of the Exodus narrative,and “On the Resurrection Narrative,” which points to the contradictions in thedifferent accounts of the resurrection. Lessing responds to the first in a somewhatflippant manner, offering alternative accounts of the Red Sea crossing, which serve toundermine the orthodox assumption of biblical inerrancy. He approaches the secondby distinguishing between contradictions among eyewitnesses and among second-hand reports based upon the transmission of the eyewitness accounts. By noting thatboth types of contradiction are to be expected, Lessing explains away the discrepanciesin the different Gospel accounts of the resurrection but, again, at the cost of totallyundermining the doctrine of biblical inerrancy.

1 The definitive account of neology and Lessing’s relation to it is Aner (1964). I discuss Lessing’s critique ofneology in Allison (1966), 83–95.

2 I discuss the views of Reimarus in Allison (1966), 42–9.3 Lessing, (1956) 7, 830–5; (2005), 75–9.

REASON, REVELATION, AND HISTORY IN LESS ING AND KANT 255

It was this last fragment that ignited the major attack from the side of Lutheranorthodoxy.4 The first published response was a pamphlet entitled On the Evidence of theProofs for the Truth of the Christian Religion by Johann David Schumann, who reaffirmed thetraditional doctrine of inerrancy, basing his argument onOrigen’s discussion of the “proof ofthe spirit and of power” in Contra Celsum. Lessing replied in his best-known polemicalwriting, “On theProof of the Spirit and of Power,”where he frames the issue in terms of thecontrast between direct experience and second-hand reports, apparently ignoring thenotorious problems regarding miracles, which were emphasized by Hume and othereighteenth-century thinkers. I believe, however, that Lessing is best read as proceedingstrategically, accepting for the sake of argument the possibility ofmiracles and their probativepower for those who believe they experience them, in order to underscore his main point.The latter is expressed in his famousmetaphor of the “broad andugly ditch,”which he notesironically that he “cannot get across, nomatter how often and earnestly I have tried tomakethe leap.”5 Although there is a good deal of ambiguity in Lessing’s account, it seems clearthat his basic point is the incommensurability between historical truths, nomatter howwellauthenticated, and religious truths, particularly those which appear to transcend the reach ofreason.6 From this point of view, the miraculous nature of the events being narratedbecomes a side issue. Lessing expresses the central issue with a rhetorical question:

[I]f I have no historical objection to the fact that Christ raised someone from the dead, mustI therefore regard it as true that God has a Son who is of the same essence as himself? Whatconnection is there between my inability to raise any substantial objection to the evidence for theformer, and my obligation to believe something which my reason refuses to accept?7

Given this incommensurability, it was no longer possible either to attack Christianity inthe manner of Reimarus or defend it by the traditional appeal to historical evidence. Andsince for Lessing the neological position, which combined the rejection of much of thecontent of the Christian revelation with a defense of its historicity, was not an option,there was need for a radical reconceptualization of the relation between the Judeo-Christian revelation and history. This is the task that Lessing undertook in Education.

(B) Education: Lessing published the first fifty-three paragraphs of this curious work in1777, and the entire work, consisting of one hundred paragraphs together with aneditorial preface to which he affixed his own name, anonymously in 1780. As the titlesuggests, its central theme is the analogy between revelation and education, whichLessing uses first to counter Reimarus’s rejection of the revealed nature of the OldTestament and then to provide a sketch of a developmental view of the history andfuture destiny of the Judeo-Christian tradition.

4 The initial responses from the orthodox camp were directed against the arguments of Reimarus ratherthan Lessing’s counter-assertions. The first to recognize that Lessing was the most dangerous enemy wasJohann Melchior Goeze. I analyze their controversy in Allison (1966), 107–20.

5 Lessing (1956), 8: 14; (2005), 87.6 For a discussion of the issue see Michalson (1985).7 Lessing (1956), 8: 13; (2005), 86.

256 PART IV

Neither the analogy itself nor his use of it was original with Lessing. The first to use itwas Clement of Alexandria, in response to Gnostic criticisms of the Old Testament.Much like deists such as Reimarus, the Gnostics challenged the revealed character ofthe Old Testament on the grounds that it contains an overly anthropomorphicconception of God and fails to include a doctrine of immortality. Clement respondedby characterizing the Old Testament as a preparatory revelation, accommodated to theintellectual level of the ancient Hebrews.8 A similar conception, together with adoctrine of stages of revelation, was also developed by Origen and Iranaeus.9 Inmodern times the analogy was used by Locke to counter deistic objections that areasonable God would not make salvation dependent upon the acceptance of anhistorical revelation, which must of necessity remain inaccessible to the bulk ofhumankind. Locke responds by admitting that, inasmuch as the truths it contains areaccessible to reason, the Christian revelation was not absolutely necessary; while alsoinsisting that it had an essential pedagogical function, since without it most peoplewould never be able clearly to apprehend these truths.10

Since these thinkers evidently took the analogy between revelation and educationstraightforwardly as expressing a communication by a transcendent God, the questionarises whether Lessing likewise understood it in this manner. Although there is somedisagreement, the dominant view is that the work’s reference to the revelatory actionsof a transcendent God is to be taken as an exoteric cloak for an esoteric, decidedly non-theistic doctrine.11 Having argued previously that Lessing’s theological position in-volves a combination of a Leibnizian perspectivalism with a broadly Spinozisticconception of God, my own reading falls into this camp.12 Here, however, I shalltry to show how, within the text itself, Lessing provides clues which strongly suggestthat he is using the analogy between revelation and education to deconstruct ratherthan to defend the traditional conception of revelation.13

I shall begin with an overview of the structure of the work. As noted, it consists ofone hundred numbered paragraphs preceded by an editorial preface in Lessing’s ownname. Prefixed to the entire text is a citation from St. Augustine to which I shall return.In his preface, Lessing remarks that the author stands on a height from which he can seefurther than his contemporaries, but that his vision is still not absolutely clear. The latterpoint is expressed poetically by the suggestion that the author sees “from that immea-surable distance, which a soft evening glow neither wholly conceals.”14 Presumably,what he sees in the “immeasurable distance” is religious truth; while the “soft evening

8 See Bigg (1866), 55–6.9 For a discussion of this conception in Origen, see Bigg (1866), 223; for Tertullian and Iranaeus see

Harnack (1961), 103 and 305–7.10 Locke (1965), 156–82.11 For a useful recent survey of the literature on the topic, see Wessel (1977), 17–46.12 I argue for this in detail in Allison (1966), 121–35.13 I provided an earlier statement of this line of analysis in Allison (1975).14 Lessing (1956), 8: 590; (2005), 218.


glow” refers to the contemporary perspective of the Enlightenment. In other words,Lessing (the editor) characterizes the (unnamed) author as someone who views thereligious situation from a determinate place within religious history rather than fromsome transcendent position. Moreover, by so locating the author Lessing is implicitlyplacing himself (as editor) at a somewhat higher, if not totally transcendent, standpoint.This is reflected in his query:

Why should we not see in all the positive religions simply the process whereby the humanunderstanding . . . can alone develop, and will develop further still, instead of reacting with eithermockery or anger to one of them? If nothing in the best of worlds deserves this scorn . . . why shouldthe religions alone deserve it? Can God`s hand be at work in everything except in our errors?15

The body of the text develops the analogy between revelation and education as adevice for interpreting the history of religion. It is suggested that “Education isrevelation imparted to the individual; and revelation is education which has been,and still is, imparted to the human race” (}2). This analogy is justified in terms of its“advantages” for theology, which turn out to be precisely the considerations that led itsprevious proponents to adopt it, viz., the defense of the revealed character of the OldTestament and the determination of a function for revelation that is compatible withthe view that it does not supply any truths that are not, in principle, accessible tohuman reason. The latter task is accomplished by appealing to an essentially Platonicconception of education, according to which:

Education gives the individual nothing which he could not also acquire by himself; it merelygives him what he could acquire by himself, but more quickly and easily. Thus revelationlikewise gives the human race nothing which human reason, left to itself, could also not arrive at;it merely gave it, and gives it, the most important of these things sooner. (}4)16

Applying this analogy, the Mosaic and Christian revelations are viewed as successivestages in the moral and religious education of humanity. The work ends with thesuggestion of a third and presumably final stage, characterized as the era of the “EternalGospel,” a glimmer of which was already perceived by “certain enthusiasts of thethirteenth and fourteenth centuries,” who “erred only in proclaiming that its comingwas so close at hand” (}87).

15 Lessing (1956) 8: 590–1; (2005), 218.16 It is often objected that there is a contradiction between this claim and the suggestion (}77) that despite

its dubious historical foundations, the Christian revelation may have led to the discovery of truths “whichhuman reason could never have arrived at on its own.” There is no contradiction, however, if we keep inmind the Platonic-rationalistic conception of education and the closely associated notion of an “occasion.”Qua rational, these “revealed” truths are not derived from without, and in this sense education (revelation)does not provide us with any truths which we could not, in principle, acquire for ourselves. But inasmuch asrevelation provides the occasion, which first enables us to recognize these truths, without it human reasonmight never, in fact, have come to recognize them. For further discussion of this, see Allison (1966), 158–9,and (1975), 186.

258 PART IV

So much for the plot; the question is what we are to make of it. In order to addressthis issue, I shall focus on two clues. The first is provided by the above-mentionedquotation from St. Augustine, which reads: “All these things are in some respects trueprecisely because they are in other respects false.”17 It is taken from a portion ofAugustine’s Soliloquies in which he is discussing how the notions of truth and falsityare applicable to things such as works of art and it constitutes the essence of Reason’s(Augustine’s spokesperson) answer. Reason also provides several examples that aredesigned to illustrate this principle. Typical of these is the painting of a horse. The pointis simply that in order for such a painting to count as a “true picture,” that is, as a workof art, it cannot contain a real horse. As a work of art, its “truth” depends upon itssuccess in creating an illusion and, in this sense, on its being “false.”Moreover, Reasonsuggests that this applies not only to works of art, but also to thing such as jokes andmirror images. With regard to this whole class of things, it may be said that, “Toestablish their truth, the only thing in their favor is that they are false in some otherregard. Hence, they never succeed in being what they want or ought to be, as long asthey refuse to be false.”18

It is not difficult to recognize the applicability of Augustine’s principle to the Mosaicand Christian revelations as depicted in Education. One need only consider the contrastbetween the implicitly rational, albeit historically conditioned, content of these revela-tions and the form in which they present themselves to their recipients. As the authornotes, speaking of the traditional Christian “mysteries,” “When they were revealed . . .they were not yet truths of reason; but they were revealed in order to become suchtruths” (}73). The point is that in order eventually to be recognized as containing suchtruths, these mysteries had to present themselves initially as being other than they reallyare, that is, as divine revelations rather than as human insights. Accordingly, like works ofart, these revelations may be said to be in some respects true, that is, successful pedagogicalinstruments or vehicles for communicating truths about God, human nature, and ourrelationship to God (}77), only insofar as they are also in other respects false.The second clue is embedded in the structure of the work and concerns what it

omits as much what it includes. Taken as a whole, the work traces the religiousdevelopment of humanity in terms of three concepts: “God,” “morality,” and “im-mortality.” According to the basic schema, each of these concepts is grasped in anincreasingly adequate manner at the successive stages of development. Thus, theancient Israelites initially had a relatively crude conception of God (}34), and it wasonly under the Persian influence that they were led to a genuine monotheism (}35).Once this insight was attained, however, the people, returning from the Captivity,were able to realize that this conception was implicit in their Scriptures all along (}40).With regard to morality and immortality, the situation was somewhat different; for,despite certain hints and allusions, the Old Testament really lacks a doctrine of

17 Augustine (1948). 18 Ibid.


immortality and because of this its moral teachings do not rise above a doctrine oftemporal rewards and punishments (}22).The New Testament reflects a higher level of moral and religious consciousness,

since it contains an explicit doctrine of immortality, of which Christ is described as “thefirst reliable and practical teacher” (}58). This goes together with the spiritualization ofmorality, i.e., an emphasis on an inward purity of heart rather than mere externalobservance (}61), which is accompanied by a development in the concept of God.Accordingly, the author suggests that the doctrine of the Trinity might be viewed asexpressing the truth that “God cannot possibly be one in the sense in which finite thingsare one, and that this unity must also be a transcendental unity which does not excludea kind of plurality.” In support of this speculation, it is argued that, as a perfect being,God must have a complete representation of himself, which, as such, must containeverything that is in him including his necessary existence. And, it is further suggestedthat those who wished to popularize this idea could have chosen no better metaphorthan to describe this necessarily existing replica as “a Son whom God begets frometernity” (}73).19

The anticipated final stage of education is referred to as the time of the EternalGospel, which the author notes is promised already in the New Testament (Revelations14:6) and which was proclaimed by Joachim di Fiori, a twelfth-century Italian mysticand his followers with their doctrine of the three ages of the world. Given the groundplan of Education, the reader is led to expect that this new gospel will involve adevelopment in the conceptions of God, morality, and immortality, analogous to theprogress from the Old to the New Testaments. With regard to morality, this is preciselywhat we find. We are told that at that time “man . . . will have no need to borrow themotives for his actions from this future; when he will do good because it is good, notbecause it brings arbitrary rewards . . . ”(}85). Correlatively, the belief in an afterlife isreplaced by a doctrine of metempsychosis in which the same individual appears manytimes and is thus able to progress through the whole course of education.20 Althoughthe author is notably silent concerning the conception of God pertaining to this finalstage, given the organizing principle at work it is not difficult to supply it for oneself.Just as the Christian conception of God was deemed an improvement over the Israeliteone because it provided a more adequate conception of the divine unity, whichincluded plurality within it, it can be inferred that a still more adequate conceptionwould include an even greater plurality within it. And since the greatest conceivable

19 Lessing sketched a similar view in “The Christianity of Reason,” which was written sometime in the1750s. The same Leibniz–Wolffian conception of existence as a perfection, which Kant famously criticized,plays a pivotal role in the argument in both texts; though in the “Christianity of Reason” Lessing goesbeyond the account in Education, completing the Trinitarian picture by introducing the conception of theHoly Spirit as the harmony between Father and Son; (1956) 7: 197–200; (2005), 25–9.

20 As is evidenced by a remark appended to the fragment “That More than Five Senses are Possible forHuman Beings,” Lessing appears to have seriously entertained the doctrine of metempsychosis. See (1956),8: 579; (2005), 182–3.

260 PART IV

plurality would encompass all things, this is tantamount to the Spinozistic tenet that“Whatever is, is in God, and nothing can be or be conceived without God.”21

If this reading is correct, it indicates that in Education Lessing implies, withoutexplicitly affirming, his adherence to a broadly Spinozistic conception of God.22

Moreover, with such a conception there is obviously no longer room for the traditionalview of revelation as an intervention in human history of a transcendent God, which iswhy I suggested that Lessing uses the revelation–education analogy as a tool fordeconstructing the concept of revelation itself. At the same time, however, it mustbe kept in mind that this is only half the story. The other and equally important half isLessing’s conservatism regarding the content of the Christian revelation, properlyunderstood. Since humanity, save perhaps for a few particularly precocious pupilssuch as Lessing, is not yet ready for the eternal gospel, his advice to his contemporariesis to emulate the procedure of the Israelites after their return from captivity and re-examine the lessons from their primer that had not yet been fully mastered. We shallsee that a similar dual message is found in Kant.

IIThe central question in Kant’s philosophy of religion is: “What may I hope for?” Heregarded this as one of the three central questions of philosophy, the others being: “Whatcan I know?” and “What ought I to do?” which are brought together in the underlyingquestion: “What is man?”23 Accordingly, Kant’s philosophy of religion is linked as closelyto anthropology as to theology. Indeed, in the Preface to the first edition ofReligionKantmakes the connection between the question of hope and the anthropological questionevident, when he characterizes the project of the work as “to make apparent the relationof religion to a human nature partly laden with good dispositions and partly with evilones.”24 Here, as elsewhere, Kant understands by religion the recognition of duties asdivine commands; but the type of religion inwhich he is interested is a purely rational one,which remains “within the boundaries of mere reason.”The situation is complicated, however, by an ambiguity in the concept of hope. On

the one hand, it refers to what an individual may hope for, namely, happiness inproportion to one’s worthiness of it. This is also the highest good for an individual. Onthe other hand, it concerns the collective question of what may be hoped for thehuman race as a whole. Kant terms the object of this hope “the highest good as a good

21 Spinoza (1982), 40 (Ethics, Part 1, Proposition 15).22 I say “broadly Spinozistic” because (1) Lessing’s theological forays, as well as his notorious conversations

with Jacobi, are too sketchy to establish more than a sympathy for a form of pantheism and (2) Lessing alsoadhered to a form of providentialism, which is in the spirit of Leibniz and far removed from the doctrine ofSpinoza.

23 See Kant’s letter to Carl Friedrich Stäudlin of May 4, 1793 (Br 11: 429), which he sent together with acopy of Religion. In the letter Kant informs Stäudlin that with this work he has attempted to complete thethird part of his plan (the part dealing with hope).

24 Kant, RGV 6: 11.


common to all.”25 The latter involves not merely an arrangement in which virtue andvice are appropriately awarded and punished, but in which humanity fulfills itsvocation [Bestimmung]. Both aspects are addressed in Religion; for whereas parts oneand two deal respectively with the nature of radical evil and the possibility of individualredemption, parts three and four are concerned with the problem of the collectiveredemption of humanity and the role therein of the Church.Although the latter two parts deal more directly with the issues of interest to both

Kant and Lessing, inasmuch as Religion purports to contain a single line of argument, itis necessary to begin with a consideration of what precedes. This includes the doctrineof radical evil; but having discussed it at length elsewhere, I shall here simply assumeKant’s thesis that radical evil, understood as a propensity to reverse the proper ofincentives, subordinating that of morality to self love, is an imputable, yet ineradicablefeature of human nature.26 Thus, I shall begin with a consideration of the argument ofthe second part ofReligion, which is concerned with the struggle between the good andevil principles for dominion [Herrschaft] over the human being.

(A) The nature of the moral struggle: Unlike most Enlightenment moralists, includingLessing, who believed in the possibility of moral progress for both the individual andthe race through the development of reason, Kant insists that the starting point for anysuch development is evil rather than mere ignorance or lack of intellectual develop-ment, from which it follows that moral progress cannot be regarded as linear. Rather,its very possibility presupposes a radical dispositional change, which has seemed tomany as a bootstrapping operation of colossal proportions. Simply put, the problem isto understand how a disposition that is evil in the radical sense specified by Kant couldremake itself into one that is good.27

Setting that issue aside, what is of interest to us here is that in discussing the demandfor such a change, which he justifies on the familiar grounds that ought implies can,Kant appeals to the conception of a prototype of moral perfection, which he char-acterizes in Johannine, terms, as “in him [God] from all eternity” and as “God’s only-begotten-Son, ‘the Word’ . . . through which all other things are, and without whomnothing that is made would exist.”28 Despite this biblical terminology, Kant makes itclear that he understands the prototype as an idea of reason.29 Its function is to providea model of moral perfection that we can emulate. Kant suggests that this language isappropriate because we are not the authors of this idea and cannot explain its presencein our reason. Accordingly, it is more fitting to think of it as having come down to us

25 Ibid., 67.26 My most recent treatment of this topic, Allison (2002), is included in this volume.27 For a thoroughgoing critique of Kant’s account see Michalson (1990). I discuss some of the issues

involved in Allison (1990), 136–45, 162–79.28 Kant, RGV 6: 60.29 For an analysis of Kant’s conception of the prototype and the role that it plays in his argument, see

Firestone and Jacobs (2008), 155–80.

262 PART IV

from heaven and as taking up not merely the form but also the burdens of humanity.Moreover, Kant points out that it is only by representing it in the latter way that it canbe effective as a model for emulation; for a perfectly holy being, who merely assumedthe form of a human being without also being subject to the burdens and temptationsthat are inseparable from the human condition, could hardly provide such a model.30

Kant speaks of a “practical faith in this Son of God (so far as he is represented as having takenup human nature).”31 This faith is not in the historical existence of Jesus as the Christ butin a human being’s capacity to emulate this prototype and thereby become wellpleasing to God. In a similar vein, Kant also speaks of the objective practical realityof this idea, which resides not in a Christian (or Platonic) heaven but in our morallylegislative reason.32 To attribute objective practical reality to this idea is to say that it isnot an empty ideal but a valid norm which we have a duty to strive to satisfy.Kant notes three difficulties, however, which even after the required dispositional

change, cast doubt on our capacity to realize the ideal of being well pleasing to God: (1)Since all human deeds fall short of the holiness required, how can the dispositioncompensate for the insufficiency of one’s deeds? (2) How can we be sure that we willnot relapse into our old sinful ways? (3) Since we all start from evil and since there is nopossibility of making up for this in the future with a surplus of good works, how canone repay the previously acquired debt of sin?Of these, the last, which Kant characterizes as the most serious, deserves our attention

because it is central to his analysis of the function of revelation. Although Kant’s accountis notoriously obscure, it seems to turn on two main points. The first is that it is in thevery process of dispositional change, whichKant describes in Pauline terms as the castingoff of the old man and the birth of the new, that the requirements of divine justice forthe sins of the old man are met. The idea is that the act of conversion or dispositionalchange is at the same time both the birth of the new man and the death of the old, whothereby receives his deserved punishment. The second point is that this punishment forthe dispositional guilt of the oldman does not endmatters; for, given that the propensityto evil is inextirpable, even the newman continues to fall short of the strict requirementsof holiness. It is this problem that Kant addresses by appealing to the prototypepersonified as the son of God. His suggestion is that as a result of adopting the gooddisposition, which emulates that of the prototype, God grants to the newman a graciousdispensation from the debt of sin.33

Kant’s introduction of the concept of grace is jarring in two respects. First, it appearsto be an unwelcome intruder into a religion within the boundaries of mere reason.Second, it threatens to undermine Kant’s ethics of autonomy. Although a consider-ation of the second worry would take us too far afield, the first must be addressed, sinceit speaks directly to the coherence of Kant’s project in Religion.

30 Kant, RGV 6: 61. 31 Ibid., 62. 32 Ibid. 33 Ibid., 74–5.


The crucial point is that Kant’s conception of grace is the counterpart of radical evil;so if the latter has a place with a religion within the boundaries of mere reason, then theformer does also. Recall that the object of hope is to become well pleasing to God.That this condition is attainable is the object of a practical faith. The problem,however, is that, given our inextirpable propensity to evil, this goal is unattainableby human effort, even though the pursuit of it is morally required. Although this doesnot violate the ought implies can principle, since our obligation is to do everything inour power to attain this goal rather than actually to attain it, it leaves the hope inbecoming well pleasing to God without any basis. It is at this point that the conceptionof grace enters the story, supposedly providing grounds for such hope.Nevertheless, Kant is extremely guarded in his appeal to this conception. He

considers both the effects of grace on its recipient, and the means whereby God impartsit as parergonal to a religion within the boundaries of mere reason.34 Indeed, he isacutely aware of the dangers involved in appealing to these conceptions, understood asdogmatic attempts to explain the inscrutable workings of God. Not only are suchappeals theoretically vacuous, since they involve an illegitimate venture into thetranscendent, but, more insidiously, they draw attention away from where it properlybelongs, namely, the moral demands upon the individual. Despite this, however, Kantaccepts as part of a rational faith the “general supposition that grace will work in uswhat nature cannot if we have just made use of that nature (i.e., of our own forces)according to possibility.”35

(B) The church and the kingdom of God: The distinctive feature of the third part of Religionis the shift from an individualistic to a social-historical analysis of the human predica-ment, which brings with it a relocation of hope from the individual to the human raceas a whole. This shift is motivated by the thought that the work of Christ, viewed as apersonification of the good principle, succeeded in destroying the dominion (i.e., theclaim to legitimacy) of the evil principle, as personified by Satan, but not in conqueringit. Thus, while no longer in bondage to sin, humanity still must struggle with it.Although the expression is not used, the basis for this shift lies in Kant’s conception

of “unsociable sociability” as an essential feature of the human nature.36 In effect, Kantarticulates a Rousseauean equivalent of Sartre’s dictum that “Hell is other people,”though, like Rousseau, he rejects the correlative Sartrean thesis that there is “no exit.”The former is the case because a human being becomes assailed by the destructivepassions, such as envy, addiction to power, and avarice, “as soon as he is among humanbeings.”37 And the latter is true because and the latter because a solution is provided bythe idea of an ethical community. As standing under laws of virtue (i.e., non-coercivemoral laws), this is contrasted with a juridico-civil society, which stands under coercivejuridical laws. Kant maintains not only that such a community has objective practical

34 Kant, RGV 52. 35 Ibid., 191.36 Kant, IAG 8: 20–2. 37 Kant, RGV 6: 93–4.

264 PART IV

reality as an idea of reason, but also that there is a collective duty, applying to the entirehuman race, to abandon an ethical state of nature and to join such a community.38

Kant’s initial concern is to develop the concept of this community. He notes that,like any community, it must be governed by public laws; but as an ethical community,these laws must themselves be purely ethical, that is, non-coercive. And since the onlyconceivable law-giver for such an ideal community is God, an ethical community canbe conceived only as a people of God. The complete realization of the idea of such acommunity can never achieved by human beings, but, insofar as it is possible, it willtake the form of a church, which, if it is to be the true church, that is, the closestpossible approximation of the ideal of an ethical community under human conditions,must have the marks of universality, purity, freedom, and unchangeableness.39

After specifying the criteria of a true church, Kant turns to a consideration of thechurch as a human institution. At this point, his concern is not with the Christian orany actual church, but with something like a Weberian ideal type. Underlying Kant’saccount is the contrast between a pure religious faith based on moral principles and anhistorical faith based on facts. While insisting that only the former pertains to a universalchurch, Kant laments that “due to a peculiar weakness in human nature” such faithalone is insufficient to found a church.40 This weakness is a propensity to regard thetask of becoming pleasing to God as consisting in the performance of certain servicesbeyond (or instead of) fulfilling the requirements of morality, and it leads to the need toincorporate into the constitution of a church certain statutory requirements. Sinceunlike moral requirements, these cannot be discerned by mere reason, they must beviewed as revealed. And since whether something is a divine revelation is a question offact concerning an historical occurrence, this means that the actual faith of a church,which Kant terms an “ecclesiastical faith,”must include an historical as well as a rational(moral) component.Kant expresses the need of any actual church for an extra-moral, historical compo-

nent by characterizing the belief in a revelation as a “vehicle” [Leitmittel] for thepresentation of the universal religion of reason.41 He also claims, however, that achurch that is progressing toward its ultimate telos of becoming the true church mustbe conscious of this historical element as merely a vehicle and therefore as somethingthat must ultimately be dispensed with.42 Kant had already expressed the essentials ofthis view eighteen years earlier in a letter to Lavater, where, after acknowledging thatrevelation and miracles may have been needed at the time of Christ to awaken faith, heinsisted that “once the true religious structure [a pure religion of reason] had been builtup so that it could maintain itself in the world, then the scaffolding must be taken

38 Ibid., 97.39 Ibid., 101–2. Kant here links these marks with the four titles from the table of judgments in the first

Critique.40 Ibid., 103.41 Ibid., 115; 146, 116.42 Ibid., 115.


down.”43 Accordingly, even though it will involve a lengthy process, Kant saw thehistorical task for the Christian Church to be the eventual removal of this scaffolding.Kant’s explanation of why this removal must be a lengthy process is grounded in his

analysis of what he terms a “saving faith” [seligmachenden Glauben]. The concept of sucha faith contains two conditions. One concerns what the individual cannot do, namely,undo the immoral conduct that has already been done; the other concerns what theindividual can and ought to do, namely, convert to a new life in conformity with duty.The former is a faith in God’s redemptive grace; the latter a faith in a capacity to leadthe good life.44 Since both are necessary, the question becomes which has priority.Although Kant presents the question of precedence in the form of an antinomy,

which recapitulates the familiar dialectic between works and grace, it is impossiblehere to pursue the details of this account.45 For present purposes, it must suffice tonote two points. First, Kant links the first question with the historical portion ofreligion, since this is something that could only be learned through a divine revelation,whereas common human reason is capable of knowing what one must do to be-come pleasing to God. Second, Kant characterizes this question as theoretical ratherthan practical.It follows from this that the need for an extra-moral element, and therefore a belief

in an historical revelation, will remain only as long as an answer to the theoreticalquestion (which can only be acquired through revelation) is itself regarded as aningredient in a saving faith. But since Kant held that it is contrary to human (practical)reason to make acceptance of a theoretical proposition an ingredient in (not to mentiona condition of) a saving faith, he concludes that we can expect that the sufficiency ofthe morally based faith will eventually become universally recognized; so that “at lastthe pure faith of religion will rule over all.”46

In the second division of part three, Kant turns to a consideration of what historyteaches us about humanity’s progress toward this goal. This is a particularly rich section,which must be seen in connection with Kant’s philosophy of history and serves as afocal point of a comparison of Kant’s views on the historical development of religionand Lessing’s Education.47 I shall here confine myself, however, to two points, savingthe comparison with Lessing for the next section.The first is that Kant’s history of religion, which is really an ecclesiastical history, does

not allot a place for ancient Judaism. This is because he thought that its faith was apurely political one in the establishment and restoration (through the Messiah) of theJewish state. As partial evidence for this, he cites (in agreement with Reimarus and in

43 Kant, Letter to J. C. Lavater, April 28, 1775, Br 10: 177.44 Kant, RGV 6: 115–16.45 For a critical discussion of this antinomy and Kant’s resolution see Michalson (1990), 100ff.46 Kant, RGV 6: 121.47 The relation between Kant’s philosophy of religion and his philosophy of history is a much discussed

topic in the literature. Perhaps the seminal work on the subject is Troeltsche (1904). More recent treatmentsinclude Despland (1973) and Yovel (1980).

266 PART IV

opposition to Lessing) ancient Judaism’s lack of a doctrine of immortality.48 Kantacknowledges that Judaism eventually became a genuine religion but only throughthe same foreign influences to which Lessing referred. Accordingly, the ecclesiasticalhistory in which Kant was interested, the central theme of which is the struggle forsupremacy between a pure religion of reason and one based on an historical faith,begins with Christianity.The second point is that this history also ends with Christianity. In other words,

rather than being a stage in the development of the religious consciousness, the historyof the Church is this history. This is because for Kant Christianity was from itsinception in principle a pure moral religion, which entails that there is no room forany development beyond it. Rather, the development that Kant envisages is a purelyimmanent one, in which the Church gradually discards its statutory features and comesto be recognized as the repository of a purely rational faith.

(C) Kant’s pure rationalism: The fourth part of Religion is concerned with the contrastbetween a genuine (i.e., moral) and a spurious service to God. Its philosophical interest,however, consists largely in the contrast Kant draws between various standpoints vis-a-vis religion, the distinction between a natural and a revealed religion, and the dualstatus of Christianity.Kant begins by defining religion (subjectively considered) as “the recognition of all our

duties as divine commands.”49 This, in turn, leads to Kant’s non-standard characterizationof the distinction between revealed and natural religion. The former is characterized asone in which we must recognize something as a divine command (learned throughrevelation) before we can acknowledge it as a duty; the latter is one in which we mustrecognize something as a duty before we can regard it as a divine command.50 In light ofthis Kant sets out the following taxonomy of theological positions:

Anyone who declares natural religion as alone morally necessary, i.e. a duty, can also be calledrationalist (in matters of faith). If he denies the reality of any supernatural divine revelation, he iscalled a naturalist; should he, however, allow [zulassen] this revelation, yet claim that to takecognizance of it and accept it as actual is not necessarily required for religion, then he can benamed pure rationalist; but if he holds that faith in divine revelation is necessary to universalreligion, then he can be called pure supernaturalist in matters of faith.51

This classification has been the source of disagreement regarding both the number ofpositions Kant distinguishes and the one he assigns to himself. Although some inter-preters distinguish four views (rationalism, naturalism, pure rationalism, and supernat-uralism), I believe that naturalism and pure rationalism are species of rationalism, sincethey both hold that natural religion alone is morally necessary.52 Accordingly, Kant

48 Kant, RGV 6: 125–7. 49 Ibid., 153–4.50 Ibid., 154. 51 Ibid., 154–5.52 Those who distinguish between four views include Collins (1967), 159–61, and Despland (1973), 220.


actually distinguishes three positions: two forms of rationalism and pure supernatural-ism. And since it is non-controversial that Kant was not a supernaturalist, the issuebecomes what sort of rationalist he took himself to be. Inasmuch as Kant does not denythe possibility (or actuality) of a supernatural revelation, he is not a naturalist, whichleaves us with pure rationalism.53 Against this it has been argued, however, that Kant isnot a pure rationalist because he did not accept (believe in) the actuality of anyhistorical revelation.54 Here everything turns on the translation of “zulassen.” If it istaken in the strong sense to mean believing in the reality of such a revelation, quasupernatural intervention in history, then it is at best unclear that Kant held such aview. But if one takes the term in a weaker sense to mean something like not denyingsuch claims, which would involve being agnostic rather than dogmatically negativeabout historical revelation, it seems clear that this represents Kant’s view.To further complicate matters, Kant introduces a new distinction between natural

and learned religion, which concerns the basis of their communicability rather thantheir origin. Underlying the distinction is Kant’s view that the true religion must beuniversally communicable. Such communicability characterizes natural religion, sinceit is based solely on universal human reason; but a learned religion cannot, as such, beuniversally communicable, since its propagation depends upon a contingent access toits oral proclamations and written texts. Although Kant does not explicitly say so, itseems clear that the class of revealed religions is coextensive with the class of learnedones.Immediately after drawing this distinction, Kant concludes that a religion can be

objectively natural and subjectively revealed. This would occur “if it is so constituted thathuman beings could and ought to have arrived at it on their own through the mere use oftheir reason, even though they would not have come to it as early or as extensively as isrequired . . . ”55 At first glance, this seems to yield a contradiction similar to the one thatis often alleged to hold between }4 and }77 of Lessing’s Education.56 The problem isthat if reason not only could but ought to have arrived at the truths contained in naturalreligion on its own, there seems to be no need for revelation. There is no contradic-tion, however, if one emphasizes Kant’s qualification that unaided human reasonwould not have arrived at these truths “as early or as extensively as is required.”Accordingly, the need for revelation, understood as a vehicle, is perfectly compatiblewith a religion that is objectively natural.Kant endeavors to confirm this thesis by considering Christianity as an example of a

religion which may properly be considered as both natural and learned. By naturalreligion Kant now understands morality combined with the postulates of God andimmortality, that is, to say, the tenets of his own moral religion.57 This religion is“natural” because it is based on a pure practical concept of reason, which accounts for

53 I am here in agreement with Collins and Despland. See above note 52.54 This position is taken by Wood (1991a), 10–12. 55 Kant, RGV 6: 155.56 See above note 17. 57 Kant, RGV 6: 157.

268 PART IV

its universality. To actually attain such universality, however, it must take the form of avisible church, which is the task that Kant assigns to Jesus and his followers. In essence, heclaims that Jesus was the first teacher of Kantian morality, including the principle that oneshould do one’s duty “from no other incentive except the unmediated appreciation ofduty itself.”58 And to this was added the doctrine of immortality, regarded as a reward forrather than an incentive to morality.59 Kant notes that these pure moral teachings werecombined with an appeal to miracles and allusions to the older Mosaic revelation asforeshadowing the teachings of Jesus, but these are regarded as devices for bringing thenew religion to the attention of the unlearned rather than as integral parts of it.60

Since in addition to containing the precepts of a pure religion of reason Christianityis obviously a learned faith, Kant does not have to argue for this part of his thesis.Rather, the aim of his brief discussion of Christianity as a learned religion (or, moreproperly, faith61) is to clarify the proper relationship between its two components.Echoing his account of radical evil, Kant sees the issue as one of primacy. In short, theproper relationship is one in which the moral element is deemed primary and thelearned elements provide merely the means, “though a most precious one” forspreading and maintaining the natural religion even among the ignorant.62 Kantterms such a faith the “true service [Dienst] of the church.”63 Conversely, the false orcounterfeit service [Afterdienst] is one in which primacy is given to belief in the factualelements, the authentication of which depends on historical evidence. And the over-coming of the tendency to view Christian faith in this way defines the Church’s still tobe completed historical task.

IIIHaving sketched the views of Lessing and Kant regarding the relation betweenreason and revelation, I shall conclude with some comparative reflections. AlthoughKant was undoubtedly familiar with many of Lessing’s writings, since his references toLessing are spare and the extent of this influence difficult to determine, I shall set asidethe question of a direct influence and focus instead on a comparison of their views onthe topics discussed in the first two parts of this essay.64

58 Ibid., 160.59 Ibid., 161.60 Ibid., 162–3.61 At RGV 6: 164 Kant suggests that, insofar as its doctrine is based upon facts and not mere concepts of

reason, the Christian religion is more properly termed the Christian faith. Previously Kant had claimed thatwhile there can be only one (true) religion, there can be several kinds of faith (RGV 6: 107–8).

62 Kant, RGV 6: 165.63 Ibid.64 Setting aside references to Lessing by Kant’s correspondents and places where Lessing’s name appears

more than once on the same or adjoining pages, there are only ten references to Lessing in the AkademieAusgabe. Moreover the only one of these that is of any philosophical import is in the third part of Theory andPractice, where Kant defends Lessing`s view of historical progress for the human race against Mendelssohn’scritique of that notion in Jerusalem (TP 8: 307–8). In Religion, while not mentioning Lessing, Kant strongly


To begin with, there are four major areas of agreement. The first is the sharp linethey draw between religious and historical truth. Lessing’s dictum that “contingent truthsof history can never become the proof of necessary truths of reason” is a principle that is alsoaffirmed by Kant in substantially similar terms. The second is a historicized view ofrevelation, according to which an implicitly rational content is initially presented in away that is accessible to those who are not yet in position to grasp it in its rational form.The third is that a significant segment of the history of religion is viewed as a process inwhich the rational nature of this content is progressively realized, leading to theeventual overcoming of the need for any reliance on its revealed character. AlthoughLessing uses the metaphor of a pedagogical tool and Kant that of a vehicle (or ascaffolding), they held similar views regarding the content and function of revelation.Finally, even though both thinkers maintained that a reliance on the Christian revela-tion as the ultimate source of religious truth was destined to be eventually outgrown,both were insistent upon its continued relevance. Thus, both were conservative whencompared with the views of many of their contemporaries. Moreover, both weresharply criticized for their conservatism at the same time as they were being attacked bythe orthodox establishment for their heretical views.65

There are also a number of noteworthy areas of disagreement, however, which reflectsharp differences in their respective styles, motivations, and philosophical commitments.Although extraordinarily learned, particularly in patristics, Lessing was a self-proclaimed“amateur of theology, not a theologian.”66 And much the same could be said regardingphilosophy. Thus, despite havingmade an in-depth study of both Leibniz and Spinoza, hisown position seems to have been somewhat eclectic, expressed in suggestive fragmentsrather than systematic works.67 Correlatively, Lessing’s intent in joining the theologicalcontroversies of the 1770s seems not to have been to argue for a specific position, but toserve as a gadfly, with the aim of shifting the debate to a new and more fundamentalquestion, namely, the nature of religious truth and its relation to an historical revelation.As far as theology is concerned, Kant might well have accepted Lessing’s self-

characterization as an amateur.68 Obviously, however, the same cannot be said regard-ing philosophy; for whereas Lessing’s philosophical forays tend to be fragmentary and

criticized Reimarus’s (to whom he refers as the “Wolfenbüttel fragmentarist”) account of the resurrectionnarratives (RVG 6: 81n). For an account of Kant’s debt to Lessing, which tends to minimize the degree of hisinfluence, see Bohatec (1966), 51ff., 453–5.

65 Prominent among Lessing’s critics for his apparent betrayal of Enlightenment principles was his brotherKarl, who speaking not only for himself, but also for Lessing’s Berlin friends, Mendelssohn and Nicolai, wasdisturbed by Lessing’s theological writings in the early seventies. Lessing’s letters to his brother during thisperiod are among the major sources for understanding his views at the time. For a discussion of this see Allison(1966), 83–8. Critics of Kant’s account of radical evil in Religion include Schiller and Goethe. On this seeAllison (1990), 146–7 and 270 note 1.

66 Lessing (1956), 8: 167; (2005), 122.67 For my account of Lessing’s philosophical studies, which includes the beginning of a translation of

Leibniz’s New Essays, see Allison (1966) 50–79.68 For an account of the theological influences on Kant and a discussion of the literature on the topic, see

Bohatec (1966), 19–32.

270 PART IV

suggestive rather than thorough and systematic, Kant is the systematic thinker parexcellence. Indeed, behind Religion lies a complete moral theory, which includes viewsregarding God, freedom, and immortality. Moreover, Kant had two distinct doctrinalgoals in Religion: (1) to provide his definitive response to the question: “What mayI hope for”; and (2) to establish not merely the compatibility, but the union betweenreason and Scripture.69 Lessing differs from Kant on each of these points. While Kantwas a theist (though his theism was grounded in practical rather than theoretical reason)and libertarian, and regarded immortality as a postulate of practical reason, Lessing wasevidently a pantheist of sorts,70 a determinist,71 and at least entertained a doctrine ofmetempsychosis.72 And while he claimed to find a rational kernel in Christian doc-trines that were rejected by the neologists, Lessing would not have gone as far as Kantin proclaiming the unity of reason and (Christian) Scripture.This last point is evident from their respective treatments of the historical role of

Christianity. As we have seen, Lessing assigned a middle role to Christianity, regardingit as the successor to Judaism and the prelude to the Eternal Gospel. Among otherthings, this reflected his Leibnizian perspectivalism, which finds an element of truth inevery position; his resistance to the idea of judging every revealed religion by an a priorinorm, such as one which requires that it contain a doctrine of immortality; and hisnon-theistic conception of God, which created space for a post-Christian theology.Kant, by contrast, did apply an a priori norm to religion and in light of it denied thatancient Judaism was a genuine religion, because it was neither based on moralprinciples nor included a doctrine of immortality. Accordingly, Christianity for Kant,as in principle a pure moral religion, marked a radical break with Judaism rather than adevelopment of it. And since Kant was a theist and regarded Christianity as already amoral religion, he saw no need for a development beyond it. Rather, the onlyhistorical development to which Kant attached any importance was one withinChristianity itself, through which it gradually becomes conscious of its pure moralcontent and casts off the extrinsic factors connected with the belief in its revealed status.Viewed through Kantian spectacles, perhaps the most interesting aspect of Lessing’s

position inEducation is his suggestion that an ethic of autonomy (one in which “man . . .willdo good because it is good, not because it brings arbitrary rewards”) is part of the content of afuture Eternal Gospel rather than of the moral teachings of the New Testament. This isreflected in Lessing characterization of Jesus as the “the first reliable and practical teacher” ofthe immortality of the soul rather than, as he effectively was for Kant, of autonomy. Andgiven Kant’s exhaustive division of moral theories into those based on the principle of

69 Kant, RVG 6: 13.70 See note 23 above.71 Since it does not enter directly into Lessing’s counter-assertions to Reimarus, the dispute engendered

by the publication of the fragments, or Education, I did not discuss Lessing’s determinism in the first section.Kant was undoubtedly familiar with it, however, through Jacobi’s report of his conversations with Lessing.I discuss Lessing’s determinism in Allison (1966), 73–5 and 146–7.

72 See note 21 above.


autonomy and those based on some form of heteronomy, this entails that for Kant theChristian ethic, as Lessing understood it, was a form of heteronomy, inasmuch as it regardedreward in a future life as the true moral incentive. Moreover, if this were the case, it wouldalso follow that Christianity could not be regarded as a pure moral religion in Kant’s sense.In view of his emphasis on autonomy and rejection of an ethics based on future

rewards and punishment as a species of heteronomy, one may wonder why Kantinsisted that any genuine religion must contain a doctrine of immortality. The answerlies in his conception of the highest good for an individual, namely, happiness inproportion to one’s worthiness of it, which we have seen Kant describes as the object ofhope. Since the main function of religion for Kant is to offer a morally based hope, andsince such hope is inseparably connected with belief in a future life, any religionworthy of the name must include among its tenets a doctrine of a future life. Butsince the hope for an eternal happiness presupposes that one is worthy of it, and sincethis worthiness requires that one has been motivated by the thought of duty rather thanan expected reward, this hope must be sharply distinguished from the moral incentive,which alone renders it compatible with an ethic of autonomy.73

Although there is no counterpart in Lessing to Kant’s complex moral psychology, histreatment inEducationof the threeChristian doctrines,which he cites as examples of truthsthat have been too hastily discarded, makes for an interesting contrast with Kant. The firstand most important of these truths for Lessing is the doctrine of the Trinity (}73). Asalready noted, Lessing saw this as expressing a deep speculative truth about the unity ofGod, albeit one that will only find its full development in the Eternal Gospel. Despite theprominence he gives to the prototype envisaged as the son of God, Kant only treats thedoctrine of the Trinity in passing and, as one might expect, he construes it in purely moralrather than in speculative terms. Simply put, Kant suggests that our moral relation to Godrequires the assignment of three distinct functions to him, which may be symbolized bydistinguishing among three personalities: lawgiver, ruler or executive power, and judge.74

The second doctrine towhich Lessing refers is original sin, with regard to which he asksrhetorically: “What if everything should finally convince us that man at the first and loweststage of his humanity, is quite simply not sufficiently in control of his actions to be able tofollow moral laws?”(}74). As the passage indicates, by original sin Lessing understandsnothing more than an original incapacity, which is supposedly overcome through aprocess of intellectual development. This reflects the Leibnizian intellectualistic andperfectionist ethic and its associated determinism;whereas for Kant radical evil (his versionof original sin) is grounded in a misuse of freedom rather than a mere incapacity or lack ofperfection.75 As a determinist, Lessing obviously has no place for such a conception.

73 Kant’s clearest statements of his views on this topic are contained in the first part of TP 8: 278–89.74 RGM 6: 139–42; and SF 7: 38–9.75 Lessing gives an indication of his intellectualist and perfectionist ethic in the final paragraphs of “The

Christianity of Reason,” Lessing (1956), 7: 200; (2005), 28–9.

272 PART IV

The final doctrine that Lessing considers is that of the vicarious satisfaction and hisaccount is of a piece with his treatment of original sin. Once again, he presents it in theform of a rhetorical question:

What if everything should finally compel us to assume that God, despite the original incapacity ofman, nevertheless chose rather to give him moral laws and to forgive him all transgressions inconsideration of his Son—i.e. [ . . . ]the independently existing sum of his own perfections, incomparison with which and in which every imperfection of the individual disappears—than notto give him them and thereby to exclude him from all moral happiness, which is inconceivablewithout moral laws? (}75)

On this view, what is forgiven in the name of the Son, who is himself conceived asGod’s complete (and existent) representation of himself, are simply man’s unavoidableimperfections rather than anything for which he is actually responsible. Accordingly,“satisfaction” in this sense is morally non-problematic. By contrast, for Kant a vicarioussatisfaction or forgiveness of sin is deeply problematic, since it is an undeservedcancelling of the debt of sin, which nevertheless must be accepted as a condition ofhope in being well pleasing to God, even though it remains inexplicable by human(practical) reason.76

Moreover, it is with respect to their attitudes toward the closely intertwinedChristian conceptions of sin and forgiveness that we find the deepest differencebetween the religious-theological views of Lessing and Kant. In addition to his crypticaccount of the core Christian conceptions in Education, Lessing gestured at a defense ofthe orthodox doctrine of eternal punishment, which was actually a defense of Leibniz’sdefense of this doctrine against a critique by Eberhard, but this was more an exercisein philosophical esotericism than a serious endeavor to find a serious moral content inthe doctrine.77 By contrast, Kant was committed to the more ambitious project offinding a legitimate place for these core conceptions in a “pure moral religion” or,what amounts to the same thing, a “religion within the boundaries of mere reason.”Whether he was successful in this endeavor is another question, which cannot be dealtwith here.

76 Together with the call and election, Kant refers to the doctrine of satisfaction as a “mystery” (RGV: 6:143). I have added the qualifier “practical” because the problem with this doctrine for Kant lies in its apparentconflict with the clear dictates of morality rather than with its theoretical inscrutability.

77 Lessing’s “defense,” which is contained in his essay “Leibniz on Eternal Punishments,” is in response toEberhard’s critique of Leibniz’s treatment of the doctrine of eternal punishment in the Theodicy. See Lessing(1956), 7: 454–88; (2005), 30–60; and for my analysis Allison (1966), 86–8.


Bibliography

Al-Azm, Sadik J. (1972) The Origins of Kant’s Argument in the Antinomies, Oxford: OxfordUniversity Press.

Albrecht, Michael and Hinske, Nobert (eds.) (1973) Was ist Aufklärung? Beiträge aus denBerlinischen Monatsschrift, Darmstadt: Wissenschaftliche Buchgesellschaft.

Allison, Henry E. (1966) Lessing and the Enlightenment: A Study of G. E. Lessing’s Philosophy ofReligion and its Place within the Context of 18th Century Thought, Ann Arbor: University ofMichigan Press.

—— (1973) The Kant–Eberhard Controversy, Baltimore and London: Johns Hopkins UniversityPress.

—— (1975) “Lessing’s Conception of Revelation as Education,” in Studies in Eighteenth CenturyCulture, ed. Harold Pagliaro, Madison: University of Wisconsin Press, 183–93.

—— (1976) “The Non-Spatiality of Things in Themselves for Kant,” Journal of the History ofPhilosophy, 14, 313–21.

—— (1980) “Kant’s Critique of Spinoza,” in Studies in Philosophy and the History of Philosophy:The Philosophy of Baruch Spinoza, ed. Richard Kennington, Washington D.C.: CatholicUniversity of America Press, 99–227.

—— (1983) Kant’s Transcendental Idealism: An Interpretation and Defense, New Haven: YaleUniversity Press.

—— (1986) “Morality and Freedom: Kant’s Reciprocity Thesis,” Philosophical Review, 95,393–425.

—— (1990) Kant’s Theory of Freedom, Cambridge: Cambridge University Press.—— (1991a) “Kant’s Antinomy of Teleological Judgment,” Southern Journal of Philosophy, 30Supplement, The Spindel Conference, 25–42.

—— (1991b) “Kant’s Doctrine of Radical Evil,” in Akten des Siebenten InternationalenKant-Kongresses, ed. G. Funke, Bonn: Bouvier, 51–72.

—— (1996a) Idealism and Freedom: Essays on Kant’s Theoretical and Practical Philosophy, Cambridge:Cambridge University Press.

—— (1996b) “The Gulf between Nature and Freedom and Nature’s Guarantee of PerpetualPeace,” in Proceedings of the Eighth International Kant Congress Memphis 1995, ed. HokeRobinson, Milwaukee: Marquette University Press, 37–49.

—— (1996c) Review: Kant’s Compatibilism by Hud Hudson, The Philosophical Review, 105,125–7.

—— (1997a) “We Can Act Only Under the Idea of Freedom,” Proceedings and Addresses of theAmerican Philosophical Association, 71, 39–50.

—— (1997b) “Beauty and Duty in Kant’s Critique of Judgment,” Kantian Review, 1, 53–81.—— (1998a) “Beauty as Mediator between Nature and Freedom,” in Philosophie in synthetischer

Absicht, ed. Marcelo Stamm, Stuttgart: Klett-Cotta, 539–64.—— (1998b) “Commentary on Section Nine of the Antinomy of Pure Reason,” in Kritik der

reinen Vernunft: Ein kooperativer Kommentar, ed. M. Willaschek and J. Mohr, Frankfurt: VittorioKlostermann, 465–90.

—— (2000a) “Kant’s Conception of Aufklärung,” in The Proceedings of the Twentieth WorldCongress of Philosophy, Modern Philosophy, ed. Mark D. Gedney, vol. 7, Philosophy Documen-tation Center, Bowling Green State University, 35–44.

—— (2000b) “Where Have all the Categories Gone? Reflections on Longuenesse’s Reading ofKant’s Transcendental Deduction,” Inquiry, 43, 67–80.

—— (2000c) “Is the Critique of Judgment ‘Post-Critical’?” in The Reception of Kant’s CriticalPhilosophy: Fichte, Schelling, & Hegel, ed. Sally Sedgwick, Cambridge: Cambridge UniversityPress, 78–92.

—— (2001a) Kant’s Theory of Taste: A Reading of the Critique of Aesthetic Judgment, Cambridge:Cambridge University Press.

—— (2001b) “The Critique of Judgment as a ‘True Apology’ for Leibniz,” in Kant unddie Berliner Aufklärung, Akten des IX. Internationalen Kant-Kongresses, ed. V. Gerhardt,R.-P. Horstmann, and R. Schumacher, vol. 1, Berlin and New York: Walter de Gruyter,286–99.

—— (2001c) “Ethics, Evil, and Anthropology in Kant: Remarks on Allen Wood’s Kant’s EthicalThought,” Ethics, 111, 594–613.

—— (2002) “On the Very Idea of a Propensity to Evil,” Journal of Value Inquiry, 36, 337–48.—— (2003) “Reflective Judgement and the Application of Logic to Nature: Kant’s Deductionof the Principle of Purposiveness as an Answer to Hume,” in Strawson and Kant, ed. Hans-Johann Glock, Oxford: Oxford University Press, 169–83.

—— (2004) Kant’s Transcendental Idealism, revised and enlarged edition, New Haven: YaleUniversity Press.

—— (2005) “Kant and the Two Dogmas of Rationalism,” in A Companion to Rationalism, ed.Alan Nelson, Oxford: Blackwell Publishing, 343–59.

—— (2006a) “Transcendental Realism, Empirical Realism and Transcendental Idealism,”Kantian Review, 11, 1–28.

—— (2006b) “Kant on Freedom of the Will,” in The Cambridge Companion to Kant and ModernPhilosophy, ed. Paul Guyer, Cambridge: Cambridge University Press, 381–415.

—— (2008) Custom and Reason in Hume: A Kantian Reading of the First Book of Hume’s Treatise,Oxford: Oxford University Press.

—— (2009a) “Teleology and History in Kant: The Critical Foundations of Kant’s Philosophy ofHistory,” in Kant’s Idea for a Universal History with a Cosmopolitan Aim: A Critical Guide, ed.Amelie Oksenberg Rorty and James Schmidt, Cambridge: Cambridge University Press,24–45.

—— (2009b) “Reason, Revelation, and History in Lessing and Kant,” Internationales Jahrbuch desDeutschen Idealismus, 35–57.

—— (2011) Kant’s Groundwork for the Metaphysics of Morals: A Commentary, Oxford: OxfordUniversity Press.

—— (forthcoming) “Kant’s Practical Justification of Freedom,” in Kant on Practical Justification,ed. Sorin Baiasu and Mark Timmons, Oxford: Oxford University Press.

Altmann, Alexander (1973) Moses Mendelssohn: A Biographical Study, Alabama: University ofAlabama Press.

—— (1982) Die trostvolle Aufklärung: Studien zur Metaphysik und politische Theorie Moses Mendels-sohns, Stuttgart-Bad Cannstatt: Frommann-Holzboog, 276–85.

BIBL IOGRAPHY 275

Ameriks, Karl (1978) “Kant’s Transcendental Deduction as a Regressive Argument,” Kant-Studien, 69, 273–87.

—— (1982) Kant’s Theory of Mind: An Analysis of the Paralogisms of Pure Reason, Oxford:Clarendon Press.

—— (1992) “Kantian Idealism Today,” History of Philosophy Quarterly, 9, 329–42.—— (2000) Kant and the Fate of Autonomy, Cambridge: Cambridge University Press.Anderson-Gold, Sharon (1984) “Kant’s Rejection of Devilishness: The Limits of HumanVolition,” Idealistic Studies, 14, 35–48.

Aner, Karl (1964) Die Theologie der Lessingzeit, Hildesheim: Georg Olms Verlagsbuchhandlung.Augustine, Saint (1948) Soliloquies, Writings of St. Augustine, vol. 1, trans. T. F. Gilligan, NewYork: Cima Publishing Co.

Baker, Judith (1988) “Counting Categorical Imperatives,” Kant-Studien, 79, 389–406.Baum, Manfred (1990) “Kants Prinzip der Zweckmässigkeit and Hegels Realisierung desBegriffs,” in Hegel und die Kritik der Urteilskraft, ed. Hans Friedrich Fulda and Rolf-PeterHorstmann, Stuttgart: Klett-Cotta, 158–73.

Beck, Lewis White (1960) A Commentary on Kant’s Critique of Practical Reason, Chicago: Univer-sity of Chicago Press.

—— (ed.) (1963) Kant on History, Indianapolis: Bobbs-Merrill.—— (1965) Studies in the Philosophy of Kant, Indianapolis: Bobbs-Merrill.—— (1967) “Can Kant’s Synthetic Judgments be Made Analytic?” in Kant: Disputed Questions,ed. Moltke S. Gram, Chicago: Quadrangle Books.

—— (1969) Early German Philosophy, Cambridge, MA: Harvard University Press.—— (1975) The Actor and the Spectator, New Haven: Yale University Press.—— (1987) “Five Concepts of Freedom in Kant,” in Philosophical Analysis and Reconstruction:

Festschrift to Stephan Korner, ed. J. T. J. Srzednicki, Dordrecht: Martinus Nijhoff, 35–51.Beyerhaus, Gisbert (1921) “Kants ‘Program’ der Aufklärung aus dem Jahre 1784,” Kant-Studien,26, 1–15.

Bigg, Charles (1866) The Christian Platonists of Alexandria, Oxford: Clarendon Press.Bird, Graham (1996) “McDowell’s Kant: Mind and World,” Philosophy, 71, 219–43.Bittner, Rüdiger (1989) “Das Unternehmen einer Grundlegung zur Metaphysik der Sitten,” in

Grundlegung zur Metaphysik der Sitten, Ein kooperativer Kommentar, ed. Ottfried Hoffe, Frankfurtam Main: Klostermann, 13–30.

—— (1996) “What is Enlightenment?” in What is Enlightenment? Eighteenth-Century Answers andTwentieth-Century Questions, ed. James Schmidt, Berkeley: University of California Press.

Bohatec, Josef (1966) Die Religionsphilosophie Kants in der ‘Religion innerhalb der Grenzen der blossenVernunft’, Hildesheim: Georg Olms Verlagsbuchhandlung.

Buchdahl, Gerd (1969a) Metaphysics and the Philosophy of Science, Cambridge, MA: MIT Press.—— (1969b) “The Kantian ‘Dynamic of Reason’ with Special Reference to the Place ofCausality in Kant’s System,” in Kant Studies Today, ed. L. W. Beck, La Salle, IL: OpenCourt, 187–208.

—— (1974) “The Conception of Lawlikeness in Kant’s Philosophy of Science,” inKant’s Theoryof Knowledge, ed. L. W. Beck, Dordrecht: Reidel, 128–50.

Butts, Robert E. (1984) Kant and the Double Government Methodology, Dordrecht: Reidel.Collins, James (1967) The Emergence of Philosophy of Religion, New Haven & London: YaleUniversity Press.

276 BIBL IOGRAPHY

Crusius, Christian August (1964) Entwurf der nothwendigen Vernunft = Wahrheiten, wiefern sie denzufälligen entgegen gestetzt (1745); reprinted in Christian August Crusius Die philosphischenHauptwerke, vol. II, ed. Gorgio Tonelli, Hildesheim: Georg Olms Verlagsbuchhandlung,3–344.

—— (1969) Anweisung vernünftig zu leben (1744). Reprinted in Christian August Crusius Diephilosphischen Hauptwerke, vol. I, ed. Gorgio Tonelli, Hildesheim: Georg Olms Verlagsbuch-handlung.

Davis, Kevin R. (1973) “Kant’s Different ‘Publics,’ ” Kant-Studien 83, 170–84.Dennett, Daniel (1973) “Mechanism and Responsibility,” in Essays on Freedom of Action, ed. TedHonderich, London: Routledge & Kegan Paul, 157–84.

—— (1984) Elbow Room: The Varieties of Free Will Worth Wanting, Cambridge, MA: MIT Press.—— (1987) The Intentional Stance, Cambridge, MA: MIT Press.Despland, Michel (1973) Kant on History and Religion, Montreal and London: McGill-Queen’sUniversity Press.

Duncan, A. R. C. (1957) Practical Reason and Morality: A Study of Immanuel Kant’s Foundations forthe Metaphysics of Morals, London: Thomas Nelson and Sons.

Düsing, Klaus (1968) Die Teleologie in Kants Weltbegriff, Kant-Studien Ergänzungshefte 96, Bonn:H. Bouvier.

—— (1971) “Das Problem des hochsten Gutes in Kants praktischer Philosophie,” Kant-Studien,62, 32–3.

—— (1990) “Naturteleologie and Metaphysik bei Kant and Hegel,” in Hegel und die Kritikder Urteilskraft, ed. Hans Friedrich Fulda and Rolf-Peter Horstmann, Stuttgart: Klett-Cotta,141–57.

Eberhard, J. A. (1968) Philosophisches Magazin, Erster Band, Halle: Johann Jacob Gebauer, 1789;reprinted in Aetas Kantiana, Bruxelles: Culture et Civilisation.

Ewing, A. C. (1924) Kant’s Treatment of Causality, London: Kegan Paul, Trench, Trubner & Co.Fackenheim, Emil (1956–7) “Kant’s Concept of History,” Kant-Studien, 48, 381–98.Falkenstein, Lorne (1995) Kant’s Intuitionism: A Commentary on the Transcendental Aesthetic,Toronto: University of Toronto Press.

Fichte, J. G. (1988) Early Philosophical Writings, trans. and ed. Daniel Breazeale, Ithaca andLondon: Cornell University Press.

—— (1994) Introductions to the Wissenschaftslehre and Other Writings, trans. and ed. DanielBreazeale, Indianapolis and Cambridge: Hackett Publishing Company.

—— (2000) Foundations of Natural Right, trans. Michael Bauer, ed. Frederick Neuhouser,Cambridge: Cambridge University Press.

—— (2005) The System of Ethics, trans. and ed. Daniel Breazeale and Günter Zoller, Cambridge:Cambridge University Press.

Firestone, Chris L. and Jacobs, Nathan (2008) In Defense of Kant’s Religion, Bloomington &Indianapolis: Indiana University Press, 155–80.

Floyd, Juliet (1998) “Heautonomy: Kant on Reflective Judgment and Systematicity,” in Kant’sÄsthetik, Kant’s Aesthetics, L’esthetique de Kant, ed. Herman Parret, Berlin and New York:Walter de Gruyter, 192–218.

Freudiger, Jürg (1993) Kants Begründung der praktischen Philosophie. Bern, Stuttgart and Wien:Verlag Paul Haupt.

BIBL IOGRAPHY 277

Friedman, Michael (1992) “Causal Laws and the Foundations of Natural Science,” in TheCambridge Companion to Kant, ed. Paul Guyer, Cambridge: Cambridge University Press,161–97.

—— (1995) “Exorcising the Philosophical Tradition: Comments on JohnMcDowell’sMind andWorld,” Philosophical Review, 105, 427–67.

Gadamer, Hans (1999) Truth and Method, 2nd revised edition, trans. Joel Weinsheimer andDonald G. Marshall, New York: Continuum Publishing.

Garve, Christian (1783) Philosophische Anmerkungen und Abhandlungen zu Ciceros Buchern von denPflichten, Breslau: Wilhelm Gottlieb Korn.

Ginsborg, Hannah (1990) The Role of Taste in Kant’s Theory of Cognition, New York and London:Garland Publishing.

Grier, Michelle (2001) Kant’s Doctrine of Transcendental Illusion, Cambridge: Cambridge Univer-sity Press.

Guyer, Paul (1979) Kant and the Claims of Taste, Cambridge, MA: Harvard University Press.—— (1983) “Kant’s Intentions in the Refutation of Idealism,” Philosophical Review, 92, 329–83.—— (1987) Kant and the Claims of Knowledge, Cambridge: Cambridge University Press.—— (1990a) “Reason and Reflective Judgment: Kant on the Significance of Systematicity,”

Noûs, 24, 17–43.—— (1990b) “Kant’s Conception of Empirical Law,” Proceedings of the Aristotelian Society, supp.vol. 64, 220–42.

—— (1993) Kant and the Experience of Freedom: Essays on Aesthetics and Morality, Cambridge:Cambridge University Press.

—— (2000) Kant on Freedom, Law, and Happiness, Cambridge: Cambridge University Press.Hamann, J. G. (1965) Briefwechsel, vol. 5 (1783–5), ed. Arthur Henkel, Frankfurt: Insel Verlag.Harnack, Adolph (1961) History of Christian Dogma, vol. 2, trans. Neill Buchanan, New York:Dover.

Hegel, G. W. F. (1959) Enzyklopädie der philosophischen Wissenschaften im Grundrisse (1830), ed.Friedhelm Nicolin and Otto Poggler, Hamburg: Felix Meiner Verlag.

—— (1968) Glauben and Wissen, Gesammelte Werke, vol. IV, ed. Hartmut Buchner and OttoPoggler, Hamburg: Felix Meiner Verlag.

—— (1969) Science of Logic, trans. A. V. Miller, London: George Allen & Unwin.—— (1971)Vorlesungen über die Geschichte der Philosophie III, Frankfurt am Main: SuhrkampVerlag.

—— (1977) Phenomenology of Spirit, trans. A. V. Miller, Oxford: Oxford University Press.—— (1991) Elements of the Philosophy of Right, trans. H. B. Nisbet, ed. Allen Wood, Cambridge:Cambridge University Press.

Heimsoeth,Heinz (1967) “Freiheit undCharakter nach denKantReflexionenNr. 5611–5620,” inTradition und Kritik: Festschrift für Rudolf Zocher zum 80 Geburtstag, ed. Wilhelm Arnold andHermann Zeltner, Stuttgart: Friedrich Fromann, 123–44.

Hill, Thomas (1985) “Kant on theRationality ofMoral Conduct,” Pacific Philosophical Quarterly, 66.—— (1992) Dignity and Practical Reason in Kant’s Moral Theory, Ithaca and London: CornellUniversity Press.

Horstmann, Rolf-Peter (1989) “Why Must there be a Transcendental Deduction in Kant’sCritique of Judgment?” in Kant’s Transcendental Deductions, The Three ‘Critiques’ and the ‘Opuspostumum’, ed. Eckart Forster, Stanford: Stanford University Press, 157–76.

278 BIBL IOGRAPHY

Howell, Robert (2004) “The Conundrum of the Object and Other Problems from Kant,”Kantian Review, 8, 115–36.

Hudson, Hud (1994) Kant’s Compatibilism, Ithaca and London: Cornell University Press.Hume, David (1999) Enquiry Concerning the Human Understanding, ed. Tom L. Beauchamp,Oxford: Oxford University Press.

—— (2000) A Treatise of Human Nature, ed. David Fate Norton and Mary J. Norton, Oxford:Oxford University Press.

Irwin, Terence (1984) “Morality and Personality: Kant and Green,” in Self and Nature in Kant’sThought, ed. Allen Wood, Ithaca and London: Cornell University Press, 31–56.

Jolley, Nicholas (1999) Locke: His Philosophical Thought, Oxford and New York: Oxford Univer-sity Press.

Kaehler, Klaus (1991) “Comment on Henry E. Allison: Kant’s Antinomy of TeleologicalJudgment,” The Southern Journal of Philosophy, 30 Supplement, 43–8.

Kitcher, Philip (1981) “How Kant almost wrote ‘Two Dogmas of Empiricism’ (and why hedidn’t),” Philosophical Topics, 12, 217–49.

Korner, Stephan (1987) “Reply to Professor Beck,” in Philosophical Analysis and Reconstruction:Festschrift to Stephan Korner, ed. J. T. J. Srzednicki, Dordrecht: Martinus Nijhoff, 52–8.

Korsgaard, Christine (1996) Creating the Kingdom of Ends, Cambridge: Cambridge UniversityPress, 3–23.

Kuypers, K. (1972) Kants Kunsttheorie und die Einheit der Kritik der Urteilskraft, Amsterdam andLondon: North-Holland Publishing Company.

Langton, Rae (1998) Kantian Humility: Our Ignorance of Things in Themselves, Oxford: OxfordUniversity Press.

Leibniz, G.W. (1951) Theodicy, Essays on the Goodness of God, the Freedom of Man, and the Origin ofEvil, trans. E. M. Huggard, London: Routledge & Kegan Paul.

—— (1956) The Leibniz–Clarke Correspondence, ed. H. G. Alexander, Manchester: ManchesterUniversity Press.

—— (1981) New Essays on Human Understanding, trans. and ed. Peter Remnant and JonathanBennett, Cambridge: Cambridge University Press.

—— (1989) Philosophical Essays, trans. and ed. Roger Ariew and Daniel Garber, Indianapolis:Hackett.

Lessing, Gotthold Ephraim (1956) Gesammelte Werke, 10 vols., ed. Paul Rilla, Berlin: Aufbau-Verlag.

—— (2005) Lessing: Philosophical and Theological Writings, trans and ed. H. B. Nisbet, Cambridge:Cambridge University Press.

Locke, John (1965) On the Reasonableness of Christianity as Delivered in the Scriptures, ed. GeorgeW. Ewing, Chicago: Henry Regnery Company.

—— (1975) An Essay concerning Human Understanding, ed. Peter H. Nidditch, Oxford: Clar-endon Press.

Longuenesse, Beatrice (1998) Kant and the Capacity to Judge, trans. Charles T. Wolfe, Princeton:Princeton University Press.

—— (2000) “Kant’s Categories and the Capacity to Judge: Responses to Henry Allison and SallySedgwick,” Inquiry, 43, 91–110.

McDowell, John (1994) Mind and World, Cambridge MA: Harvard University Press.McFarland, J. D. (1970) Kant’s Concept of Teleology, Edinburgh: Edinburgh University Press.

BIBL IOGRAPHY 279

McLaughlin, Peter (1989) “What is an Antinomy of Judgment?” in Proceedings of the Sixth Interna-tional Kant Congress, vol. ii/2, ed. Gerhard Funke and Thomas M. Seebohm,Washington, D.C.:Center for Advanced Research in Phenomenology & University Press of America, 357–67.

—— (1990) Kant’s Critique of Teleology in Biological Explanation Antinomy and Teleology, Lewiston,NY: Edwin Mellen Press.

—— (2003) “Newtonian Biology and Kant’s Mechanistic Concept of Causality,” in Kant’sCritique of the Power of Judgment: Critical Essays, ed. Paul Guyer, Lanham, Boulder, New York,and Oxford: Rowman & Littlefield Publishers, 209–17.

Marc-Wogau, Konrad (1938) Vier Studien zu Kants Kritik der Urteilskraft, Uppsala:A. B. Lundequistka Bokhandeln.

Martin, Gottfried (1955) Kant’s Metaphysics and Theory of Science, trans. Peter Lucas, Manchester:Manchester University Press.

Mendelssohn, Moses (1973) “Über die Frage: was heisst aufklären?” in Was ist Aufklärung?,Beiträge aus den Berlinischen Monatsschrift, ed. Michael Albrecht and Nobert Hinske, Darmstadt:Wissenschaftliche Buchgesellschaft, 444–51.

Mertens, Helga (1975) Kommentar zur ersten Einleitung in Kants Kritik der Urteilskraft, Munich:Johannes Berchman Verlag.

Michalson, Gordon E. Jr. (1985) Lessing’s “Ugly Ditch”: A Study of Theology and History,University Park, PA and London: Pennsylvania State University Press.

—— (1990) Fallen Freedom, Kant on Radical Evil and Moral Regeneration, Cambridge: CambridgeUniversity Press.

Moran, Dermot (2000) “Hilary Putnam and Immanuel Kant: Two ‘Internal Realists’?” Synthese,123, 65–104.

Nell (O’Neill), Onora (1975) Acting on Principle: An Essay on Kantian Ethics, New York andLondon: Columbia University Press.

Neuhouser, Frederick (1990) Fichte’s Theory of Subjectivity, Cambridge: Cambridge UniversityPress.

—— (2000) Foundations of Hegel’s Social Theory: Actualizing Freedom, Cambridge, MA: HarvardUniversity Press.

O’Neill, Onora (1989) Constructions of Reason, Cambridge: Cambridge University Press.Paton, H. J. (1958) The Categorical Imperative, London: Hutchinson & Co.Philonenko, Alexis (1988) “L’antinomie teleologique chez Kant,” Revue de Metaphysique et de

Morale, 82, 13–37.Pinder, Tillman (1986) “Kant’s Begriff der transzendentalen Erkenntnis,” Kant-Studien, 77, 1–40.Pippin, Robert (1995) “Avoiding German Idealism: Kant and the Reflective JudgmentProblem,” in Proceedings of the Eighth International Kant Congress, Memphis 1995, ed. HokeRobinson, Milwaukee: Marquette University Press, 977–97.

Prauss, Gerold (1974) Kant und das Problem der Dinge an sich, Bonn: Bouvier.Prichard, H. A. (1909) Kant’s Theory of Knowledge, Oxford: Clarendon Press.Quine, W. V. O. (1963) From a Logical Point of View, New York: Harper & Row.Quinn, Philip (1988) “In Adam’s Fall, We Sinned All,” Philosophical Topics, 16, 89–118.Reich, Klaus (1939) “Kant and Greek Ethics (II),” trans. W. H. Walsh, Mind, 48, 446–63.Rousseau, Jean-Jacques (1964) “Discourse on the Origin and Foundations of Inequality AmongMen,” in The First and Second Discourses, trans. Roger D. Masters and Judith R. Masters, ed.Roger D. Masters, New York: St Martin’s Press, 101–228.

280 BIBL IOGRAPHY

Schmidt, James (1989) “The Question of Enlightenment: Kant, Mendelssohn, and the Mittel-wochgesellschaft,” Journal of the History of Ideas, 50, 269–91.

—— (1992) “What Enlightenment Was: How Moses Mendelssohn and Immanuel KantAnswered the Berlinische Monatschrift,” Journal of the History of Philosophy, 30, 77–101.

—— (ed.) (1996) What is Enlightenment? Eighteenth-Century Answers and Twentieth-CenturyQuestions, Berkeley: University of California Press.

Schneewind, J. B. (1998) The Invention of Autonomy, Cambridge: Cambridge University Press.Schonecker, Dieter and Wood, Allen (2004) Kants “Grundlegung zur Metaphysik der Sitten,” Ein

einführender Kommentar, 2. Auflage, Paderborn, München, Wien, and Zürich: Verlag Ferdi-nand Schoningh.

Schopenhauer, Arthur (1958) The World as Will and Representation, vol. 1, trans. E. F. J. Payne,Indian Hills, CO: The Falcon’s Wing Press.

—— (1960) Essay on the Freedom of the Will, trans. Konstantin Kolenda, Indianapolis and NewYork: Bobbs-Merrill Company.

—— (1965) On the Basis of Morality, trans. E. F. J. Payne, Indianapolis and New York: Bobbs-Merrill Company.

—— (1974) The Fourfold Root of the Principle of Sufficient Reason, trans. E. F. J. Payne, LaSalle, IL:Open Court Publishing.

Schulz, E. G. (1974) “Kant und die Berliner Aufklärung,” Akten des 4 Internationalen Kant-Kongresses,Mainz 1974, ed. Gerhard Funke, Berlin and New York: Walter de Gruyter, 60–80.

Silber, John (1960) “The Ethical Significance in Kant’s Religion,” in Religion within the Limits ofReason Alone, trans. T. M. Greene and H. H. Hudson, New York: Harper & Row, lxxix–cxxxiv.

Spinoza, Baruch (1982) The Ethics and Selected Letters, trans. Samuel Shirley, Indianapolis: HackettPublishing Company.

Stratton-Lake, Philip (1993) “Formulating Categorical Imperatives,” Kant-Studien, 84, 317–40.Strawson, P. F. (1962) “Freedom and Resentment,” Proceedings of the British Academy, 48, 1–25.—— (1966) The Bounds of Sense: An Essay on Kant’s Critique of Pure Reason, London: Methuen.Stroud, Barry (1994) “Kantian Argument, Conceptual Capacities and Vulnerability,” in Kant and

Contemporary Epistemology, ed. Paolo Parrini, Dordrecht, Boston, and London: Kluwer Aca-demic Publishers, 231–51.

—— (1996) “The Charm of Naturalism,” Proceedings and Addresses of the American PhilosophicalAssociation, 70, 43–55.

Thompson, Kevin (1995) “The Antinomy of Teleological Judgment and the Concept of anIntuitive Intellect: Transformation and Conflict,” Proceedings of the Eighth International KantCongress, vol. 2, Milwaukee: Marquette University Press, 445–52.

Timmermann, Jens (2007) Kant’s Groundwork of the Metaphysics of Morals: A Commentary,Cambridge: Cambridge University Press.

Tonelli, Giorgio (1957–8) “Von den verschiedenen Bedeutungen des Wortes Zweckmässigkeitin der Kritik der Urteilskraft,” Kant-Studien, 49, 154–66.

Troeltsche, Ernst (1904) “Das Historische in Kants Religionsphilosophie,” Kant-Studien, 9,21–154.

Turbayne, Colin (1955) “Kant’s Refutation of Dogmatic Idealism,” Philosophical Quarterly, 5,225–44.

BIBL IOGRAPHY 281

Tuschling, Burkhard (1990) “Intuitiver Verstand, absolute Identität, Idee. Thesen zu Hegelsfrüher Rezeption der Kritik der Urteilskraft,” in Hegel und die Kritik der Urteilskraft, ed. HansFriedrich Fulda and Rolf-Peter Horstmann, Stuttgart: Klett-Cotta, 174–88.

—— (1992) “The System of Transcendental Idealism: Questions Raised and Left Open in theKritik der Urteilskraft,” System and Teleology in Kant’s Critique of Judgment (Spindel Conference1991), 30, Supplement, The Southern Journal of Philosophy, ed. Hoke Robinson, 109–27.

Vaihinger, Hans (1892) Commentar zu Kants Kritik der reinen Vernunft, zweiter Band, Stuttgart:W. Spemann.

—— (1935) The Philosophy of ‘As If ’, trans. C. K. Ogden, 2nd edition, London: Kegan Paul,Trench, Trubner & Co.

Van Cleve, James (1999) Problems from Kant, Oxford: Oxford University Press.Weil, Eric (1963) Problemes kantiens, Paris: J. Vrin.Wessel, Leonard (1977) G. E. Lessing’s Theology: A Reinterpretation, The Hague: Mouton.Weyand, Klaus (1964) Kants Geschichtsphilosophie, Ihre Entwicklung und ihr Verhältnis zur Aufklär-

ung, Kant-Studien Ergänzungshefte 85, Koln: Kolner Universitäts-Verlag.Willaschek, Marcus (1992) Praktische Vernunft, Handlungstheorie und Moralbegründung bei Kant,Stuttgart: J. B. Metzler.

Wolff, Christian, Feiherr von (1997) Vernünfftige Gedancken von Gott, der Welt und der Seele desMenschen, Auch allen Dingen überhaupt, 5th edition (1781), reprinted by Georg Olms.

Wolff, Robert Paul (1973) The Autonomy of Reason: A Commentary on Kant’s Groundwork of theMetaphysic of Morals, New York: Harper & Row.

Wood, Allen (1970) Kant’s Moral Religion, Ithaca, NY: Cornell University Press.—— (1990) Hegel’s Ethical Thought, Cambridge: Cambridge University Press.—— (1991a) “Kant’s Deism,” in Kant’s Philosophy of Religion Reconsidered, ed. Philip J. Rossi andMichael Wreen, Bloomington and Indianapolis: Indiana University Press.

—— (1991b) “Unsociable Sociability: The Anthropological Basis of Kantian Ethics,” Philosoph-ical Topics, 19, 325–51.

—— (1999) Kant’s Ethical Thought, Cambridge: Cambridge University Press.—— (2007)“Debating Allison on Transcendental Idealism,” Kantian Review, 12, 1–10.Yovel, Yirmiahu (1980) Kant and the Philosophy of History, Princeton: Princeton University Press.Zammito, John H. (1992) The Genesis of Kant’s Critique of Judgment, Chicago and London:University of Chicago Press.

Zumbach, Clark. (1984) The Transcendent Science: Kant’s Conception of Biological Methodology, TheHague: Martinus Nijhoff.

282 BIBL IOGRAPHY

Index

affection 39agency see rational agencyAlbrecht, Michael 229nAltmann, Alexander 229nAmeriks, Karl 69n, 83n, 116n, 120–1Analogies of Experience 169–70, 201nsecond analogy 148, 151, 183, 202–4

analytic judgment 50–8see also the predicate-in-notion principle

Analytic of the Teleological Power ofJudgment 174–6

animality, predisposition to 101anthropology 118, 219, 261and evil 99, 103–4, 107–8

Antinomy of Pure Reason 15–30, 78, 120–3first antinomy 16, 18–19, 78–1, 122second antinomy 16, 19–21third antinomy 21–8, 122, 144, 147–51,186, 238

fourth antinomy 28–30mathematical vs. dynamic 17, 82, 122

Antinomy of the Teleological Power ofJudgment 175, 201–14

on mechanism 201–4, 210, 212–14and the supersensible 213–14teleology in the 208–14

appearances 21as contrasted with things in themselves 64, 68,74, 95–6, 152–3

apperception 91see also unity of apperception

association (Humean) 34, 54, 114autonomy see freedom

Baker, Judith 128nBaum, Manfred 165n, 175Baumgarten, A. G. 142beauty 250Beck, Lewis White 54n, 58n, 73n, 81, 116–17,

121, 124n, 137n, 139n, 140n, 141n,151n, 230n

Berkeley, George 69Beyerhous, Gisbert 229nBigg, Charles 257nbiology 175, 206, 212–13, 242–4Bird, Graham 95nBittner, Rüdiger 126n, 231Bohatec, Joseph 269–70nBreazeale, Daniel 152nButts, Robert E. 201n 206n

capacity to judge (Vermogen zu urteilen) 33, 37–8,41–2, 44

categorical imperative, theconstruction of the complete concept of128–30, 134

distinct formulations of 124n, 127–36(FA) formula of autonomy 129, 131–3(FH) formula of humanity 128–9, 131–3,135, 247–8

(FLN) formula of the law of nature 127, 128,130–3, 134–5

(FRE) formula of a realm of ends 124n, 130n(FUL) formula of universal law 124n, 127n,130, 132–3, 134–5

and freedom 112–16and rational agency 127–31the singleness of 124–6, 132–6(UF) universal formula 135nand universalizability 130–5

categories (pure concepts of understanding)appearances subsumed under 32, 44–5as applied in every judgment 32–6, 45–6as concepts of an object in general 32, 42,45, 73

as logical forms 33and (human) sensibility 37–42transcendental use of 73unschematized (pure) 35, 35n, 80n

causality see Analogies of Experience, secondanalogy; free will problem; mechanism

characterempirical 219–20empirical vs. intelligible 24–7, 108,149, 160

Cicero 126, 128Clarke, Samuel 60Clement of Alexandria 257community, ethical (church) 222–3, 253,

264–7, 269compatibilism 87, 105, 121, 137, 139, 148,

151, 271conceptsand cognition see discursivity thesisand comparison 180completely determined 50, 51–2, 54intension and extension 58, 62, 180, 195and judgment 42, 54–6, 179–80, 195of objects in general see objectsand the reflective power of judgment 33, 45,172–3, 178–9, 193–6, 214

conditions and the unconditioned 15–16,79–80, 147–8

contradictionprinciple of 138, 139–140practical 131

Copernican Revolution 52, 56and idealism 64, 166see also critical philosophy; transcendentalidealism

critical philosophy 1, 78, 95–6, 165–7and dogmatism 18–19, 52, 199–200, 208–9

Crusius, C. A. 7, 137, 140–5, 146, 150, 159Culture 221, 224–5, 230, 249–50

Davis, Kevin R. 231nDialectic of the Teleological Power of

Judgment 176, 185, 202, 204Dennett, Daniel 5, 88–90, 92Descartes, Rene 143, 160, 175n, 208design 207–11

see also purposivenessDespland, Michel 239n, 266n, 267n, 268ndeterminism 25, 81–2, 88, 121, 137, 144, 146,

147–51, 158–9, 271, 272discursivity see concepts; spontaneity;

understandingdiscursivity thesis 47, 52, 55–6, 209and analytic judgment 58and space 62–3see also reducibility principle

Dostoevsky, Fyodor 10, 231Dummett, Michael 123Duncan, A. R. C. 127–9Düsing, Klaus 165n, 214n, 220nduty 104–5, 118, 228

Eberhard, J. E. 58, 189, 199, 254, 273education (Bildung) 94, 97–8, 226, 230, 258empiricism 52, 55ends 129–30in themselves 131–2see also purposiveness

Enlightenment, the 99, 254, 258, 262, 270nenlightenment 99, 229–35see also Unmündigkeit

ens realissimum 29essencenominal vs. real 173, 190, 192see also natural kinds

ethical rigorism 100–1, 103–4evil 99–109and freedom 104–6, 109and the moral law 101–2radical 99–100, 106–8, 262, 264, 269see also unsocial sociability

Ewing, A. C. 202n, 203n

fact of reason 110–11, 116–20, 147, 186Falkenstein, Lorne 76nFichte, J. G. 7, 152–5, 227final cause 208–14

see also teleologyFloyd, Juliet 188nformin general 38–40of judgment see the logical forms of judgment

form of intuitionas contrasted with formal intuitions 38–9, 46see also forms of sensibility

forms of sensibility 39, 62–3and objects 74in the transcendental deduction 33, 37–40,45–8

and transcendental idealism 64–6, 72, 75–8and the transcendental synthesis of theimagination 39–40, 46–8

see also space, timefreedom:as autonomy 112, 115–16, 118–20, 123, 137,145–7, 232, 271–2

and causal determinism 26, 96, 137, 144,147–51

in Kant’s predecessors 137–42in Kant’s successors 151–61and morality 89–90, 100, 105–6, 110–13,115, 119, 140, 145–7, 160–1, 186, 217

naturalist conceptions of 88–90, 121and nature 22, 97, 115, 217–18practical 111n, 113, 143–4practical justification of 28n, 82, 92, 97,110–23, 186, 198–9

and rational agency 24, 26–8, 90–1, 92, 97,111–13, 116, 119, 123, 144–5

as spontaneity 22–3, 25–6, 91, 113–16, 137,142–4, 148–50

transcendental 23, 29, 111n, 115n,142–3, 186

as a transcendental idea 22, 90, 111–20,143, 150–1

see also rational agency; transcendental idealismFreud, Sigmund 118Freudiger, Jürg 128nfree will problem 21–8, 81–2, 87–8, 96, 105,

149–51, 156–8, 186, 218, 238see also freedom

Friedman, Michael 96, 168n

Gadamer, Hans 10, 235Garve, Christian 126geometry 61Ginsborg, Hanna 172–3, 178–9, 194God 110, 138, 222, 257–73Goethe, Johann, Wolfgang von 270n

284 INDEX

Goeze, J. M. 256nGrace 263–4Grier, Michelle 29n, 79n, 122nGuyer, Paul 8, 75, 83n, 121, 126n, 167n, 171n,

178n, 183n, 217

Hamman, J. G. 231happiness 104–5, 107, 221, 222, 249, 272Harnack, Adolph 257nheautonomy 170, 188, 200, 212Hegel, G. W. F. 7, 8, 94, 155–8, 165–7, 174,

176, 205n, 226, 238, 245Heimsoeth, Heinz 140nhighest good 220–3, 252, 261–2, 272Hill, Thomas Jr. 116n, 127nHinsky, Nobert 229nhistoryphilosophy of 94, 223, 237, 251–3and religion 254, 255–61, 265, 267–9,270, 271

Hobbes, Thomas 208holy will 105Horstmann, Rolf-Peter 8, 167, 169Howell, Robert 76nhumanity, predisposition to 101–2, 106–7Hume, David 8, 60, 140, 177, 182, 184,

186–8, 256

Ideal of Pure Reason 30idealismBerkeleyan 69post-Kantian 39–40, 151–8, 165–7see also transcendental idealism

Ideas 236illusion, transcendental 79–81, 122, 206–7imagination 48

see also transcendental synthesis of theimagination

immortality 110, 255, 257, 259–60, 269,271, 272

imperatives 127nimputation (imputability) 26–8, 99, 101, 102,

108, 145–7, 150incentives 100; see also incorporation thesisincorporation thesis 93, 100, 102, 114, 144–5induction 177, 181–2, 183, 187–8infinite divisibility 20–1infinityin the antinomies 16, 18–21, 79–80and concepts 50–2

inner sense 39, 41intellect see understandingintuitionand cognition 52–4; see also discursivity thesisas immediate and singular representation 53intellectual 53pure 38, 59, 62–3

as essentially sensible (in humans) 53and synthetic judgment 51

Iranaeus 257Irwin, Terence 82I think, the 93

Jesus Christ 263, 264, 269, 271–2Jacobi, F. H. 232, 233Jolly, Nicholas 191nJudaism 266–7, 271judgmentsaesthetic 241analytic see analytic judgmentand apperception 91and the analytic-synthetic distinction 57–9and the categories 32–3, 34, 42form of see the logical forms of judgmentas fundamental unit of thought 54, 56of perception and experience 32–3, 34,45, 179

reflective see reflective power of judgmentand truth 34

justice, divine 263

Kant’s works(A) Anthropology from a Pragmatic Point of

View 94, 230–2, 234(BA) “An Answer to the Question: What isEnlightenment?” 229–32

(EE) First Introduction to the Critique of thePower of Judgment 170–1, 177–8, 182–3,194–6, 210, 251n

(EF) Toward Perpetual Peace: A PhilosophicalProject 223–8, 237

(G) Groundwork of the Metaphysics of Morals 92,110–20, 124–36, 186, 198–9, 247

(IAG) “Idea of a universal history from acosmopolitan point of view” 236–9, 250,252

(JL) The Jäsche Logic 53, 54, 57–8, 181–2, 234(KpV) Critique of Practical Reason 110–11,116–20, 186

(KrV) Critique of Pure Reason 15–30, 31–48,56–7, 61–3, 67–72, 78–83, 110, 111n, 115,120–3, 147–8, 167–70, 183–5, 187,196–8, 239

(KU) Critique of the Power of Judgment (KU) 94,165–76, 177–86, 189, 193–200, 201–14,217–22, 224, 226, 234–5, 237, 242–53

(MAN) Metaphysical Foundations of NaturalScience 202

(MS) The Metaphysics of Morals 110(ND) New Elucidation of the First Principles of

Metaphysical Cognition, TheoreticalPhilosophy 142

(PRO) Prolegomena to Any FutureMetaphysics 34, 70n, 179

INDEX 285

Kant’s works (cont.)(RGV) Religion within the Boundaries of Mere

Reason 100–9, 111, 253, 262–9(RSV) “Review of Schulz’s Attempt at anintroduction to a doctrine of morals for allhuman beings regardless of differentreligions” 111–12, 113–14, 119

(SF) The Conflict of the Faculties 234n(TP) “On the common saying: That may becorrect in theory, but it is of no use inpractice” 223

(UE) On a discovery whereby any new critique ofpure reason is to be made superfluous by an olderone 189

(VEF) Preliminaries to Perpetual Peace 224n(WHD) “What does it mean to orient oneselfin thinking?” 233–4

Kitcher, Philip 49Korner, Stephen 151nKorsgaard, Christine 125n, 131nKuypers, K. 224n

Langton, Rae 76nlaws (empirical) 167–73, 173, 174–6, 177–8,

181, 183–4, 194laws, moral see the moral lawLeibniz, G. W. 7, 8, 49, 56, 199–200, 208on freedom 111n, 136–9, 145, 150, 160on natural kinds 191–3, 195on pre-established harmony 189, 196, 199on religion 270, 272, 273on sensibility and the intellect 61, 65on space 20–1, 60, 71–2

Lessing, G. E. 232, 254–61, 262, 266, 267,269–73

Lessing, Karl 270nLinnaeus, Carl 173, 181, 196, 248Locke, John 8, 190–3, 195, 257logical forms of judgment, the 31–2, 34–5, 42, 45Longuenesse, Beatrice 1, 3, 31–48, 193Luther, Martin 159n

McDowell, John 5, 91, 93–96McFarland, J. D. 168n, 201nMcLaughlin, Peter 202n, 203n,

205n, 242nMalebranche, Nicolas 55–6manifoldof intuition 35–6, 38, 45–6as requiring synthesis 48

Marc-Wogau, Konrad 204n, 206n, 210nmatter, corpuscularian theory of 190–1maxims 124and rational agency 26see also incorporation thesis

mechanismprinciple(s) of 202–4, 205, 206, 212–14as regulative 203, 206

and teleology 175, 203, 206, 210, 212–14,242, 248–9

Mendelssohn, Moses 10, 229–30, 232–3, 235,269n, 270n

Mertens, Helga 170nMichalson, Gordon E. Jr. 256n, 262n, 266nmoral intuitionism 116–17moral law, theand evil 104–6and freedom 92, 116–20, 146and rational agency 126, 146–7realizability of 218–23see also categorical imperative

moral worth 118myth of the given 92, 96

naturalism 87–98and evil 99, 107–9and free will see freedomand spontaneity 93–4

natural kinds 171–3, 178, 181, 190–3, 194, 196nature 239, 243

see also freedom; the purpose of nature;purposiveness of nature

necessary being 28–30necessityand freedom 22–23, 139–40, 142

Neuhauser, Frederick 154n, 155nNew Testament, the 254, 260Newton, Isaac 60, 71–2, 76–7Nietzsche, Friedrich 118, 159nnorms and rational agency 25–7, 91, 92–4, 113,

129–30, 186noumenal world 95noumenon see things in themselves

objective validity 34, 35, 44nobjective reality 35, 44npractical 222

objectsin general 32, 73–5; see also things inthemselves

of intuition 44of judgment 34Object and Gegenstand 35, 43–4

Old Testament 254–9O’Neill, Onora 7, 125, 132–3, 135–6ontology 68, 71–2, 74, 82, 121, 140, 175, 207,

212norganisms (organized beings) 21, 208–14, 239,

242–4Origen 257original sin 99, 272

see also radical evil

Paton, H. J. 124n, 126n, 131nperpetual peace 219–20, 221, 223–8,

237, 250

286 INDEX

personality, predisposition to 101–2Pinder, Tillman 70nPippen, Robert 166nPlato 160postulates of pure practical reason 110, 120, 268power of judgment (Urteilskraft) 33, 43, 179

see also reflective power of judgmentpredicate-in-notion principle 49, 50redispositions (Anlagen) 101–2, 238

see also animality, humanity, personalityprinciplesmathematical vs. dynamic 17regulative 122, 167–77, 179, 187–8, 193,198, 199–200, 206, 207n

propensity (Hang) 102, 103prototype (moral) 262–3prudence 119purpose (final end) of creation 220–1, 248,

249, 251purpose (ultimate end) of nature, the 238, 245,

247–9, 249–51purposiveness of nature 238, 242–4in the Antinomy of TeleologicalJudgment 208–14

formal (logical) vs. material (real) 171, 210,214, 240

in the formation of empirical concepts 172–3,178–81, 194–6

intrinsic vs. extrinsic 241, 244–6and mechanism 204, 210–14, 242and ontological commitment 175, 207–9,212n, 244

and the realization of morality 218–23, 224as regulative 174–5, 179, 185–6, 199transcendental deduction of 171–4, 182–8,196–9, 210

the transcendental principle of the 166,170–2, 177–82, 185–8, 194–200, 198

see also systematicity; designPutnam, Hilary 68n, 122

Quine, W. V. O. 3, 50, 57Quinn, Philip 102n

rational agency 23–8, 87–8, 128–31, 144–5, 247see also categorical imperative

rationalism 49–66, 232and the analytic-synthetic distinction 51, 57and freedom 137–45regarding religion 267–9and sensibility 55–6, 65as theocentric 52, 64–6

realismempirical 68, 83, 121transcendental see transcendental realism,transcendental idealism

realm of ends 130n, 222, 253

reasonin its empirical employment 18–19practical 26, 91, 97, 114, 117, 153as regulative 167–70, 174–7

reciprocity thesis 112, 115, 116, 146reducibility principle 49, 50–1, 61, 64

see also discursivity thesisreflection, logical act of 33, 45, 186

see also concepts, formation ofreflective power of judgmentand concept formation 43, 178–82, 193–6and the contingency of fit between particularand universal 174, 185, 198, 210–11, 213,218–19

and induction 186–8and purposiveness 170, 200, 204–8, 211,218–20, 239–42

see also concepts; purposivenessregulative principles see principlesReich, Klaus 126n, 128Reimarus, Hermann Samuel 255, 256, 264–79nreligiondefined 267faith in 263–9and history see historyhistory of 260–1, 267–8learned 268–9and miracles 234, 256, 265and morality 259–60, 262–4, 268–9, 271–3natural 254, 255, 267–9revealed see revelationthe truth of 254–61see also community, ethical; God; revelation

representation 68–69, 73republican form of government 221, 223, 225,

228, 252respect for the moral law 100, 101, 104–5revelation 254–73, 267–9, 270the Bible’s status as 255and education 256–61in an ethical community 265

Rousseau, Jean-Jacques 106, 221, 224, 264

St. Augustine 257, 259Sartre, Jean-Paul 264Satan 264schemata 33, 35, 36, 45, 59, 73Schiller, Friedrich 270nSchmidt, James 229n, 230nSchneewind, J. B. 140n, 141nSchopenhauer, Arthur 7, 158–61Schulz, E. G. 229nSchulz, J. H. 111Schumann, J. D. 256self-deception 106self-legislation 130–1, 147, 155self-love 101–2, 104, 106–7

INDEX 287

Sellars, Wilfred 89, 91, 113sense (inner) 39sensibilityand cognition 56as effected by the understanding 37–9as passive 47, 113see also forms of sensibility

sensuous impulse 23, 24skepticism 83, 96, 182, 187–8spaceas a priori and intuitive 61–3, 72nabsolute vs. relational 60–1, 71–2concept of 41as infinitely divisible 20–1as a whole 20see also forms of sensibility

space of reasons 91, 92–4, 97, 113, 186, 199Spinoza, Benedict de 55, 139, 160, 175n, 208,

261, 270spontaneityas beginning a causal series 23, 27, 148–51practical see freedomof reason 26, 91–5, 113–15of understanding 26, 39–40, 41, 47–8, 53, 91,113–15

Straud, Barry 87, 97Strawson, P. F. 55, 77n, 94–5substance 190and matter 20–1

sufficient reason 50principle of 137–41, 159

supersensible, the 95, 213–14, 245superstition 234supreme principle of morality 126synthesisintellectual 33–6, 45see also transcendental synthesis of theimagination

synthetic a priori judgmentsproblem of 57, 59, 74and space 61

systematicity 167–76, 171–3, 175, 177–81,194–6, 212, 214, 218–19, 240–1

see also purposiveness of nature

teleologyhistorical 247–53moral 220–8, 249–53see also design; final cause; mechanism;purposiveness

Tertullian 257nthings in themselvesas noumena 78, 95–6and sensibility 39and things as such (in general) 73, 76nand understanding 64

Thompson, Kevin 175n

timeconcept of 41and transcendental synthesis ofimagination 37, 48

see also forms of sensibility, spaceTimmermann, Jens 127nTonelli, Georgio 210ntranscendental, the 70–1Transcendental Aesthetic 61–3, 71–2and the Transcendental Deduction 37

Transcendental Analytic 74, 167, 169–70, 185,197–8

Transcendental Deduction of the Categories 32,33–48

A vs. B 32, 33; see also A-Deduction,B-Deduction

and the deduction of the purposiveness ofnature 184–5, 187–8, 196–8

Transcendental Dialectic 121–2transcendental idealism 67–83, 94–6and the anthropocentric paradigm 52, 64–6, 77and the antinomies 15–16, 21, 23–4, 28,78–83, 120–3, 147–51

defined 63and epistemic modesty 75–8and freedom 22–3, 81–2, 96–8, 120–3, 143,147–51, 152

in Kant’s later works 165–6as metaphilosophical (methodological) 64, 65,72, 81, 95

neglected alternatives objection to 76and space and time 63, 68–9, 71–2and transcendental realism 63–4, 67–5, 95–6;

see also transcendental realism“two-aspect” vs. “two-world” view of 64,66, 67, 81–2, 96

transcendental realism 67–75, 122, 207nand argument for transcendentalidealism 67–72, 79–83, 95

and things in themselves 74–5, 148transcendental synthesis of the imagination 33,

35, 37–42, 46–8and the categories 40–2

Troelsche, Ernst 266ntruthand judgment 34in rationalism 50

Tuschling, Burkhard 7, 165–7, 173–4, 175n,176, 198n

understandingdiscursivity of 174, 209(human) faculty of 37–8, 41, 44, 47, 52,113–14, 174–6, 184, 185, 190–1, 197,203–4

intuitive 165–6, 174–6, 204, 211, 213–14pure 75

288 INDEX

unity of apperception 185, 197–8and the categories 32–4and forms of sensibility 37, 46–7

unity of reason 91Unmündigkeit (tutelage) 230–1unsociable sociability 99, 106–9, 219, 227,

253, 264–5

Vaihinger, Hans 120virtue 108–9

Weil, Eric 238Wessel, Leonard 257nWeyand, Klaus 228nwilland enlightenment 233the good 220

Wille vs. Willkür 147, 158see also freedom, rational agency, reason,practical

Wolff, Christian 7, 70, 137, 139–40, 142Wolff, Robert Paul 129nWood, Allen 5, 7, 69n, 99–100, 102n, 106–7,

124n, 125–6, 127n, 130, 135n, 135–6, 158,219n, 222n, 223, 238n, 252n, 268n

worldage and size of 16, 18–19, 78–81,122as thing in itself 19, 79–80, 122

Yovel, Yirmiahu 228n, 238n, 266n

Zammito, John H. 173Zollner, J. F. 229Zumbach, Clark 207n, 211n, 212n

INDEX 289

henry e. allison-essays on kant-oxford university press (2012).pdf

Documents

university of oxford

idea of freedom

antinomy of pure reason

transcendental realism

rights department

reective judgment

blank essays

transcendental idealism