Kant on Perception: Naive Realism, Non-Conceptualism, and the B-Deduction

Download Kant on Perception: Naive Realism, Non-Conceptualism, and the B-Deduction

Post on 22-Jan-2017

214 views

Category:

Documents

2 download

TRANSCRIPT

The Philosophical Quarterly Vol. 64, No. 254 January 2014ISSN 0031-8094 doi: 10.1093/pq/pqt019Advance Access Publication 5th November 2013KANT ON PERCEPTION: NAIVE REALISM,NON-CONCEPTUALISM, AND THE B-DEDUCTIONBy Anil GomesAccording to non-conceptualist interpretations, Kant held that the application of concepts is not necessaryfor perceptual experience. Some have motivated non-conceptualism by noting the affinities between Kantsaccount of perception and contemporary relational theories of perception. In this paper, I argue (i) thatnon-conceptualism cannot provide an account of the Transcendental Deduction and thus ought to berejected; and (ii) that this has no bearing on the issue of whether Kant endorsed a relational account ofperceptual experience.Keywords: Kant, perception, non-conceptualism, relational, representational.I. INTRODUCTIONRecent debates in the philosophy of perception have focused on the contrast be-tween relational and representational theories of perceptual experience. For initialpurposes, the following rough characterisation will suffice: relational theoriesare those that hold that the phenomenal character of perceptual experienceessentially involves the obtaining of a non-representational relation that holdsbetween subject and perceived objects. Representational theories are thosethat hold that the phenomenal character of perceptual experience essentiallyinvolves representational properties that determine accuracy conditions forthe perceptual state.Interest in these debates has been prompted by the recent developmentand defence of nave realist relational theories. Such views hold that the non-representational relation involved in perceptual experience is one that subjectsstand in to ordinary material objects and their properties. Versions of this viewwere popular amongst the early 20th-century Oxford Realists, but it is therecent work of John Campbell, Mike Martin, and others that has brought theproposal back into the philosophical landscape.11 CookWilson, J. (1926) Statement and Inference, with Other Philosophical Papers. Oxford: ClarendonPress; Prichard, H. A. (1909) Kants Theory of Knowledge. Oxford: Clarendon Press; Campbell, J.C The Author 2013. Published by Oxford University Press on behalf of The Scots Philosophical Association and the Universityof St Andrews. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/2 ANIL GOMESThis debate in the philosophy of perception intersects with a recent issue ofinterpretation in Kants theoretical philosophy. Kant famously holds that thereare two stems to human cognition: a passive faculty of sensibility and an activefaculty of the understanding. The former presents us with objects by meansof intuitions; the latter enables thought by means of concepts. But thoughtswithout content are empty and intuitions without concepts are blind: only fromtheir unification can cognition arise (A51/B76).2 An important question to askis how we should understand the relation between intuitions and concepts andwhat contribution each makes to our perceptual consciousness of the world.Following a series of papers by Robert Hanna and Lucy Allais, answersto these questions have split into two broad camps. The traditional concep-tualist interpretation holds that the application of concepts is necessary forthe perceptual presentation of empirical objects in intuition. In contrast, thenon-conceptualist interpretation of Allais and Hanna holds that intuitions canpresent us with empirical objects without any application of concepts.This terminology is somewhat unhelpful since the terms conceptualist andnon-conceptualist are used in the philosophy of perception literature to pickout varieties of representational theories: conceptualist theories hold that per-ceptual experience involves properties that represent the world as being someway and that subjects who undergo such experiences need possess the con-cepts required to specify the content of those experiences; non-conceptualisttheories hold that perceptual experience involves properties that represent theworld as being some way but deny that subjects need possess the conceptsrequired to specify the content of those experiences.3 This use should not beconfused with the terminology used by those involved in the debate about howto understand Kants theoretical philosophy. In the rest of this paper, I willuse the terms conceptualism and non-conceptualism solely in the Kantiansense.How does the debate about the nature of perceptual experience relate to theKant debate? There is no immediate correspondence between positions in onedebate and positions in the other. Yet those on both sides of the Kant debateoften assume that conceptualist interpretations are committed to ascribing toKant a representational account of perceptual experience. This is importantbecause one way of motivating non-conceptualism goes via the claim that(2002) Reference and Consciousness. Oxford: Oxford University Press; Martin, M. G. F. (2002) TheTransparency of Experience, Mind and Language, 4: 376425; Brewer, B. (2006) Perception andContent, European Journal of Philosophy, 14: 16581.2 References to the Critique of Pure Reason [Kant, I. (1998) Critique of Pure Reason, ed. andtrans. P. Guyer and A. Wood. Cambridge: Cambridge University Press.] follow the standardA/B pagination. All other references are to the volume and page of the Akademie edition ofKants works Kant, I. (1902) Kants Gesammelte Schriften, ed. Koniglich Preussische Akademie derWissenschaften, vols. 129. Berlin: de Gruyter.3 Crane, T. (1992) TheNon-Conceptual Content of Experience, in T. Crane (ed.)The Contentsof Experience, pp. 13657. Cambridge: Cambridge University Press. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 3Kants account of the perceptual presentation of empirical particulars shouldbe read on the model of a relational account of perception.4 The assumptionappears to be that if Kants account of intuition is relational, then he cannothave thought that the application of concepts is necessary for the perceptualpresentation of empirical particulars.My primary concern in this paper is to suggest that non-conceptualism isfalse: Kant holds that the application of concepts is necessary for the percep-tual presentation of empirical objects in intuition. But I will also show that thishas no implications for the question of whether Kant endorsed a representa-tional or relational theory of perception. Relational theories of perception arecompatible with conceptualism.Why would one think that conceptualism required a representational ac-count of perceptual experience? It is true that influential conceptualist in-terpretations have ascribed to Kant a representational account of perceptualexperience, but it is hard to find an explicit argument in the literature for thissupposed link.5Here is one line of thought: according to conceptualist interpretations,Kant held that the application of concepts is necessary for the perceptualpresentation of empirical objects. The reason for endorsing this claim is thatKant takes intuitions to depend on acts of synthesis. And acts of synthesisare undertaken by the understanding: they take the manifold of intuition andcombine it according to rules. These rules are concepts of the understanding.Combining the manifold of intuition in accordance with rules thus involvesapplying concepts in intuition. And if concepts are applied in intuition, thenperceptual experience represents the world as being a certain way. Thus thereasons that motivate a conceptualist interpretation of Kant also motivateascribing to him a representational theory of perception.6In the final section of this paper I will examine this argument. Before that,in 2, I will set out the debate between conceptualist and non-conceptualistinterpretations and draw attention to the considerations that motivate eachside of the debate. In 3, I will draw on the B-edition of the TranscendentalDeduction to provide some reason for thinking that non-conceptualism is false.Finally, in 4 I will show that this has no bearing on the question of whetherKant endorsed a relational theory of perception: conceptualism is compatiblewith relational theories.4 Allais, L. (2009) Kant, Non-Conceptual Content and theRepresentation of Space, Journal ofthe History of Philosophy, 47: 383413; 38792; Allais, L. (2010) Kants Argument for TranscendentalIdealism in the Transcendental Aesthetic, Proceedings of the Aristotelian Society, 110: 4775; 5862;Allais, L. (2011) Idealism Enough: Response to Roche, Kantian Review, 16: 37598; 37983.5 McDowell, J. (1998) Having the World in View: Sellars, Kant, and Intentionality, Journalof Philosophy, 95: 43191; Abela, P. (2002) Kants Empirical Realism. Oxford: OUP.6 Ginsborg, H. (2006) Kant and the Problem of Experience, Philosophical Topics, 34: 59106;pp. 647 presents a particularly clear exposition of this line of thought. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/4 ANIL GOMESII. NON-CONCEPTUALISMAccording to non-conceptualist interpretations of Kants theory of cognition, wecan be perceptually presented with particulars without any input from theactive faculty of the understanding.7 Since Kant introduces the understandingas a faculty for judging and tells us that all judgement proceeds via concepts(A69/B94), this is often expressed as the claim that one can be perceptually pre-sented with particulars without the application of concepts. Non-conceptualistreadings hold that for Kant, the application of concepts is not necessary forour being perceptually presented with outer particulars.8Non-conceptualism is opposed by those who hold that, for Kant, the ap-plication of concepts is necessary for perceptual experience.9 But we need tobe careful as to what is meant by the application of concepts. Conceptualistinterpreters often distinguish two aspects of the activity of understanding10:the understanding as rule-giver for the synthesis of the manifold in intuitionand the understanding as discursive combiner of concepts in judgement. Itis the former, and not the latter, which conceptualist interpretations take tobe necessary for perceptual experience. Thus if the phrase application ofconcepts is reserved for the latter activitythe deployment of concepts injudgementthen there is no bar to conceptualists accepting Allaiss claimthat the application of concepts is not necessary for our being perceptuallypresented with outer particulars.11 Rather, the sense in which conceptualiststake the application of concepts to be necessary for perceptual experience isthat they take the perceptual presentation of objects in intuition to require aperceptual synthesis of the manifold of intuition, undertaken by the under-standing in accordance with concepts.One way to characterise this dispute is over where best to limn the domainof the understanding. Traditional conceptualist readings of Kants theory ofcognition see the understanding as reaching all the way out to perception itself:the understanding is active in perceptual experience because the application ofconcepts is required for the perceptual presentation of outer particulars. Non-conceptualists hold that the understanding is required only for subjects toengage in a certain form of thought: the perceptual presentation of particulars7 Allais, Kant, Non-Conceptual Content and the Representation of Space; Allais, L. (2012)Transcendental Idealism and the Transcendental Deduction, in D. Schulting and J. Verburgt(eds) Kants Idealism: New Interpretations of a Controversial Doctrine. Dordrecht: Springer; Hanna, R.(2001) Kant and the Foundations of Analytic Philosophy. Oxford: OUP; Hanna, R. (2005) Kant andNonconceptual Content, European Journal of Philosophy, 13: 24790.8 Allais, Kant, Non-Conceptual Content and the Representation of Space, p. 394.9 McDowell, Having the World in View; Abela, Kants Empirical Realism.10 Longuenesse, B. (1998) Kant and the Capacity to Judge, p. 63. Princeton, NJ: Princeton Uni-versity Press.11 Allais, Kant, Non-Conceptual Content and the Representation of Space, p. 394. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 5can take place in the absence of concepts, but we require input from theunderstanding in order to cognise them in a certain way.We can mark this distinction by distinguishing the conditions necessary toengage in a certain sort of thought about objects and the conditions necessaryto be perceptually presented with such particulars. Conceptualist readings ofKant hold that the application of concepts is necessary for the perceptualpresentation of particulars; non-conceptualists hold only that the applicationof concepts is necessary for us to think about objects in a certain way. AsAllais puts it, once we draw a distinction between the perception of a distinctparticular and cognition of an object in the full-blown Kantian sense of an object,[non-conceptualists] can allow that Kant does not see concepts as necessaryfor the basic intentionality of perceptionthat fact that perception presentsus with distinct particular things.12 And this is compatible with thinking thatthe understanding is required for cognition.How should we decide between these views? Conceptualism is sometimesmotivated by appeal to the opening paragraphs of the Transcendental Logicin which Kant distinguishes sensibility from the understanding and notes theirinterrelation (A5052/B7476). But this will not suffice. Kants claim in thesepassages is only that the unification of sensibility and the understanding isrequired for cognition [Erkenntnis] (A51/B76), and this falls importantly short ofclaiming that their co-operation is required for perception itself. Thus Kantsoft-quoted claim that thoughts without concepts are empty, intuitions withoutconcepts are blind (A51/B75) need not be read as claiming that intuitionwithout concepts do not amount to perception, but rather that intuitionswithout concepts are incapable of yielding cognition.One might attempt to draw a link from these passages to perception by not-ing that Kant held experience [Erfahrung] to be a kind of cognition requiringthe understanding (Bxvii), for if the understanding is active in cognition, andif experience is a kind of cognition, then the understanding is active in expe-rience. But Kants use of the term experience is not continuous with that ofcontemporary philosophers of perception and it is open for non-conceptualiststo hold that at least some of Kants uses of the term pick out a form of judge-ment made on the basis of perceptual experience rather than the experienceitself (e.g. B166, A176/B218, A189/B234). On this reading, those passages inwhich Kant claims that Erfahrung requires the active, combinatorial input ofthe understanding (A93/B126) show only that the understanding is requiredfor a certain sort of empirical judgement or thought. There is nothing thus farthat threatens the non-conceptualist claim about perception.It is for this reason that the debate between conceptualists and non-conceptualists has largely focused on whether the application of concepts12 Allais, Transcendental Idealism and the Transcendental Deduction, p. 41. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/6 ANIL GOMESis required for intuition [Anschauung]. In contrast to his use of term Erfahrung,Kant tells us explicitly that visual perception is a form of objective empiricalintuition [An13 AA07:154; cf. Prol.14 AA04:283] and that empirical intuition isthe means by which we are perceptually presented with objects [A180/B222;Prol. AA04:283]. Thus if the application of concepts is required for Anschauungitself, this would seem to tell against non-conceptualism: the discursive activ-ity of the understanding would be involved in the very perception of distinctparticulars and not just required for their cognition in thought.Are there any reasons to think that one can be presented with particularsin intuition absent any function of the understanding? Particular attention hasbeen paid to a passage at A90/B123 in which Kant raises the possibility thatappearances could after all be so constituted that the understanding wouldnot find them in accord with the conditions of its unity, before concluding that[a]ppearances would nonetheless offer objects to our intuition, for intuition bynomeans requires the functions of thinking (A90/B123).15 Non-conceptualiststake Kant to be raising a genuine metaphysical possibility here, one that issignalled by his claim that objects can indeed appear to us without necessarilyhaving to be related to the functions of the understanding (A89/B122). But analternative is to take these passages as expressing a mere epistemic possibilitythat will later be shown not to be a genuine metaphysical possibility at all. Andan epistemic possibility is compatible with the conceptualist reading.Is the metaphysical reading supported by the fact that Kant uses the in-dicative can [konnen] in the formulation at A89/B122, as opposed to thesubjunctive could [konnten] at A90/B123?16 The issue is not clear. Guyer andWood note a passage in the Reflexionnen (177678, AA19:122317) where Kantformulates an epistemic possibility without using the subjunctive.18 And, moregenerally, the use of the indicative in the formulations at A89/B122 may becompatible with the three paragraphs that end that section (A8992/B122124) operating under an assumed for all we know operator. We are not forcedto treat the possibility expressed at A90/B123 as metaphysical.More compelling, tomymind, areKants scatteredwritings about the natureof non-human animal (hereafter: animal) engagement with the world. Kanttakes animals to be sensible beings that are incapable of discursive thought13 Kant, I. (2007) Anthropology from a Pragmatic Point of View, in R. B. Loude and G.Zoller (eds) Anthropology, History and Education. Cambridge: Cambridge University Press.14 Kant, I. (2002) Prolegomena to Any Future Metaphysics, in H. Allison and P. Heath (eds)Theoretical Philosophy after 1781. Cambridge: Cambridge University Press.15 Hanna, Kant and Nonconceptual Content, pp. 24950; Allais, Kant, Non-ConceptualContent and the Representation of Space, pp. 3878.16 ibid., p. 387, n. 13.17 Reprinted in Kant, I. (2005) Notes and Fragments, eds and trans C. Bowman, P. Guyer, andF. Rauscher, p. 222. Cambridge: Cambridge University Press.18 Critique of Pure Reason, p. 725, n. 17. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 7[An AA07:196]. Thus if one wants to make it plausible that Kant held thatwe can be perceptually presented with objects through sensibility alone, it isnatural to consider Kants views on animal consciousness. And there are anumber of passages in which Kant appears to suggest that animals can beperceptually presented with objects through sensibility, despite the fact thatthey do not have the resources to conceptualise such objects. These passageshave been taken to support a non-conceptualist reading of Kant.19Consider Kants rejection, repeated at various places throughout his writ-ings, of Descartess view of animals as merely mechanical. In the Critique of thePower of Judgement, Kant says that animals also act in accordance with represen-tations (and are not as Descartes would have it, machines) [CJ 20 AA05:464n;cf. MV 21 AA28:449, FS22 AA02:330331]. Kant takes this to mark a differencebetween his and Descartess views on animal cognition and such acting is oftenexplicated as involving perceptual acquaintance with particulars in the world.The most significant remark is found in Kants discussion of the differentlevels of cognition in his lectures on logic where he says that [a]nimals areacquainted with objects too, but they do not cognize them [JL23 AA09:645;see also VL24 AA24:846]. And whilst he denies that the ox has a distinct con-cept of its stall, he is clear that the animal perceives it [FS AA02:59].25 Thesepassages suggest that animals can be perceptually aware of particulars in theenvironment without any involvement of the understanding. I take this to bea significant consideration in support of non-conceptualism.However, Kants comments on animal consciousness are varied and widelydispersed, and there are prominent passages central to the first Critique thatappear to tell against the non-conceptualist reading. Let me highlight twofrom the Transcendental Analytic. The first concerns the role of synthesisin the representation of intuitions; the second concerns Kants aims in theTranscendental Deduction of the Categories. In each case I will set out why the19 Allais, Kant, Non-Conceptual Content and the Representation of Space, pp. 4067.20 Kant, I. (2001) Critique of the Power of Judgment, eds P. Guyer and E. Matthews. Cambridge:Cambridge University Press.21 Kant, I. (2001) Metaphysics Vigilantius, in K. Ameriks and S. Naragon (eds and trans)Lectures on Metaphysics. Cambridge: Cambridge University Press.22 Kant, I. (1992) The False Subtlety of the Four Syllogistic Figures, in D. Walford andR. Meerbote (eds and trans) Theoretical Philosophy 17751770. Cambridge: Cambridge UniversityPress.23 Kant, I. (1992) Jasche Logic, in J. M. Young (ed. and trans.) Lectures on Logic. Cambridge:Cambridge University Press.24 Kant, I. (1992) Vienna Logic, in J. M. Young (ed. and trans.) Lectures on Logic. Cambridge:Cambridge University Press.25 See Naragon, S. (1990) Kant on Descartes and the Brutes, Kant-Studien, 81: 123; McLear,C. (2011) Kant on Animal Consciousness, Philosophers Imprint, 11 (15) for further discussion andtextual evidence. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/8 ANIL GOMESmaterial has been thought to support a conceptualist reading before providingthe non-conceptualist response.26(1) Synthesis: the most telling passages in support of the idea that the under-standing is involved in intuition are those in which Kant describes the roleof synthesis in the representation of intuition (A98-A107, A117f.). Synthesisis the activity of putting different representations together with each other(A77/B103); it is a necessary ingredient (A120n) in the perception of objectsbecause otherwise the manifold of intuition would be dispersed and separatein the mind (A120); without synthesis we would have only unruly heaps ofrepresentations [Vorstellungen] (A121). But the same function which gives unityto the various ideas in a judgement also gives unity to the mere synthesis ofvarious ideas in an intuition (A7980/B105106) and all combination is an actof the understanding (B130). Thus the perceptual presentation of particularsin intuition involves a discursive act of the understanding.The non-conceptualist response to these passages is to deny that all synthesisis a result of the understanding: as Allais puts it, synthesizing is not the sameas conceptualizing.27 And though there are passages in which Kant presentssynthesis as an act of the understanding, he most often ascribes its func-tion to the imagination (A78/B104, A118, A119, A120, A123, A124, B151), anintermediate faculty that has aspects of both sensibility and the understand-ing. Hanna similarly takes synthesis to be a lower-level spontaneous cognitivepower that falls under the remit of sensibility: sensibility is only relatively pas-sive, but not entirely passive . . . by virtue of its expressing a mental power forspontaneous synthesis, or mental processing.28 The non-conceptualist claimis that the synthesis of intuitions can take place absent any function of the un-derstanding and that the mere recognition of processing activity is not enoughto show the involvement of the understanding in perception.(2) The Deduction: Kants stated aim in the Transcendental Deduction is toshow that without their [the categories] presupposition nothing is possible asobject of experience (A93/B125). For the objective validity of the categories,as a priori concepts, rests on the fact that through them alone is experiencepossible (A93/B126). Traditional readings of the Transcendental Deductionhave taken these passages to support the claim that the categories are conditionson the possibility of experience, a conclusion that has been taken as equivalentto the conceptualist claim that the application of concepts is required forperceptual experience. On this reading, demonstrating the objective validity of26 See Griffith, A. (2012) Perception and the Categories: A Conceptualist Reading of KantsCritique of Pure Reason, European Journal of Philosophy, 20: 193222, for further discussion of thesepassages.27 Allais, Kant, Non-Conceptual Content and the Representation of Space, p. 396.28 Hanna, Kant and Nonconceptual Content, p. 249. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 9the categories requires showing how intuitions already involve the actualisationof categorial capacities.29Allaiss response is to dispute this account of the Deduction. Drawing on thedistinction between the conditions necessary to perceive a particular and thosenecessary to cognise an empirical object, she claims that the Deduction aimsonly to show that the categories are necessary conditions on the possibility ofthinking about objects in a particular way (as persisting, causal unities) and notconditions on being presented with particulars in perception.30 As we havealready noted, Kant takes experience [Erfahrung] to be a kind of cognitioninvolving the understanding. Thus when Kant says here that without thecategories presupposition nothing is possible as an object of experience, thenon-conceptualist takes him to be making a claim about the necessity ofthe categories for thinking of objects in a particular way. On Allaiss non-conceptualist reading, the Deduction aims only to show that the categories areconditions on a certain sort of thought.How should we weigh these competing considerations? In what follows Iwill set out a reason for thinking that the understanding must be involved inintuition for Kant, and therefore that the perceptual presentation of objectsinvolves a discursive act of the understanding. The issue turns on a set ofpassages in 2026 of the B-Edition of the Transcendental Deduction ofthe Categories and the role that the argument of the Deduction is intendedto play in Kants philosophy. This will comprise my defence of the claimthat the understanding is active in perceptual experience, contrary to thenon-conceptualist reading, and that the application of concepts is thereforerequired for perceptual experience.III. THE B-DEDUCTIONAs has been well documented, the argument in the B-Deduction consists oftwo separate stages. In 1519, Kant focuses on the role that the categoriesplay as intellectual conditions on empirical representation arguing that [a]llsensible intuitions stand under the categories, as conditions under which alonetheir manifold can come together in one consciousness (B143). This is becausethe combination of the manifold in general can never come to us throughthe senses (B129); such combination is possible only if all the manifold ofintuition stand under conditions of the original synthetic unity of apperception(B136); bringing representations under this synthetic unity requires a processof synthesis; and the rules that govern this synthesis are the categories (B143).29 McDowell, Having the World in View, Lecture II; Abela, Kants Empirical Realism.30 Allais, Transcendental Idealism and the Transcendental Deduction, pp. 416. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/10 ANIL GOMESThus the manifold in a given intuition also necessarily stands under thecategories (B143).In the second part of the B-Deduction, Kant shifts his focus from thecategories as intellectual conditions on representation to the way in whichobjects come before our sense (B160). The intention is to show that fromthe way in which the empirical intuition is given in sensibility that its unitycan be none other than the one the category prescribes to the manifold of agiven intuition (B1445). The main argument in support of this claim occursprimarily in 26 with reference to material outlined in 24 and Kant is explicitthat this second step is needed to complete his proof (B144145).31Kants argument in these passages centres on the role that space and timeplay in our representation of empirical particulars. As the forms of humansensible intuition, space and time structure the manifold of appearance sincesuch amanifold can only occur in accordance with this form (B161). But spaceand time are represented by us not only as forms of sensible intuition, but alsoas intuitions themselves, and therefore as possessing a unity of the manifoldof empirical intuition within them. This unity precedes all concepts, thoughto be sure it presupposes a synthesis, which does not belong to the senses butthrough which all concepts of space and time first become possible (B161n).This presupposed synthesis is one in which the understanding determinesthe sensibility (B161n). So the unity of space and time is to be explainedwith reference to the effect of the understanding upon sensibility itself: theunderstanding plays a role in our representation of particulars as situated inspace and time.How should we understand these passages? The second part of the De-duction isolates a form of synthesis that is involved in some aspect of ourrepresentation of particulars. Since Kant is explicit that this form of synthesisis one that proceeds from the understandinghe calls it figurative or transcen-dental (B151) and describes it as an effect of the understanding on sensibility(B152)it is not open to the non-conceptualist to claim that this is a processof combination that does not involve the understanding:32 Kant states unam-biguously that the transcendental synthesis described in 24 is one in whichthe understanding determines the sensibility (B160n).If such synthesis is governed by the understanding, is it required for the per-ceptual presentation of particulars in space and time or only for our cognitionof objects? Allais claims the latter: she takes these passages in the second partof the B-Deduction to show only that there is a way of thinking about objectsas spatial that requires the input of the understanding. As sensible beings,31 In what follows I draw on my reading of the Transcendental Deduction set out in Gomes,A. (2010) Is Kants Transcendental Deduction of the Categories Fit for Purpose? Kantian Review,15: 11837.32 Allais, Kant, Non-Conceptual Content and the Representation of Space, pp. 3967. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 11we are presented with objects as spatially arrayed through the form of ourintuition and this enables the perceptual presentation of particulars distinctfrom us and situated about us in the environment. But there is also a wayin which concept-using creatures such as ourselves think about space whenrepresenting the world as objective and to say that our representation of spaceis transformed by the understanding through a transcendental synthesis is tosay only that there is a way of thinking about space that requires the activityof the understanding. This is compatible with the absence of the understand-ing in the perceptual presentation of particulars as spatially and temporallyarrayed.33In order to see why this non-conceptualist reading of 2226 is inadequate,we need to fix upon the purpose of the Transcendental Deduction. On Al-laiss reading, Kants aim in the Deduction is to show that the categories arenecessary conditions on a certain sort of thought. This is a scaling back ofthe traditional import of the Deduction and Hannah Ginsborg complains thatit threatens to trivialize Kants central project in the Critique, or at least todiminish its interest and importance.34 This is unfair: as Allais points out itwould be interesting if a certain way of thinking about objectsa way thatKant thinks to be both necessary and a prioriwere required in order for us toascribe properties to persisting objects in the world.35 So the retreat from iden-tifying conditions on experience to identifying conditions on thought does notrecede to triviality even if it remains a retrenchment of traditional ambitions.What is more important, however, is that this scaling back prevents theTranscendental Deduction from providing a response to Humean concernsabout the justified application of a priori concepts.36 As James Van Cleve haspointed out, there is a difference between showing that we must apply thecategories and that the categories must apply: one may slip without noticingfrom one to the other, but between the two there is no small distance. Itis the distance between our using a category and its being instantiated, orbetween making a judgement and it being true.37 Humean scepticism aboutthe justified application of a priori concepts will not be answered by showingonly that we must apply the categories to experience, for that is compatiblewith the falsity of any such application. Kant needs the stronger claim: thatthe categories must apply.Allaiss reading of the Transcendental Deduction supports only the weakerclaim: that we must make use of the categories in making judgements about33 Allais, Transcendental Idealism and the Transcendental Deduction, pp. 478.34 Ginsborg, Kant and the Problem of Experience, p. 62.35 Allais, Transcendental Idealism and the Transcendental Deduction, p. 50.36 HannahGinsborg makes this point in Ginsborg, H. (2008) Was Kant a nonconceptualist?,Philosophical Studies, 137: 6577; compare Strawson, P. F. (1966) The Bounds of Sense, pp. 734, 85.London: Methuen.37 Van Cleve, J. (1999) Problems from Kant, p. 89. Oxford: OUP. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/12 ANIL GOMESthe world as containing persisting, causal unities. But it is compatible withthis conclusion that all the judgements we so make are false. And if this werethe case, our thinking about the world would be subject to an unavoidableerror: we would be compelled, of necessity, to think of the world as containingpersisting substances, capable of existing unperceived and standing to eachother in causal relations; but none of these judgements about the world wouldbe accurate. Without the stronger conclusion that the categories must applyto experience, the Deduction cannot be used to answer Humean scepticismabout justification. The result is not simply a curtailment in the argumentsambitions but a neutering of its force.VanCleves distinction offers us away to read the argument thatmakes senseof the B-Deduction and its relation to Humean scepticism.38 The first partof the proof, 1519, argues for the claim that we must apply the categories,whilst 2226 complete the argument by showing that the categories mustapply. Kants claim is that since the unity of space and time arises from aprocess of transcendental synthesis, that which is given in space and timestands under the unity of apperception. And in virtue of so standing, it isconstituted so as to require synthesis in accordance with a priori rules of theunderstanding, namely the categories. It is the fact that both transcendentalsynthesis and categorial synthesis originate in the understanding that explainswhy the categories must apply.Note that the fact that wemust apply the categories is not independent of thefact that the categories must apply: if it were so independent, the result wouldbe what Kant calls a kind of preformation-system of pure reason (B168), aview on which it is only accidentally true that our application of the categoriesis objectively valid. On the reading offered here, the interdependence of ourapplication of the categories and their required application consists in the factthat both transcendental and categorial synthesis originate in the understand-ing: it is the nature of the understanding that explains both why we must applythe categories and why the categories must apply.39This reading focuses on the way in which transcendental synthesis accountsfor a certain aspect of our empirical intuitions: it thus claims a role for theunderstanding in the perceptual presentation of empirical particulars. If therewere no other activity for the understanding than the application of concepts,then the case against conceptualism would be complete. But Kant describesthe unity of space and time conferred by transcendental synthesis as precedingall concepts (B160n.), a remark that reaffirms the claims of the Aesthetic thatconcepts do not contribute towards our representation of space and time(A245/B39; A312/B47). This suggests that although 2226make the case38 This is the reading offered in Is Kants Transcendental Deduction of the Categories Fitfor Purpose?39 Thanks to a referee for raising this point. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 13for the involvement of the understanding in the perceptual presentation ofparticulars, they do not yet show that perception involves the application ofconcepts.This can seem incoherent: how can there be activity of the understandingthat precedes all concepts when Kant introduces the understanding as thefaculty for judging via concepts (A69/B94, A126)? This is a delicate topic.There is a more fundamental characterisation of the understanding in theB-Deduction as the capacity for apperception (B133134n), and Kant else-where suggests that this capacity is more basic than the categories (A401).Beatrice Longuenesse and others have claimed that Kant is committed toa pre-discursive, and therefore pre-conceptual, function for the understand-ing and one way to understand this is as a non-discursive exercise of thecapacity of apperception distinct from the discursive exercise at work injudging via concepts.40 The prospect of the understanding operating on ourrepresentation of space and time other than through the application of con-cepts leaves open the possibility of a position that is non-conceptualist in letter ifnot in spirit: one on which the perceptual representation of particulars involvesa pre-discursive act of the understanding without involving the application ofconcepts.However, we can now return to 1519 to complete the case against thenon-conceptualist. 2226 show that the categories must apply by isolatinga role for the understanding in the perceptual presentation of particulars assituated in space and time. But if this argument is to work, it must be thecase that the transcendental synthesis discussed in 2226 originates in thesame understanding as the categorial synthesis discussed in 1519: only sowill Kant have a guarantee that what is given in space and time is such asto be necessarily subject to the categories. Thus we must read the processof synthesis offered in the first part of the B-Deduction as originating in theunderstanding and proceeding according to the categories. The synthesis ofthe manifold of intuition takes place according to the categories, contrary tothe non-conceptualist suggestion.On this way of reading the Transcendental Deduction, 1519 show thatwe must synthesise the manifold of intuition in accordance with the categoriesand 2226 show that sensible intuition is constituted such that it must besynthesised in just this way. This is exactly how Kant presents the result of thefirst part of his proof in the summary at 20 where he concludes: the manifoldin a given intuition also necessarily stands under the categories (B143). I have40 Longuenesse, Kant and the Capacity to Judge, p. 211f; Longuenesse, B. (2000) Kants Categoriesand theCapacity to Judge: Responses toHenryAllison and Sally Sedgwick, Inquiry, 43: 91110; IsKants Transcendental Deduction of the Categories Fit for Purpose? pp. 1301; Land, T. (2011)Kantian Conceptualism, in G. Abel and J. Conant (eds) Rethinking Epistemology, pp. 197239.Berlin: de Gruyter. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/14 ANIL GOMESsuggested that we must take this at face value if Humean sceptism is to beforestalled: the categories are active, for Kant, in perceptual experience.Is this consideration decisive? There are two caveats to be borne in mind.The first concerns Kants views on animal consciousness. As mentioned above,there are grounds for thinking that Kant held that animals could be perceptu-ally presented with empirical objects absent any function of the understanding.To the extent that one finds these grounds compellingas I dothen an onusremains on conceptualist interpretations to explain how their account of hu-man perceptual experience is compatible with the thought that non-humananimals can be perceptually presented with objects absent the application ofconcepts. I will not attempt such an explanation here but it is worth notingthat an interpretative debt remains to be discharged.The second concerns the purpose of the Transcendental Deduction. Inraising this objection to non-conceptualist readings of Kant, I have assumedthat the role of the Deduction is to respond to Humean worries about ourjustified application of a priori concepts to experience. And one may contestthis claim. There are other, more local, sources for the Deduction to whichAllaiss reading is responsive. In his comments on Kants Inaugural Dissertation,Johann Heinrich Lambert accepts that knowledge arises out of two entirelydifferent and, so to speak, heterogenous sources, so that what stems from the onesource can never be derived from the other, but questions to what extentthese two ways of knowing are so completely separated that they never come to-gether [C41 AA10:105]. Similarly Marcus Herz, in his Observations on SpeculativePhilosophy from 1771, asks how external things can agree with our intellectualrepresentations.42 These challenges to Kants pre-critical position raise thequestion of how it is that a priori concepts can be applied to experience, notwhether they can ever be accurately applied so.Allaiss reading responds to the concerns of Lambert and Herz: it portraysKant as concerned to show why it is that we must use a certain set of a prioriconcepts in making judgements about objects in the world. And one mighthold that the justification of such application is accomplished elsewhere in theCritique. But if one thinks that the role of the Deduction is to combat Humesproblem [Prol. AA04:259261]or, perhaps more accurately, if one thinks thatHumes problem concerns not only our possession and application of a prioriconcepts but also our justified application of themthen the understandingmust be involved in intuition and, in particular, the manifold of intuition mustbe synthesised in accordance with the categories. This gives us reason to thinkthat, for Kant, the categories need be employed in perceptual experience and41 Correspondence, ed. and trans. A. Zweig. Cambridge: Cambridge University Press, 1999.42 Reprinted in Watkins, E., ed. (2009) Kants Critique of Pure Reason: Background Source Materials,p. 299. Cambridge: Cambridge University Press. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 15therefore that the application of concepts is necessary for perceptual experiencecontrary to the non-conceptualist claim.IV. NAIVE REALISMWhat are the implications of this discussion for the question of whether Kantheld a relational or representational theory of perceptual experience? Theshort answer is: not much. But in order to see why, it will be useful to set outsome definitions.Let the phenomenal properties of an experience be those in virtue of whichthere is something it is like to have an experience. The phenomenal character of anexperience consists of its phenomenal properties. We type experiences by theirphenomenal character: two experiences are of the same fundamental kind ifand only if they have the same phenomenal character.43 Relational theories ofperceptual experience are those on which the phenomenal properties of theexperiences involved in perception essentially involve non-representationalrelations to objects. Representational theories of perceptual experience are thoseon which the phenomenal properties of perceptual experiences essentiallyinvolve representational properties.One question about these definitions is where to place the account of per-ception defended by John McDowell.44 Relational theorists often cite him as aproponent of a representational theory of perceptual experience, on groundsthat McDowell takes perception to have a certain sort of content.45 McDowelldisputes this characterisation: he takes his view to show that a proper account ofthe relational aspect of perception cannot do without representational notionsand thus that relational views are compatible with perceptual experience hav-ing content.46 I will not pursue this issue here. Although McDowells accountof perception is interesting and important, the considerations to be outlinedbelow hold even for those who understand the relational aspect of perceptionin wholly non-representational terms.In a series of recent papers, Lucy Allais has made the case that there areaffinities between Kants account of the perceptual presentation of empiricalparticulars and contemporary relational accounts of perception. She is reluc-tant, for fears of anachronism, to straightforwardly ascribe a relational theoryof perceptual experience to Kant, but she nevertheless holds that one finds in43 Soteriou, M. (2005) The Subjective View of Experience and Its Objective Commitments,Proceedings of the Aristotelian Society, 105: 17790, p. 194.44 McDowell, J. (1994) Mind and World. Cambridge, MA: Harvard University Press; Havingthe World in View.45 Brewer, B. (2007) Perception and Its Objects, Philosophical Studies, 132: 8797.46 McDowell, J. (2013) Perceptual Experience: Both Relational and Contentful, EuropeanJournal of Philosophy, 21: 14457. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/16 ANIL GOMESKant some of the ideas which [relational] theories of perception are trying tocapture.47 Let me state some of her considerations.First, there is Kants notion of intuition. Allais thinks that we cannot makesense of the Kantian notion of intuition without recognising the relationalaspect of his account of perception.48 On Allaiss reading, intuitions are singu-lar and immediate (A320/B377, A713/B741); they are object dependent (Prol.AA04:281, B72); and their role is to give us objects in such a way that we canthink about them (A23/B39, A239/B298). These considerations do not forceupon us a relational reading of Kantian perception, but they can be easilycaptured on a view that takes intuitions to involve relations of acquaintancethat immediately present empirical objects to consciousness in such a way thatthey can be the subjects of thoughts.Secondly, there is the argument for transcendental idealism in the Tran-scendental Aesthetic. Many commentators have noted a gap in this argument:Kant moves from claims about our representation of space to conclusionsabout space itself. Allais argues that this gap disappears if Kantian intuitionsinvolve a relational component.49 That is, ascribing to Kant a relational ac-count of intuition explains why he took himself to be justified in moving fromclaims about our representation of space to conclusions about space itself.Finally, there is the Refutation of Idealism. Kant there takes himself toestablish the reality of outer objects, those objects that Descartes thoughtdoubtful, for the consciousness of my own existence is at the same time animmediate consciousness of the existence of other things outside me (B276).This thought is echoed in some of the claims made by relational theorists.50And Allais claims that although he [Kant] does not explicitly situate his viewin terms of the theories of perception we discuss today, [in the Refutationof Idealism] he clearly commits himself to a key part of the direct realist orrelational position.51My purpose here is not to evaluate these considerations but to considerwhat implications they have for the debate between conceptualist and non-conceptualist interpretations. Allais appeals to such considerations in moti-vating her non-conceptualist reading52 and it is common to find those whoreject Allaiss non-conceptualism also rejecting her case for a relational aspectto Kants account of perception.53 One might think, then, that a relationalaccount of Kantian perception entails a non-conceptualist interpretation,47 Allais, Idealism Enough: Response to Roche, p. 380.48 ibid, p. 381.49 Allais, Kants Argument for Transcendental Idealism in the Transcendental Aesthetic,pp. 5862.50 Gomes, A. (forthcoming) Kant and the Explanatory Role of Experience, Kant-Studien.51 Allais, Idealism Enough: Response to Roche, p. 382.52 Allais, Kant, Non-Conceptual Content and the Representation of Space, pp. 38792.53 Ginsborg, Was Kant a nonconceptualist?. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 17and vice versa: that conceptualist interpretations require ascribing to Kanta non-relational account of experience. In what follows I will show that this ismistaken.Why might one think that there is a link between the conceptualistnon-conceptualist debate and Kants account of perception? I suggested in 1 thatonemight think that the reasons that support a conceptualist interpretation alsosupport ascribing toKant a representational account of perceptual experience.Let us formulate this line of thought by means of the following argument:(1) Intuitions involve acts of synthesis.(2) Acts of synthesis combine a manifold in accordance with rules.(3) If an act of synthesis is undertaken by the understanding, then the manifoldis combined in accordance with concepts.(4) All acts of synthesis are undertaken by the understanding.(5) The manifold of intuition is combined in accordance with rules (from 1and 2).(6) The manifold of intuition is combined in accordance with concepts (from3, 4, and 5).(7) If the manifold of intuition is combined in accordance with concepts, thenintuition represents the world as being a certain way.For the purposes of this discussion, let us accept that if (7) is true, then perceptualexperience, for Kant, involves representational properties. And I take (6) to beequivalent to the conceptualist claim discussed in 2 and 3.Non-conceptualists respond to this argument by rejecting (4): they hold thatsome acts of synthesis are not undertaken by the understanding. This allowsthem to reject both (6) and (7): they can deny both that the application ofconcepts is necessary for experience and that perception involves the presenceof representational properties. I have provided some reason above for thinkingthis rejection to be problematic. If this were the only way to take seriouslythe relational considerations voiced by Allais, we would be wise to dispute hermotivations.Conceptualists endorse (1)(6). This allows them two options for acceptingthe relational considerations. First, they can reject (7). Or secondly, they canclaim that (7) is itself compatible with ascribing to Kant a relational theory ofperception. It is this second option that I will set out here.To see that (7) is compatible with a relational theory of Kantian perception,we need to make some further distinctions. It is common in the philoso-phy of perception literature to distinguish strong and weak representationaltheories of perceptual experience. Strong representational theories hold thatall of the phenomenal properties of perceptual experiences are representa-tional properties. Weak representational theories hold that at least some ofthe phenomenal properties of perceptual experiences are representational at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/18 ANIL GOMESproperties. This latter position leaves open the possibility of perception pos-sessing non-representational phenomenal properties. Such an option is fa-miliar to us from discussions about whether experiences have qualia, sincequalia are non-representational sensational phenomenal properties of expe-rience, but weak representationalism itself is silent on the nature of the non-representational phenomenal properties that it allows.An analogous distinction applies to relational theories of perception. Strongrelational theories hold that all of the phenomenal properties of perceptual ex-perience are non-representational relations; weak relational theories hold thatat least some of the phenomenal properties of perceptual experiences are non-representational relations. Some nave realists explicitly commit themselves tostrong relational theories54 but there are no grounds for thinking that all mustdo so. As Soteriou puts it:Those who appeal to non-representational properties in their account of the consciouscharacter of experience need not deny that experiences have intentional contents withveridicality conditions . . . . They might hold that the obtaining of the relevant psycho-logical but non-representational relation is an element of the conscious character ofsuccessful perception.55Weak relational theories allow that there may be non-relational elements tothe phenomenal character of perception.Once we have made these distinctions, we can see that weak representa-tional and weak relational theories are perfectly compatible: one can allowthat perceptual experience has both representational and relational phenom-enal properties. This is because possession of any one phenomenal propertydoes not exclude the possibility of having any of the others.56 On such amixed view, the phenomenal properties of perceptual experience involve bothrepresentational and relational elements. Thus a commitment to the pres-ence of representational phenomenal properties does not require a rejectionof relational phenomenal properties.What are the implications for the argument above? The claim in (7) takesa stand on whether experience, for Kant, has representational properties. Itthus commits Kant to some form of representational theory. But it is silent onthe question of whether experience also has relational phenomenal properties.It is thus compatible with ascribing to Kant a weak relational theory.How does this bear on the considerations appealed to by Allais? That willdepend on whether the considerations appealed to support the claim thatperceptual experience has nothing other than relational elements, or whetherthey support only the weaker claim that perceptual experience has, for Kant,54 Brewer, Perception and Its Objects, p. 89.55 Soteriou, M. (2010) Perceiving Events, Philosophical Explorations, 13: 22341, p. 225.56 Martin, M. G. F. (1998) Setting Things Before the Mind, in A. OHear (ed.) Current Issuesin Philosophy of Mind, pp. 15779. Cambridge: Cambridge University Press. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/KANT ON PERCEPTION 19at least some relational component. I take it to be clear that they only do thelatter: so long as Kantian experience has some relational component, we canaccount for all the features that move Allais. And this means that one who doesfind such considerations suasive has grounds only to ascribe to Kant a weakrelational theory of perception. And a weak relational theory is compatiblewith (7).The result for conceptualist interpretations is that even if they are com-mitted to the claim that perceptual experience has, for Kant, representationalphenomenal properties, this is perfectly compatible with holding that such ex-perience also has, for Kant, relational phenomenal properties. They are thuscompatible with ascribing to Kant a weak relational theory.Is conceptualism compatible with strong relational theories? That dependson the prospects for rejecting (7). I will not explore that option here, but itis worth noting that (7) is not without question. There are many activitiesthat are conceptual in the sense of requiring the application of concepts, theoutcomes of which are not themselves conceptual or representational. (Thinkof building a car.) And it is not clear why combining the manifold of intuitionin accordance with concepts requires intuition itself to possess representationalproperties, as opposed, say, to some other mental state. At the very least, moreneeds to be said about what is involved in (7) if we are to be confident thatconceptualism is incompatible with a strong relational theory.None of this speaks in favour of ascribing to Kant a relational accountof perceptual experience. My intent has been simply to show that concep-tualist interpretations need not take a stand on this issue. Consideration ofthe structure and aim of the Transcendental Deduction gives reason to rejectnon-conceptualism: Kant held that the application of concepts is necessaryfor the perceptual presentation of empirical particulars. But we can rejectnon-conceptualism whilst remaining neutral on how best to capture Kantsaccount of perceptual experience.57Trinity College, Oxford, UK Anil Gomes57 Many thanks to Rory Madden, Ian Phillips, Craig French, and two anonymous refereesfor conversations and comments on this paper. at Indiana University Libraries Technical Services/Serials Acquisitions on June 27, 2014http://pq.oxfordjournals.org/Downloaded from http://pq.oxfordjournals.org/