m. pardo - the field of evidence and the field of knowledge

Upload: goncalobargado

Post on 07-Mar-2016

5 views

Category:

Documents


0 download

DESCRIPTION

M. Pardo

TRANSCRIPT

  • MICHAEL S. PARDO*

    THE FIELD OF EVIDENCE AND THE FIELD OF

    KNOWLEDGE

    (Accepted 14 July 2004)

    One is often bewitched by a word. For example, by the word know.Ludwig Wittgenstein 1

    I. INTRODUCTION

    The trial is fundamentally an epistemological event. We wantjurors and judges to know. And we want to know that theyknow. And we also want to know the conditions when theyknow, and when, if at all, these conditions obtain. In this articleI explore the relationship between epistemological theory andevidence law. Despite recent denials of the relevance of suchtheory, 2 I show its relevance and importance by demonstrating

    * Visiting Assistant Professor, Northwestern University School of Law(20032004); Visiting Assistant Professor, Chicago-Kent College of Law(20042005). My thanks for many helpful comments to Ronald Allen,Robert Burns, Jules Coleman, Anthony DAmato, Andrew Koppelman,Corinne OMelia, and Dennis Patterson, as well participants of the 2004Stanford/Yale Junior Faculty Forum (where a version of this article waspresented in the Jurisprudence and Philosophy category) and facultyworkshops at Northwestern and Chicago-Kent. My thanks also to theanonymous referees of this journal for their helpful comments.

    1On Certainty, G.E.M. Anscombe and G.H. von Wright (eds.); DenisPaul & Anscombe (tr.) (Oxford: Blackwell, 1969), p. 57e.

    2 See, e.g., Roger C. Park, Grand Perspectives on Evidence Law, Vir-ginia Law Review 87 (2001), p. 2067 (There is a question, for workadaylawyers and law professors, whether understanding the ins and outs ofGettier, Quine, and Goldmanis worth the time and eort. We all havesome sort of explicit or implicit idea of a general theory of evidencelawand the question is whether a grander theory will help achieve thegoals of evidence scholars.).

    Law and Philosophy (2005) 24: 321392 Springer 2005DOI 10.1007/s10982-004-4999-6

  • how current conceptual issues in epistemology intersect with,explain, and clarify conceptual issues in the law of evidence.

    But why exactly is knowledge important in this context?What does it add beyond truth or having reasons? Truth aloneis not good enough. True beliefs can be accidental. A coinci-dentally true factual conclusion, without any evidence to sup-port it, is not one the legal system should or would endorse. Adecisionmakers simply having good internal reasons is notgood enough either. One could have all sorts of what appear tobe good reasons from that persons perspective but unlessthose reasons have some connection with reality the conclu-sions they engender are likewise not ones the legal systemshould or would endorse. Witch trials, for example, were oftenfounded upon what appeared to those involved to be goodinternal reasons; namely, (false) empirical generalizations aboutbehavior and evidence being indicative of witchcraft.3 Somaybe we need both truth and good internal reasons?

    But even this will not be good enough in certain circum-stances. Consider the following example. Two ocers plantcocaine on an automobile driver, and they then give unrebuttedtestimony at the drivers trial that they found the cocaine aftera consensual search of the car. The driver, concerned about hisprior record coming out on cross-examination, does not testifyand oers no real defense. The factnder convicts after ndingthe ocers credible. Now, unbeknownst to everyone save thedefendant, he really did have cocaine in the car that never wasdiscovered. Thus the factnders conclusion that the defendanthad cocaine is true and the factnder has good reasons (thetestimony). Are you satised by the conviction? Or is theresomething missing? The something missing is knowledge. Nomatter what else was going on here, the factnder certainly didnot know the driver had cocaine in the car; the factndersconclusion was true but only accidentally so. The fact that abelief is not accidentally true makes it knowledge, and

    3 See Jack B. Weinstein, Some Diculties in Devising Rules for Deter-mining Truth in Judicial Trials, Columbia Law Review 66 (1966), pp. 23243.

    MICHAEL S. PARDO322

  • knowledge (or the best possible approximation of it given thecircumstances) is what a fair and just legal system demands.

    My general thesis is that a focus on knowledge non-acci-dentally true belief 4 and the related concept of epistemic jus-tication illuminates laws evidentiary practices and goals.Epistemological theory allows us to better understand thesepractices and goals and to articulate the epistemic commitmentsand choices involved. This general thesis is supported with dis-cussions of specic issues in evidence law. The point is not justthat this philosophical theory is benecial and useful (though itis) for thinking about evidence law; rather, it is that evidence lawis already implicitly and necessarily relying on it. The theorysimply makes explicit the epistemic issues already involved.

    The philosophical discussions in the rst twoparts of the paperare cashed out in the third with respect to three dierent areas.First, the theory allows us to conceptualize the processing ofjuridical evidence and makes explicit laws epistemic commit-ments. Once made explicit, these commitments provide acoherent explanation and rationale formany practices including,for example, the standards for summary judgment and judgmentas a matter of law in civil cases, the standard for review ofjudgments in criminal cases, and the right to counsel. Thesecommitments are then compared historically and with alterna-tive modern systems federal laws current epistemologicalcommitments with respect to factnder conclusions (which willbe explained as a form of weak internalism) are more in line withcontemporary epistemic understanding than the unnecessarilystrong internal commitments imposed on early common-law andmodern civil-law factnders. These unnecessarily strong com-mitments also prop-up the recent and philosophically sophisti-cated model of jury decisionmaking with respect to scientic

    4 I refer to knowledge as non-accidental true belief because it appearsto be a general, relatively non-controversial description that captures whatmost accounts of knowledge have in common and on which most episte-mologists would agree. I am not suggesting that this description provides thenecessary and sucient conditions for knowledge.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 323

  • expert testimony put forth by Scott Brewer.5 Epistemologicaltheory reveals why Professor Brewers model is false with regardto jury decisionmaking and, due to its incorrect epistemicassumptions, does not lead to the philosophical conclusions hesupposes. Themodel collapses when these incorrect assumptionsare removed and is thus an unwarranted attack, on epistemo-logical grounds, on the competence of jurors.

    Second, epistemological theory is used to articulate a posi-tive theory of probative value. The probative value of evidencein the juridical-proof context is shown to be a conclusion thatrefers to the strength an evidential premise provides an infer-ential conclusion in a particular multi-premise context. Thisview better explains probative value than the current view thatattempts to explain probative value in terms of likelihood ra-tios. In passing, this view further claries two misunderstand-ings about probative value, which are reected in the debatesabout statistical evidence. It shows that judgments cannot beepistemically justied by appealing to the probative value oftypes or kinds of evidence or by appealing solely to a supposedcausal connection or explanation between evidence and anunderlying event.

    Finally, epistemological theory makes plain and then clariesa constitutional puzzle, which is apparent but seldom recog-nized in the cases and commentary, regarding the review ofFirst Amendment facts and the Seventh Amendments Reex-amination Clause. The Courts decision in Bose Corp. v. Con-sumers Union of U.S. Inc.,6 has created analytical and doctrinalconfusion regarding how appellate courts are to review FirstAmendment facts found by juries in a manner consistent withboth Boses command (in a bench-trial case) to conduct inde-pendent review and the Reexamination Clauses command notto reexamine facts found by a jury. Epistemological theoryhelps to esh out the issue and then show why (1) if appellatejudges have superior knowledge on an issue the standard ofreview does not matter, and (2) if they do not have such

    5 Scott Brewer, Scientic Expert Testimony and Intellectual Due Pro-cess, Yale Law Journal 107 (1998), p. 1567.

    6 466 U.S. 485 (1984).

    MICHAEL S. PARDO324

  • knowledge then independent review reexamines found facts.The solution oered is a tentative one: the Court must decidewhether the problem is illusory or the First Amendmentsomehow trumps the Seventh Amendment. The analysis con-tributes by laying the issue bare; the path to take will not bedecided analytically, but rather by weighing competing interestsand values regarding the right to a jury and the First Amend-ment.

    The general idea of a relationship between epistemology andevidence law is not new. Indeed, the connection between thetwo elds has its own interesting history. The remainder of thisintroduction provides some of the history of this connection asthe context out of which the later discussions arise.

    Jeremy Bentham asserted pithily that the eld of evidence isno other than the eld of knowledge.7 In one sense, the moreimportant one, his assertion was exactly right; in another sensehe was wrong, although for reasons he could not likely haveforeseen. He was exactly right because epistemology the eldof knowledge provides analysis of the appropriate, andoverlapping, concepts and theoretical framework for thinkingabout the law of evidence and the related issues that clusteraround the process of juridical proof. These concepts include,in addition to knowledge and evidence8 themselves, for exam-ple, proof, truth, justication, warrant, belief, reliability,coherence, reason, and so on. Benthams no other, however,was much too narrow because epistemology itself at least tothe extent it once engaged in the project of discovering a priorior constructing an infallible foundation against radical skepti-cism9 is now more robust in scope and less autonomous infocus. Epistemologists today, while still focusing on thetraditional problem of skepticism, are also engaged in a sepa-rate but continuous investigation of knowledge with such elds

    7 An Introductory View of the Rationale of Judicial Evidence, Works VI(5)(1843) (John Bowring ed.; 2002).

    8 See, e.g., Peter Achinstein, The Book of Evidence (Oxford: OxfordUniversity Press, 2001).

    9 More on this shortly.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 325

  • as cognitive psychology, sociology, and the physical sciences.The eld of evidence, then, is also these elds.10

    A recent symposium on New Perspectives on Evidence re-ects these observations.11 On the one hand, the angles and dis-ciplines represented helpfully illuminated various aspects ofevidentiary and related practices, and helped to theorize thisrelatively under-theorized area in law. Yet, on the other hand, ona fundamental level undergirding these perspectives, there is anunavoidable and irreducible conceptual dimension. Evidencescholarship, as William Twining has observed, has too oftenneglected this conceptual dimension.12 This is unfortunate be-

    10 It is also those elds, such as economics, that provide insights into issuesother than knowledge. See, e.g., Ron A. Shapira, Saving Desdemona,Cardozo Law Review 22 (2001), p. 1771; ChrisWilliam Sanchirico, CharacterEvidence and the Object of Trial, Columbia Law Review 101 (2001), p. 1227;Richard A. Posner, An Economic Approach to the Law of Evidence,Stanford Law Review 51 (1999), p. 1477; Richard D. Friedman, EconomicAnalysis of Evidentiary Law: An Underused Tool, An Underplowed Field,Cardozo Law Review 19 (1998), p. 1531; but see Craig R. Callen, Othellocould not Optimize: Economics, Hearsay, and Less Adversary Systems,Cardozo Law Review 22 (2001), p. 1791; Richard Lempert, The EconomicAnalysis of Evidence Law: Common Sense on Stilts, Virginia Law Review 87(2001), p. 1619; Ron A. Shapira, Economic Analysis of the Law of Evidence:A Caveat, Cardozo Law Review 19 (1998), p. 1607.

    11 The symposium articles are collected in Volume 87 of the Virginia LawReview.

    12 William Twining, Rethinking Evidence: Exploratory Essays (Evanston,Ill: Northwestern University Press, 1990), p. 2. One article at the above-mentioned symposium, from a leading evidence scholar and a leading legalphilosopher, addressed this dimension in order to provide an appropriatetheoretical framework and philosophical foundation or ground beneaththe feet of evidence law. Ronald J. Allen and Brian Leiter, NaturalizedEpistemology and the Law of Evidence, Virginia Law Review 87 (2001), p.1492. In response, another leading evidence scholar questioned the impor-tance of this theorizing, see Park, supra note 2, and the assumed level ofphilosophical competence among readers at which the rst two were writing,see id. (I doubt that many lawyers can handle the authors references to whatphilosophers have said about each other without more foundation Thisauthor would have beneted from a more accessible treatment of the sub-ject.). In addition to the specic conclusions advanced, part ofmyaim is showwhy such theorizing is important without assuming much familiarity amongreaders of the relevant texts or concepts.

    MICHAEL S. PARDO326

  • cause the proliferation of useful empirical work on the process ofproof is necessarily founded upon and dependent on the above-mentioned epistemic concepts.

    In addition, there is some irony in the limited amount of explicitepistemological discussion in modern evidence scholarship. Classicevidence treatise writers from Gilbert through Wigmore rou-tinely included discussions of contemporary epistemological theo-ries. Some examples: Gilberts The Law of Evidence (published in1754), which discusses the epistemological views of John Locke13;Glassfords An Essay on the Principles of Evidence and TheirApplication to Subjects of Judicial Enquiry (published in 1820),which discusses the Scottish commonsense philosophers ThomasReid and Dugald Stewart14; Stephens A Digest of the Law ofEvidence (published in 1876), which discusses the views of JohnStuart Mill; and Wigmores famous treatise15; not to mentionBenthams own work on the subject.16 As Barbara Shapiro puts it,

    13 Although not as relevant for my purposes, the inuence went the otherway as well. For example, Lockes epistemology, in book IV of his EssayConcerning Human Understanding (ch. 16, sec. 10), makes use of the com-mon-law evidence rules excluding hearsay and copies of documents inarguing for the unreliability of such evidence in general. For a discussionand critique of Lockes epistemological views on testimony see C. A. J.Coady, Testimony (Oxford: Oxford University Press, 1992), pp. 199223.

    14 For a discussion of Glassfords book see M.Y. Abu-Hareira, An EarlyHolistic Conception of Judicial Fact-Finding, Juridical Review (1986), p.79. Both Abu-Hareiras article and Glassfords book are discussed in Mi-chael S. Pardo, Comment: Juridical Proof, Evidence, and PragmaticMeaning: Toward Evidentiary Holism, Northwestern University Law Re-view 95 (2000), pp. 422424.

    15 John H. Wigmore, Treatise on the System of Evidence in Trials at theCommon Law (19041905). See also John H. Wigmore, The Principles ofJudicial Proof (1913).

    16 See Twining, supra note 12, at pp. 3291 (discussing Gilbert, Bentham,Stephen, Wigmore, and others). For additional examples, see Barbara J.Shapiro, To a Moral Certainty: Theories of Knowledge and Anglo-American Juries 16001850, Hastings Law Journal 38 (1986), pp. 175191.Thayer was the exception to this pattern: Because Thayer distinguished theevidentiary method of the law from that of the natural sciences and history,it is not surprising that he did not attempt to ground his treatise on theepistemological treatises of the day. Id. at 190.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 327

  • One of themost interesting features of the treatises on evidenceis that so many authors found it necessary to place their treatmentsof legal evidence in the context of current epistemologicalthought.17 Yet, epistemology during these periods was largelyconceived as an a priori endeavor focused on the problem ofskepticism and thus, for reasons discussed below, was to someextent a mismatch for evidence scholarship. Now, also for reasonsdiscussed below, epistemologys focus is directly relevant to evi-dence theory, but it is not a primary focus of such theory, at leastnot explicitly.18 Thus historically there may be an inverse rela-tionship between relevance and perceived importance.

    Evidence doctrines focus on empiricism (e.g. testimony in the language ofperception) may to some extent be a remnant of this theorizing, whichfocused to a large extent on the classic empiricists. See William Twining,Theories of Evidence: Bentham and Wigmore (Stanford: Stanford UniversityPress, 1986), pp. 118; Peter Tillers, Webs of Things in the Mind: A NewScience of Evidence, Michigan Law Review 87 (1989), p. 26 (reviewingDavid Schum, Evidence and Inference for the Intelligence Analyst (1987))(Under the rubric of relevancy, the leaders in traditional evidencescholarship and almost all of their followers subscribed to a theory ofinference impregnated with an epistemological perspective characteristic ofnineteenth-century British empiricism.); see also Shapiro, supra, at p. 175(It seemed essential to them [the classic treatise writers] to ground the rulesof evidence, which were the bulk of the treatises, on what was considered tobe a sound theory of knowledge.) This is not to say that the turn towardempiricism was, for the most part, a bad thing. See W.V. Quine, FiveMilestones of Empiricism, in Theories and Things (1981), pp. 6772. Nordoes this deny the inuence of rationalist epistemological assumptions aswell. See L. Jonathan Cohen, Freedom of Proof, in William Twining andAlex Stein (eds.), Evidence and Proof (New York: New York UniversityPress, 1992), pp. 323.

    17 Shapiro, supra note 16, at p. 175.18 Given the necessary connection between the two elds epistemological

    issues are implicit in almost all evidence theory. Notable exceptions thatpresent explicit epistemological discussions include, e.g., L. Jonathan Co-hen, The Probable and the Provable (Oxford: Oxford University Press,1977); Allen and Leiter, supra note 12; Craig R. Callen, Hearsay andInformal Reasoning, Vanderbilt Law Review 47 (1994), p. 43; Peter Tillers,Mapping Inferential Domains, Boston University Law Review 66 (1986), p.883; Keith Burgess-Jackson, An Epistemic Approach to Legal Relevance,St. Marys Law Journal 18 (1986), p. 463; within the philosophy-of-science

    MICHAEL S. PARDO328

  • But if so, not without exceptions. Jerome Michael summedup succinctly the importance of such discussion for law:

    Since propositions are actual or potential knowledge, since proof or disproof isan aair of knowledge, since, if they are truthful, the parties to legal controversyassert, and witnesses report, their knowledge, and since knowledge is of varioussorts, you need some knowledge about knowledge.19

    Twining cites Michael over forty years later and makes thesame point.20

    My goals, then, are rst to provide some accessible knowledgeabout knowledge, and second to show how such knowledgeyields valuable understanding of and implications for the law ofevidence and the process of juridical proof. Section II providesadditional background regarding epistemology and focuses onboth traditional epistemological projects and recent develop-ments. Building on the background in Section II, Section III thenexplicates the recent work of three philosophers: LaurenceBonJour, AlvinGoldman, andRobert Brandom.My reasons forfocusing on these particular authors are threefold. First, theywellrepresent current and important conceptual issues in the eld ofknowledge. Second, as discussed in Section IV they illuminate anumber features in the eld of evidence law. Third, they serve themeta thesis mentioned earlier; namely, they esh out theunavoidable conceptual (as opposed to empirical) dimension andconcomitant need for epistemological theorizing in both elds. In

    context and its relation to scientic evidence see Brewer, supra note 5, at p.1567; Brian Leiter, The Epistemology of Admissibility: Why Even GoodPhilosophy of Science Would Not Make for Good Philosophy of Evidence,BYU Law Review (1997), p. 805; Heidi Li Feldman, Science and Uncer-tainty in Mass Exposure Litigation, Texas Law Review 74 (1995), p. 1. Andspecic epistemological issues are often discussed in the literature concern-ing the applicability of Bayes theorem, conrmation theory, and statisticalevidence.

    19 Jerome Michael, The Elements of Legal Controversy (West Publishing,1948), p. 7.

    20 Twining, supra note 12. See also Allen and Leiter, supra note 12, at p.1492 (The developments in epistemology have not, to date, been muchnoted in legal scholarship.)

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 329

  • other words, they show that knowledge ascriptions occur, inWilfrid Sellarss famous phrase, in the logical space of reasons:

    The essential point is that in characterizing an episode or a state as that ofknowing, we are not giving an empirical description of that episode or state; weare placing it in the logical space of reasons [original emphasis] 21

    This Section rst provides the broad context in which their viewsarise: the debates between internal and external epistemologicaltheories. Next, this Section describes a developed internal theory,BonJours coherentism, and a developed external theory, Gold-mans reliabilism. Finally, this Section describes ve implicationsBrandom draws from these recent epistemological debates. Ifthere is a pernicious caricature of philosophical theory as theproduct of starry-eyed, armchair-bound, empirically unin-formed, and just generally out-of-touch-with-reality reection there is an equally caricatured (and complementary) view thatonce all the physical facts, described in some specialized vocab-ulary, have been found then all the issues regarding knowledgeand evidencewill be solved. Section III is an antidote to the latter.

    Section IV turns to law. After rst making general observa-tions regarding the relevance of epistemology to evidence law,this Part examines in depth the three areas discussed in theopening paragraphs: the laws epistemology with respect to fac-tual decisionmaking, the nature of probative value, and thepuzzle regarding the review of First Amendment facts.

    II. THE FIELD OF KNOWLEDGE

    An obvious point, but: no matter what else you care about, youcare about knowledge. It is universally valuable.22 As anindividual matter, to achieve any goal or end, you need to knowseveral facts about the world and the way it works, includingwhat those goals and ends are, when and whether you achievedthem, and perhaps even when and whether you know that. Andas a social matter, your ascriptions of knowledge to others, and

    21 Wilfrid Sellars, Empiricism and the Philosophy of Mind (Cambridge,MA: Harvard University Press, 1997), p. 76.

    22 See Hilary Kornblith, Knowledge and its Place in Nature (Oxford:Oxford University Press, 2002), pp. 160161.

    MICHAEL S. PARDO330

  • theirs to you, are likewise valuable. You want to rely on theknowledge of others in your own reasoning, as true conclusionsfrom which to infer other true conclusions, and have otherslikewise rely on you. These concerns and realizations trigger thesecond-level question of what they have in common. Well,knowledge; but whats that? This question and its related issuesanimate the eld of knowledge.

    One way to begin to think about knowledge is, strange as itperhaps may rst seem, with intent. In one sense knowledge isthe ipside of intent. Both involve some sort of connection be-tween mind and world; however, the direction of t and con-comitant authority run in opposite directions knowledge runsfrom world to mind; intent runs from mind to world. To illus-trate this point consider G.E.M. Anscombes example of thegrocery shopper and the detective.23 Imagine a shopper with agrocery list (eggs, milk, etc.) lling a shopping cart with items onthe list. Meanwhile a detective is following the shopper andwriting down a list of the objects being purchased. After theshopper has checked out, if there is a discrepancy between theitems purchased and the shoppers list, the error is with the itemspurchased. But if there is a discrepancy between the detectiveslist and the items purchased, the error is with the detectives list.The rst case illustrates intent, the second knowledge.

    A second way to enter the eld of knowledge is-with the concept of knowledge itself.24 The classicaltripartite conception of knowledge is as justied, true,25

    23 G.E.M. Anscombe, Intention (Oxford: Blackwell, 1957), pp. 5657.24 Kornblith, supra note 22, at pp. 6163, however, suggests that focus on

    the concept of knowledge as opposed to knowledge itself, which he thinks ofas a natural kind like rocks and trees, is responsible for directing epistemologyin less fruitful directions. Taking knowledge to be a natural kind is a minorityview, however. See also Timothy Williamson, Knowledge and Its Limits(Oxford: Oxford University Press, 2002) (arguing that knowledge should beused to explain justication and belief, not the other way around).

    25 The concept of truth has played a feature role in much philosophicaldiscussion, but neither epistemology nor law requires a complicated the-ory of truth. See Alvin I. Goldman, Knowledge in a Social World (Oxford:

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 331

  • belief.26 The traditional focus has been on the rst element thenature of justication or warrant. In other words, some sort ofreason to ascribe knowledge to a knowing agent, following the

    Oxford University Press, 1999), pp. 4168; Michael Williams, Do We(Epistemologists) Need A Theory of Truth? Philosophical Topics 14(1986), pp. 223242; Mirjan Damaska, Truth in Adjudication, HastingsLaw Journal 39 (1998), p. 289. By truth throughout, I refer to its basic,minimal notion as descriptive success or accuracy. Goldman, supra at p.60. Much philosophical theorizing has involved trying to explain thisintuitive notion in more basic terms; the theories generally dier onwhether truth is a realist or antirealist notion, and the accounts dier onwhether they are trying to explain the semantic meaning or pragmaticsignicance of truth. Donald Davidson, however, has argued that phi-losophers should stop trying to search for more basic notions that purportto explain the meaning of truth; rather, because truth is a primitive no-tion we already understand (in minimal realist terms), we should focus onhow that concept can explain more problematic ones, such as meaning.Donald Davidson, The Structure and Content of Truth, Journal of Phi-losophy 87 (1990), pp. 279328; see also Donald Davidson, Truth Reha-bilitated, in Robert Brandom (ed.), Rorty and His Critics (Oxford:Blackwell, 2000), pp. 6574. (Cf. Williamson, supra note 24, which arguesthat knowledge should be treated as a basic concept and used to explainothers.) For other classic and representative accounts in these debates seeAlfred Tarski, The Semantic Conception of Truth, Philosophical andPhenomenological Research 4 (1944), pp. 341375; Bertrand Russell, Truthand Falsehood, The Problems of Philosophy (1962) (Oxford: OxfordUniversity Press, 1998), pp. 119130; F.P. Ramsey, Facts and Proposi-tions, in George Pitcher (ed.), Truth (Englewood Clis, NJ: Prentice-Hall,1964), pp. 1617; J.L. Austin, Truth, in Id., pp. 1831; P.F. Strawson,Truth, in Id., pp. 3253; Michael Dummett, Truth and Other Enigmas 124 (1978), pp. 124; Hilary Putnam, Two Philosophical Perspectives,Reason, Truth and History (Cambridge University Press, 1981), pp. 4974;W.V. Quine, Pursuit of Truth (Cambridge, M: Harvard University Press,1992), pp. 7782. While epistemology and law do not need complicatedaccounts of truth, it should be noted that often the desire to avoid makingtruth epistemically inaccessible may tempt the positing of implausiblemetaphysical theories of truth. On the integration of ontological andepistemological accounts see Christopher Peacocke, Being Known (Oxford:Oxford University Press, 1999), pp. 112.

    26 This conception is often rst attributed to Plato in the Meno, whichpresents the view that knowledge is true belief plus some connection or tiebetween the belief and its truth. See, S Marc Cohen et al. (eds.), Readings inAncient Greek Philosophy: From Thales to Aristotle (Indianapolis, IN:Hackett Pub. Co. Inc., 1995), pp. 185211.

    MICHAEL S. PARDO332

  • intuitive belief that someone who believes a proposition thatcoincidentally just happens to be true, without having anyreason at all for believing it is true, cannot be said to know it.The point here is that we want to rule out accidentally truebeliefs as knowledge.

    Given this focus on justication, a classic problem arises: theinnite regress of justication (and of reasons and of evidence).To be epistemically justied in believing a proposition, an agentmust have some reason or evidence for believing the proposi-tion to be true. But then why is the agent justied in believingthe underlying reason or evidence to be true? It must itself bebased on further reasons or evidence. And so on, and on, andon. The traditional response, from Descartes through to theearly-twentieth-century movement known as logical positiv-ism, was to ground knowledge on some kind of given, self-evident, or otherwise indubitable foundation. Various candi-dates were put forth, from Descartes cogito and other incor-rigible experiences to the classical empiricists sense data to thelogical positivists combination of veriable propositions andanalytic statements. For a variety of reasons, these projectsfailed. But one primary reason was because of the realizationthat all of our observations are already shaped by our concepts,and thus cannot occur prior to having those concepts.27

    27 More specically, a purported foundation either already presupposesacquired concepts and hence is not basic and infallible, or it lacks concep-tual content and hence cannot serve as a premise from which to infer furtherknowledge. This was demonstrated by exposing that these foundationalaccounts relied on oversimplied and problematic notions of the relation-ships between mind-world, language-world, and mind-language. For theprincipal attacks see Sellars, supra note 21 (exposing the myth of the given);W.V. Quine, Two Dogmas of Empiricism, From a Logical Point of View,2nd edn. (Cambridge, MA: Harvard University Press, rev. 1994), pp. 2046.Similar views are expressed by Wittgenstein in The Philosophical Investiga-tions, G.E.M. Anscombe (tr.), (Oxford: Blackwell, 1953), and more recentlyand in more developed form by Donald Davidson, see, e.g., Subjective,Intersubjective, Objective (Oxford: Oxford University Press, 2001), pp.107121, 205220, and John McDowell, Mind and World (Cambridge, MA:

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 333

  • One favorable consequence of these failed projects was todirect focus within the theory of knowledge away from theproblem of incorrigible or infallible foundations and towardboth antifoundational approaches to the traditional problemof skepticism and the exploration of other knowledgeproblems.28 These problems relate, for example, to knowledgesnature, limits, values, virtues, and relations to other conceptssuch as action, as well as toward a rethinking of epistemologicalmethodology itself.29 Some of these and other issues are dis-cussed below. In the next section, before turning to legal set-tings, I discuss the recent30 philosophical works of BonJour,Goldman, and Brandom. After that, I map insights from theirwork and epistemology generally onto evidentiary and otherlegal issues. But before doing either, and to complete thisbackground section and provide context for the next, twofurther twentieth-century developments need to be discussed.

    The rst is Edmund Gettiers seminal paper Is Justied TrueBelief Knowledge? which demonstrated that in certain cir-cumstances the justied-true-belief model was insucient toestablish knowledge.31 This was huge.He essentially smashed thegenerally accepted view of the sucient conditions for knowl-

    Harvard University Press, 1996). For general discussions of the failedfoundational projects see Michael Williams, Groundless Belief: An Essay onthe Possibility of Epistemology (Princeton: Princeton University Press,1999); Richard Rorty, Philosophy and the Mirror of Nature (Princeton:Princeton University Press, 1979).

    28 Maybe redirect focus is more accurate, as some were pursuing theseother knowledge problems prior to the rise of contemporary naturalizedepistemology. See Philip Kitcher, The Naturalists Return, PhilosophicalReview 101 (January 1992), pp. 53114. Moreover, naturalism may refer toa variety of (and varying degrees of) both methodological and substantivecommitments such that many philosophers such as Hume, Spinoza, andNietzsche fall under the label. For an excellent discussion see Brian Leiter,Nietzsche on Morality (2002), pp. 326.

    29 See id. Epistemological projects may be normative, descriptive, evalu-ative, or ameliorative. See Jonathan M. Weinberg, Shaun Nichols, andStephen Stich, Normativity and Epistemic Intuitions, Philosophical Topics29 (2001), p. 429.

    30 Recent in the context of philosophy.31 Analysis 23 (1963), pp. 121123.

    MICHAEL S. PARDO334

  • edge, a view that stretched back at least to Plato.32 (The conse-quences of this demonstration, however, are not clear,33 and aqualication sometimes seen in contemporary discussions issimply to assume away Gettier situations.34) As may alreadybe clear, a legal example of a Gettier situation would be thehypothetical presented in the Introduction in which two policeocers stop an automobile and plant cocaine on the driver.Again, at the driver-defendants trial, both ocers testify con-sistently that when they approached the car they saw the driverattempting to hide the cocaine under the car seat. A factnderconcludes that both ocers were credible witnesses. The defen-dant chooses not to testify (he has a number of prior convictionsfor crimes involving dishonesty and is worried that his convic-tions may be used to impeach his credibility35). The defense,unable to impeach the ocers, puts on no other defense. But unbeknownst to both ocers, the factnder, and everyone elsesave the defendant, at the time he was stopped the defendantactually was concealing cocaine under his seat in the car. Thefactnder comes to believe that the defendant possessed cocaine.This belief is justied (based on the ocers consistent, unre-butted testimony) and the belief happens to be true. Yet thefactnder did not know the defendant possessed cocaine. Theupshot: truth and justication are independent variables; we can

    32 See supra note 26.33 One possible implication is that knowledge is a prime state and not the

    conjunction of an internal and external one. See Williamson, supra note 24,at pp. 6592. Weinberg et al., supra note 29, present preliminary evidencethat they argue suggests intuitions regarding Gettier situations may beculturally dependent, which they then use to question the reliance on intu-itions at all in epistemological theorizing. Gettier situations are cases at themargins, however, and so situations where intuitions are likely to vary.Weinberg et al. would need to show variations in more intuitively clear casesto undermine reliance on them. The easy cases provide the starting point forepistemological theorizing.

    34 See, e.g., Laurence BonJour, The Structure of Empirical Knowledge(Cambridge, MA.: Harvard University Press, 1985), p. 5; Williams, supranote 27, at pp. 512.

    35 See Federal Rule of Evidence 609(a)(2).

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 335

  • coincidentally believe something that happens to be true for thewrong reasons.36

    The second development is the turn toward naturalizedepistemology. Naturalism is, following Brian Leiters classi-cation, the distinctive development in philosophy over of thelast thirty years [original emphasis] or so.37 In epistemologythe classic locus is W.V. Quines Epistemology Naturalized.38

    To paint with a broad brush over some complicated-but-for-my-purposes-here-irrelevant details, Quines argument wentroughly as follows. The traditional epistemological project from Descartes to the foundational empiricism of the logicalpositivists attempted to construct some sort of rst philos-ophy that would provide an independent and suitable foun-dation for our knowledge, including our scientic practices.39

    Accordingly, it would be circular for this project to appeal toscience, as this was part of what the project was attempting toprovide a foundation for in the rst place. But, this project was

    36 The examples Gettier presented were narrow; however, Gettier situa-tions are considered more broadly to include any situation, such as the one inthe above text, where one infers a true conclusion from false premises. SeeWilliams, supra note 27, at pp. 512; Michael Williams, Problems of Knowl-edge (Oxford: Oxford University Press, 2001), pp. 2829, 5052; GilbertHarman, Thought (Princeton: Princeton University Press, 1973), pp. 2123.

    37 Brian Leiter, Naturalism and Naturalized Jurisprudence, in Brian Bix(ed.), Analyzing Law (Oxford: Oxford University Press, 1998), pp. 79104.See also Brian Leiter, Naturalism in Legal Philosophy, Stanford Ency-clopedia of Philosophy (http://plato.stanford.edu/entries/lawphil-natural-ism/) (2002).

    38 W.V. Quine, Ontological Relativity and other essays (New York:Columbia University Press, 1969), pp. 6990.

    39 See, e.g., the papers collected in A.J. Ayer (ed.), Logical Positivism (NewYork: Free Press, 1959). The logical positivists attempted foundationwas notan attempt at metaphysics; indeed, they were rabidly anti-metaphysical, andtried to construct a foundation for scientic knowledge by logically analyzinglanguage in order to separate meaningful (i.e. empirically veriable) state-ments from meaningless ones. See P.M.S. Hacker, Wittgensteins Place inTwentieth-Century Analytic Philosophy (Oxford: Blackwell, 1997), pp. 3966.

    MICHAEL S. PARDO336

  • a failure.40 Therefore, no reason prevents an appeal to sciencein our epistemic theorizing. In other words, epistemology nolonger needs to be an a priori endeavor; it can now engage in aposteriori investigation along with the sciences and other dis-ciplines and, most important, rely on this additional evidence inassessing our epistemic practices. In short, philosophical theorycan (and should) rely on and be constrained by empiricalinformation.

    While doing so, however, epistemology may still remain alargely evaluative, normative endeavor.41 Following HilaryKornbliths scheme, (1) the traditional epistemologist asks howwe ought to form beliefs; (2) the empirical psychologist askshow we do in fact form beliefs; (3) the naturalized epistemol-ogist (along with others) asks if the epistemic practices by whichwe form beliefs are the means by which we ought to, and howthose practices (typically evidence gathering and reasoning)might be improved.42 Two general considerations constrain

    40 Thanks partly to Quines own previous work in attacking logicalpositivisms reliance on the analyticsynthetic distinction. See Quine, supranote 27. The rejection of the analyticsynthetic distinction more speci-cally, the notion that even our analytic truths could be revised based onexperience implies that epistemology could not only make use of scienticresults but that such results might also be used to revise our epistemologicalmethodologies and concepts as well.

    41 See Kitcher, supra note 28; Larry Laudan, Normative Naturalism,Philosophy of Science 57 (March 1990), pp. 4459. See also Kornblith, supranote 22, at p. 177 (describing philosophy as empirically informed theoryconstruction.) An exception is the purely descriptive epistemological projectsometimes attributed to Quine. See Jaegwon Kim, What is NaturalizedEpistemology? in Hilary Kornblith (ed.), Naturalizing Epistemology(1994), pp. 3355; Leiter, supra note 37. But this view is not clearly explicitin Quines original paper; he refers to reciprocal containment, not justepistemology as a chapter of descriptive science. See Quine, supra note 38, atp. 83. See also W.V. Quine, Reply to Morton White, in Hahn and Schilpp(eds.), The Philosophy of W.V. Quine (1986), pp. 664665 (Naturalization ofepistemology does not jettison the normative and settle for the indiscrimi-nate description of ongoing procedures.); Kornblith, supra note 22 at 137140.

    42 Hilary Kornblith, Introduction: What is Naturalistic Epistemology?,in Kornblith (ed.), supra note 41, at p. 14. See also Kitcher, supra note 28;Lauden, supra note 41.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 337

  • such theorizing.43 The rst is instrumental: practices, processes,etc. are evaluated for their tendency to produce true beliefs andto avoid error or ignorance.44 The second constraint recognizesthe practical limitations on potential knowing agents; it rec-ognizes that ought implies can. For example, a normativetheory may recommend one maintain consistency among onesbeliefs, but the theory would violate this constraint if one has somany beliefs that the computational complexity involved inkeeping them consistent was impossible to perform.45 Or amore extreme example that violates the constraint: one mightrecommend that the best epistemic strategy for jurors is totravel back in time and witness the event giving rise to thelawsuit.

    Even if one rejected a naturalized approach to episte-mology, such an approach (in the normative sense) is stillappropriate for epistemological theorizing within the law. Aprimary reason for rejecting a naturalized approach is be-cause one takes the skeptical threat to be the problem forepistemology; therefore, so the argument might go, it wouldbe circular to appeal to science in answering the skeptic

    43 Leiter oers these constraints in discussing jurisprudential theories.Leiter, Naturalized Jurisprudence, supra note 37. Moreover, because legalquestions are not ontologically or epistemologically distinct from factualquestions, see Ronald J. Allen and Michael S. Pardo, The Myth of the Law-Fact Distinction, Northwestern University Law Review 97 (2003), p 1769,there is no obvious dierence in the application of epistemological theory.This is not to deny the possibility of drawing a distinction at a higher levelbetween natural facts and social facts, with legal questions concerningexamples of the latter. But this distinction would not explain the distinctionbetween legal and factual adjudicative questions. See id. Much of thefactual subject matter involved in legal disputes also involves social factssuch as monetary damages, intent, duty, breach, the meaning of contractlanguage based on parol evidence, and so on.

    44 Error and ignorance are, of course, two dierent variables. One couldpossess epistemic practices that produced only true beliefs, though theyproduced so few of them that they would not allow that person to do orknow much of anything.

    45 See Christopher Cherniak, Minimal Rationality (Cambridge, MA: TheMIT Press, 1986). See also Christopher Cherniak, Computational Com-plexity and the Universal Acceptance of Logic, in Kornblith (ed.), supranote 41, at pp. 239260.

    MICHAEL S. PARDO338

  • because this scientic knowledge is precisely what the skepticwanted a justication for in the rst place. Radical skepti-cism, however, has no place in the law. The very act oftheorizing about laws epistemic practices presupposes ashared, public, knowable world. Without this supposition,the notion of law itself is unintelligible.46 (The same pointabout skepticism goes for radically subjective notions oftruth also.47) Thus, although much of the following discus-sion will concern conceptual rather than empirical issues, Iassume that any epistemological theorizing about juridicalproof or law in general must be naturalized in the sensediscussed above.48

    A nal word about skepticism. Even if it turned out that,given our actual or even potential epistemic practices, we werenever justied in believing any of our beliefs, epistemologicaltheorizing would still be relevant to evidence law. One thing itwould reveal is what practices, beliefs, judgments, etc. werebetter or worse, more or less justied than others. Given thataccurate factual judgments are obviously important to any justlegal system, a legal systems epistemic practices even if onlyin matters of degree and even if those degrees are insucient toqualify as justied or knowledge are of critical impor-tance. This is all the more so given that juridical proof is acomparative-judgment aair.49 So, in the law, then, we are inan ineluctable epistemic situation.

    46 See Damaska, supra note 25; Hilary Putnam, Pragmatism and Real-ism, Cardozo Law Review 18 (1996), p. 153. If skepticism holds, then we donot even know that there is law or underlying factual disputes to begin with,or anything to even dispute for that matter. There would be nothing for lawto do. Skepticism is a conversation stopper in this context.

    47 These, though, are often unintelligible on their own terms. Goldman,supra note 25, at pp. 340. See also supra note 25; Terry Eagleton, AfterTheory (New York: Basic Books, 2003), p. 104 (those who do not believe intruth are quite often inverted dogmatists. They reject an idea of truth thatno reasonable person would defend in the rst place.)

    48 For general discussions of naturalism with regard to jurisprudence andlegal theory see Leiter, sources cite in supra note 37.

    49 Ronald J. Allen, Factual Ambiguity and a Theory of Evidence,Northwestern University Law Review 88 (1994), p. 604; Ronald J. Allen, TheNature of Juridical Proof, Cardozo Law Review 13 (1991), p. 373.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 339

  • The story told in this Part can be summarized as follows:contemporary epistemology, for philosophical and historicalreasons, has emerged as a robust, empirically informed eldinvestigating not only traditional problems of skepticism, butalso knowledge issues on the ground, as it were. The knowledgeissues on the ground are those that arise after we accept theminimal-realist assumptions (ones evidence law must make)that the world exists independently of our senses and knowl-edge of it is possible. The works discussed in the next sectionarise within this context. They all to some extent accept a realistconception of truth50 and are exploring the connection of whatseparates knowledge from accidentally true beliefs. They dierin how to understand and explain this connection.

    III. CONCEPTUALIZING KNOWLEDGE

    This section examines contemporary works by LaurenceBonJour, Alvin Goldman, and Robert Brandom.51 I rstdiscuss the broad context in which their epistemological

    50 This is not to suggest that there arent important dierences betweenthem, primarily between Brandom on one hand and Goldman and BonJouron the other. Brandom is dicult to pin down on this point, given hiscomplex discussion of truth in Making it Explicit: Reasoning,Representing,and Discursive Commitment (Cambridge, MA: Harvard University Press,1994), pp. 299305. He might best be referred to as having a pragmaticrealist theory of truth.

    51 Specically, BonJour, supra note 34; Laurence BonJour and ErnestSosa, Epistemic Justication: Internalism vs. Externalism, Foundations vs.Virtues (Oxford: Blackwell, 2003) (This book is written in sections and re-sponses by BonJour and Sosa, not as co-authors; here I focus on BonJourssections); Goldman, supra note 25; Alvin I. Goldman, Epistemology andCognition (Cambridge, MA: Harvard University Press, 1986); Alvin I.Goldman, Discrimination and Perceptual Knowledge, Journal of Philoso-phy 73 (1976), pp. 771791; Alvin I. Goldman, What is Justied Belief?, inKornblith (ed.), supra note 41, at pp. 105130; Robert B. Brandom, Artic-ulating Reasons: An Introduction to Inferentialism (Cambridge, MA: Har-vard University Press, 2000); Brandom, supra note 50. BonJour andGoldman are concerned with epistemology proper; Brandoms discussion ofepistemology concerns its relation to his project in philosophy of language,primarily his account of semantic content, which he explains in terms ofinference.

    MICHAEL S. PARDO340

  • views arise: internal versus external theories. Then, in orderto esh out the distinction, I explicate a developed internalisttheory, BonJours coherentism, and a developed externalisttheory, Goldmans reliabilism. Finally, I discuss some epis-temological conclusions that Brandom draws from thesediering views and their relationship, which he refers to asinsights and blindspots. After that, I turn to the law.

    A. Internalism and Externalism

    Internalist and externalist epistemological theories of provideaccounts of what makes an agents true belief or judgmentjustied or warranted. They also provide accounts of whatmakes true beliefs or judgments qualify as knowing, that is, notaccidentally true. They dier in how much the accounts dependon what is internal to the agents cognitive awareness.

    Internalist theories require that the reasons or factorsneeded to epistemically justify an agents belief are accessibleto the agent, are internal to the agents cognitive perspective.These theories come in two avors: strong and weak. Stronginternalism requires that the knowing agent actually be awareof the reasons that make her belief justied, that, if chal-lenged, she is able to articulate those reasons. Weak inter-nalism requires just that the knowing agent be capable ofbecoming aware of the justifying reasons by properly focus-ing his attention, that he has within his cognitive grasp thereasons for thinking his belief is true.

    This perspective accords with our intuitive, commonsensenotion that belief acceptance must be in some sense rational; inother words, that an agent, at least when claims to knowledgeare at stake, bears some responsibility in how she formulatesbeliefs or judgments that purport to be true.52 The primaryobjection to such theories is that they demand too much ofagents by confusing rst- and second-order ascriptions ofknowledge. An agent can know something without knowing

    52 In others words, this needs to be the case for a belief or judgment to bejustied for this person. See BonJour and Sosa, supra note 51, at p. 24.Hilary Kornblith, Justied Belief and Epistemically Responsible Action,Philosophical Review 92 (1983), p. 33.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 341

  • that she knows. By requiring that agents engage (or in weakversions be capable of engaging) in this higher-level reectiveprocessing in order to have knowledge, the internalist, contraryto our commonsense notions, must deny that inarticulateadults, pre-verbal children, and certain animals53 have knowl-edge, because they cannot provide the justifying reasons, citethe relevant evidence.54

    By contrast, externalist theories account for justication andknowledge through factors external to the knowing agentscognitive awareness. Under these theories, an agents belief isjustied if it was produced for reasons or via a process ormethod that makes it objectively likely to be true. In otherwords, the theories explain both justication and knowledge byappealing to regularities or the reliability of various practicesfor producing true beliefs. The knowing agent need not beaware of this regularity or reliability relation, nor have reasonsfor thinking her belief is true, yet so long as the belief orjudgment was formed in a reliable way, the agent is still said tobe justied and to have knowledge of her true belief.55

    The externalist theories accord with our commonsenseintuitions that whether a belief or judgment counts as knowl-edge depends not just on what is going on in the agents

    53 There are thorny debates about whether non-discursive animals or pre-discursive children have beliefs. See, e.g., Davidson, supra note 27, at pp.95105. There is a weaker sense of know, however, that refers to some typeof appropriate responsiveness to an environment. An animal respondingappropriately may be said to know how to move about, in a way that is non-accidentally appropriate, even if this knowing how is not also a knowingthat. See Kornblith, supra note 22, at pp. 2869. On the distinction betweenknowing-how and knowing-that see Gilbert Ryle, The Concept of Mind(1949), pp. 2561. For a lucid and illuminating recent discussion of Witt-genstein and Sensations, Animals, and Knowledge see David H. Finkel-stein, Expression and the Inner (Cambridge, MA: Harvard University Press,2003), pp. 128152.

    54 The internalist theories may also be too weak because, in certain cir-cumstances, features external to the agent cause us to withhold ascribingknowledge even though the agent has strong reasons for believing a truebelief. Gettier situations, described above, are one example; another will bediscussed later in this section.

    55 For example, a judge or police ocer with reliable but unexplainableintuitions.

    MICHAEL S. PARDO342

  • consciousness but also depends on features of the environment.Also, unlike internalism, externalism does not require thatpotentially knowing agents be able to cite all the evidence orreasons that justify the true propositions of which the agentshave knowledge. Pure externalist theories, however, ascribeknowledge even to potentially knowing agents who act in anepistemically irresponsible manner. The standard example(BonJours56) is someone who believes himself to be clairvoy-ant, but a more realistic, legally relevant one would be as fol-lows. Imagine a judge who believes himself reliable at makingcredibility determinations based on demeanor and forms abelief based on this putative ability, yet he has (1) strong evi-dence that such determinations are in general unreliable57; (2)some (misleading) putative evidence that his recent suchdeterminations are unreliable (some defendants he recentlyconvicted based on such determinations were exonerated whensomeone else confessed (falsely)); (3) no evidence that they arereliable; (4) strong evidence that the belief in question is false(signicant evidence corroborates a witness he believes is notcredible). Nevertheless, the judge forms and accepts the belief,which happens to be true, based on demeanor, and as it turnsout, his ability to judge credibility based on demeanor is quitereliable. The externalist theory would say the judge was justiedand had knowledge even though he ignored all the evidence hehad. Thus the agent can behave quite irrationally and irre-sponsibility with respect to the evidence and still be epistemi-cally justied under a pure externalist theory.58

    My point in discussing these two perspectives is not to suggestthat one is correct or even that a choice need be made betweenthem.Rather, each illuminates features of the nature of epistemicjustication and knowledge, which, in Section IV, I use to makeexplicit laws epistemic practices regarding juridical proof andprovide a vocabulary for discussing those practices. Indeed, thevarying perspectives seem to be directing attention toward dif-

    56 BonJour, Structure, supra note 34, at pp. 4145.57 Perhaps he read and found persuasive, Olin Guy Wellborn III, De-

    meanor, Cornell Law Review 76 (1991), p. 1075.58 It severs justication from rationality, which seems, at least, prob-

    lematic. See BonJour and Sosa, supra note 51, at p. 29.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 343

  • ferent epistemological issues: they show that crucial epistemo-logical issues arise, and that justication likely depends upon,both an agents backgroundbeliefs, reasoning, and rationality onone hand as well as on how a belief or judgment was generatedand the external world on the other.59 This will become moreclear is next section, which eshes outmore-developed internalistand externalist theories. In passing, it is worth noting that thisdiscussion bears more than a supercial resemblance to H.L.A.Harts distinction between the internal and external aspects ofrules.60 A stop-sign provides a regularity from the external per-spective that allows one to predict when people will stop based onan apparent causal relationship between the sign and peoplestopping; but from the internal perspective, the same sign pro-vides a reason, a justication, for stopping. Likewise, the evi-dence, process, or method that, from the external perspective,justies a belief bymaking it objectively likely to be truemayalso,from the internal perspective, be a reason, evidence, for acceptingor retaining the belief.

    B. Coherence and Reliability

    BonJour presents an internalist theory that accounts for epi-stemic justication in terms of coherence.61 It is a traditionaltheory in that it takes skepticism as the primary epistemologicalquestion, consequently attempts to address the redress problem,and rejects naturalism in doing so. To be clear, the signicancefor the discussion here is not the success of the theory inanswering the skeptic or that the skeptical challenge is evenrelevant. Rather, the theory provides a well-developed example

    59 Or they may be attempting to articulate dierent concepts or dierentsenses of the same content. For discussions of this possibility see BonJourand Sosa, supra note 51, at pp. 3541; Williams, supra note 27, at pp. 186188 (in Afterword to 2nd edn.).

    60 H.L.A. Hart, The Concept of Law, 2nd edn. (Oxford: Oxford Univer-sity Press, 1997), pp. 8291.

    61 BonJour, Structure, supra note 34. Harman, supra note 36, also relies toa signicant extent on internal coherence. Bonjour discusses several previouscoherence theories in Structure at pp. 212229. See generally Ken Kress,Coherence, in Dennis Patterson (ed.), A Companion to Philosophy of Lawand Legal Theory (Oxford: Blackwell, 1999), pp. 533152.

    MICHAEL S. PARDO344

  • of an antifoundational, internal theory; and the concept ofcoherence, while almost certainly not the whole story, is relevantat least at some minimum level to any theory whatsoever.62

    Rather than nding some self-evident foundation on whichto ground knowledge, the coherence theory accounts forknowledge in term of the coherence of a given set or system ofbeliefs. A belief is justied to the extent it coheres with the set orsystem of beliefs. The theory answers the skeptical challenge byrejecting as myth the idea of linear justication, namely, thateach belief is justied by an independent succession of beliefs,like cars on a train.63 Rather, our beliefs frequently stand inmutual justicatory and inferential relations with each other,and it is this network that provides warrant for accepting orrejecting potential claims. The dependence (hence the defeasi-bility) of beliefs on each may be extreme (any belief dependsupon and may be defeated by any other or group of themwithin the system) or, more plausibly, may be moderate in thatbeliefs come in mutually supporting chunks or clusters.

    BonJour articulates the following minimal elements for acoherentist theory of justication. In addition to some level oflogical consistency64, the system must employ a sucient levelof inferential, evidentiary, and explanatory relations, eviden-tiary and explanatory relations being particularly importanttypes of inferential connections.65 Accordingly, coherence isincreased by the amount and the strength of the inferentialconnections between component beliefs, and it is decreased bythe presence of unexplained anomalies.

    The primary objection to the theory as explicated thus far isthat there could a number of extremely coherent, alternative

    62 On the importance of coherence for rationality and intelligibility seeDavidson, A Coherence Theory of Truth and Knowledge, in supra note 27,at pp. 137157.

    63 Kornblith rejects the linear view as the argument-on-paper thesis.Hilary Kornblith, Beyond Foundationalism and the Coherence Theory, inKornblith (ed.), supra note 41, at pp. 133134.

    64 For a naturalist viewpoint, a requirement of complete consistencywould violate the ought-implies-can requirement; it would be virtuallyimpossible. See Cherniak, supra note 45.

    65 Bonjour, Structure, supra note 34, at pp. 93101.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 345

  • systems without any contact with reality. Thus there is no reasonto think the beliefs justied in the sense of having any claim tobeing true, which is precisely what the theory should tell us. Tomeet this objection, BonJour adds in the requirement thatcoherence is a dynamic not static aair, and that a systemmust becontinually receiving a sucient amount of input from the worldthat must then be assimilated, causing varying degrees of re-sponse and occasional change in the system.66 Alternate systemsout of touch with reality will quickly come to lack the inferentialand explanatory relations when impinged upon by the informa-tional input from the outside world. Systems of belief that aremore or less coherent given this inputwill bemore or less justied,as will the component beliefs, which will in turn be more or lesslikely to be true. In others words, to oversimplify somewhat,there would be a continual process of inference to the bestexplanation67 in a global-belief sense.

    Now, the coherence theory is generally seen as a failure tothe extent it attempts to answer the traditional regress problemand respond to the possibility of radical skepticism. (EvenBonJour has conceded its failure to that end, and he has sincebegun to advocate a version of foundationalism to account forempirical knowledge.68) It failed primarily because coherencecannot do all the epistemological work by itself.69 Even if

    66 Id. at pp. 139146.67 See Gilbert B. Harman, The Inference to the Best Explanation,

    Philosophical Review 74 (January 1965), pp. 8895; Peter Lipton, Inferenceto the Best Explanation (London: Routledge, 1991); see also Harman, supranote 36, at p. vii (Coherence is a matter of explanation).

    68 BonJour and Sosa, supra note 51, at pp. 6096. Michael Williamsargues for a middle way between foundationalism and coherence, contex-tualism, see Unnatural Doubts: Epistemological Realism and the Basis ofScepticism (Princeton: Princeton University Press, 1996) and supra note 27,at pp. 159172.

    69 At least to the extent it attempts to show that justied beliefs are likelyto lead to truth in a realist, correspondence sense. One response, whichBonJour rejects, would be to posit a coherence theory of truth. BonJour,Structure, supra note 34, at pp. 157188. But this would, as he recognizes, domore harm than good by positing an implausible account of truth andreality in order to t a problematic epistemological theory. The t shouldgo the other way. See Peacocke, supra note 25.

    MICHAEL S. PARDO346

  • people (implausibly) could be aware of all or a sucientnumber of their beliefs to make a coherence assessment, aquestion would then arise as to whether an agent adequatelyknew their own mental states, which would be necessary beforecoherence could even be assessed. And even if these beliefscould be suciently known the beliefs about the beliefs(including the belief that the beliefs are known) would need tobe justied, and so on, and on.70 But, notwithstanding theselimitations, the coherence theory helps to reveal the holistic(whether extreme or moderate) nature of rational, internal be-lief formation. Although BonJours theory rejects naturalism, itresembles on a global scale what naturalized epistemology andcognitive psychology reveal about how we assimilate and drawinferences from a mass of evidence at the local level. Namely,by schematically organizing a mass of information in a way tomaximize coherence with coherence being determined byfactors beyond consistency such as completeness (absence ofanomalies), explanatory and other inferential relations, sim-plicity, and so on.71 Jurors are just one example of this largerphenomenon of drawing inferences to the best explanation.72

    From within the naturalist camp, motivated in part by thefailure of internalist theories, including coherentism, to answerthe traditional skeptical challenge, reliablism seeks to refocusepistemological discussion onto the question of which processes,methods, practices, etc., from an external perspective, are likelyto lead to objectively true belief, and to explain justication and

    70 So the regress again arises. BonJour originally posits a doxasticpresumption to stop this regress; essentially, justication rests on aconditional presumption that we adequately know the content of our ownthoughts. See BonJour, Structure, supra note 51, at pp. 101106. But henow explains further problems with even this adjustment. Id. at pp. 5659.

    71 See, e.g., John H. Holland, Keith J. Holyoak, Richard E. Nisbett, andPaul R. Thagard, Learning Inferential Rules, p. 359, and Richard Nisbettand Lee Ross, Judgmental Heuristics and Knowledge Structures, p. 282,both in Kornblith (ed.), supra note 41.

    72 See, e.g., Nancy Pennington and Reid Hastie, A Cognitive Model ofJuror Decision Making: The Story Model, Cardozo Law Review 13 (1991),p. 519.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 347

  • knowledge in that manner.73 Reliabilism, thus, seeks to containthe internalist requirement of an agents being aware of the rel-evant factors (having evidence or reasons) as being just one of thepotentially many ways that a true belief could be non-acciden-tally true and hence qualify as knowing.74 One way to show abelief was reliably produced or is reliably accepted is for the agentto be able to provide the justifying evidence or reasons that makeit likely to be true. But reasons are not the onlyway to account forknowledge; reliability allows one from an external perspective tosee an agents true belief as not produced by accident, but ratheras expected or predictable based on what the external observerknows about the reliable connection.75

    Goldman has espoused the most developed and recognizedreliabilist account, and therefore it is his theory on which Ifocus.76 A simple reliabilist theory that a belief is justied orqualies as knowledge if it was caused in a reliable way doesnot by itself tell us very much (other than that reliability is moreimportant that reasons). Goldman builds on this simple ac-count by exposing that reliability is in a certain sense arbitrarybecause it will always be subject to a variety of interests andqualications. It is only with respect to a baseline (xed bythese interests and qualications) that we can assess whether abelief was formed in a reliable way, whether the agent is justi-ed, and whether the agent knows. In other words, Goldmansexamples show that reliability is subject to a generalityproblem.

    For example, consider perception. This is a (perhapsthe) generally accepted, relatively non-controversial way to

    73 Goldman, Discrimination, supra note 51 (arguing this focus betteraccounts for our commonsense use of know.) Goldman still recognizes thetraditional, skeptical question as legitimate but downplays its importance.See Goldman, Epistemology, supra note 51, at pp. 2841.

    74 And the reverse is true. The internalist theory purports to contain theexternal: a reliable process of belief acceptance is one of the reasons an agentcould cite for why her belief is justied.

    75 Again, think of Harts distinction between the external and internalaspects of rules, supra note 60.

    76 See the Goldman sources cited in supra note 51. See also Kornblith,supra note 22, for a recent naturalized, reliabilist account.

    MICHAEL S. PARDO348

  • form reliable beliefs. But does this mean that every beliefformed via perception is justied, and, if true, known? No. Ifwe want to know whether an agents belief is justied, wewant to know whether her perception produces beliefs thatare objectively likely to be true. And if her perceptual beliefsare generally true, does this mean every belief she forms viaperception is justied, and, if true, known? No. We wouldwant to know under which conditions her perception pro-duced mostly true beliefs (in good lighting, at a close dis-tance) and when they did not (at night, without her glasses).But if we keep specifying more detailed conditions, to sift thereliable from the unreliable, eventually we will arrive at onlyone case of perception, the very one we are trying to assess.77

    This collapses the epistemological inquiries of justicationand knowledge solely into a question of whether the belief istrue. So there would be no point of inquiring into reliabilityany more. In most cases (and this is primarily the case in thejuridical-proof context) we want to know whether a beliefwas reliably formed in order to assess what do not know andwant to know whether the belief is likely to be true. Inshort, there will always be some relativity to reliability basedon how close the reliability baseline generalizes the particularcase under examination.

    In addition to generality issues, reliability inquiries also faceissues regarding relevant alternatives. Reliability turns on morethan whether a particular belief was caused by conditions thatmake the belief true or likely to be true. To illustrate this point,Goldman presented his famous barn example.78 Imagine anagent with good eyesight who, driving in his car, passes a farmand under good perception conditions sees a barn, and thusbelieves and reports to a distracted passenger look, theres abarn. The belief is true, he just passed a barn, and this factcaused him to perceive he was passing a barn. So, does he knowit? Maybe. It depends, says Goldman, on other external factors,

    77 See Goldman, Epistemology, supra note 51, at pp. 4951; Goldman,What is Justied Belief?, supra note 51, at pp. 105130.

    78 Goldman, Discrimination, supra note 51, at pp. 772773. For a similarexample involving thermometers see Goldman, Epistemology, supra note 51,at p. 45.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 349

  • regardless of whether the agent is aware of them.79 Imaginefurther the agent was traveling through an area where they justlove to make fake barn fronts (perhaps to trick passingmotorists), which from the road are indistinguishable from realbarns. The agent, however, did see a real barn, but it was theonly real barn in an area with hundreds of fake barn fronts.Under these conditions, even though his belief is true and wascaused by what would normally be reliable circumstances, it isonly accidentally true that his belief is true in these circum-stances because he would have formed the same belief even if hesaw a fake barn. So he doesnt know.80 Thus, for a belief to bereliably generated (and hence, if true, knowledge) the agent orprocess must be able to discriminate among relevant counter-factual situations that could potentially undermine the reli-ability assessment.81 In other words, an agent may qualify ashaving knowledge of a belief only if the agent has the belief viaa method suciently sensitive enough that it would not gen-erate the belief were the belief not (likely) true.82

    Goldmans example can be translated into a pressingrealistic problem regarding eye-witness identications. Per-suasive psychological literature suggests that even true iden-tications made at lineups, show-ups, or photographic arraysmay be only accidentally true based on factors external tothe initial observations conditions regarding testing

    79 Goldman is attacking simpler, pure causal theories of knowledge,including his own, for failing to recognize this potential defeasibility. Dis-crimination, supra note 51.

    80 While most would agree that the agent does not know, a more dicultquestion is whether the agents belief is justied. From an internal per-spective, the agent appears to be justied because he has what appears to begood reasons and evidence (his perceptions) and no reason to believe he is ina fake-barn rich environment. From an external perspective, however, theagent may not be justied if justication is understood to be based onobjective reliability in the particular circumstances; he is just not a reliablereporter in this area.

    81 See also Harman, supra note 36, at pp. 142154 (discussing evidenceone does not possess).

    82 This is oversimplied; there are some exceptions. See Williamson, supranote 24, at pp. 147163.

    MICHAEL S. PARDO350

  • procedures, conditions, and administrators.83 Specically,even a true identication may not qualify as knowledge if thewitness would have picked the same suspect even if thesuspect was not the one the witness initially observed.

    Before turning to Brandoms views on these issues, two re-lated points sum up the contrasting theories discussed in thisand the previous sections. The internal and external perspec-tives dier on whether evidence and reasons or objective reli-ability should be the primary epistemic focus. Second, in manysituations they will converge (think again of Hart on theexternal and internal aspects of rules); an agents reliability willprovide good evidence, a reason the agent could cite for achallenged conclusion. The situations diverge when agents arereliable but are either unaware of the reliability or believethemselves to be unreliable, or vice versa.84

    C. Reliability and Reasons

    This section discusses ve implications that Robert Brandomdraws from the above-described debate between internal andexternal theories in general and the rise of reliabilism in par-ticular.85 Whereas BonJour and Goldman are primarily pur-suing conclusions for their epistemological signicance,Brandoms views arise as part of his larger project in the phi-losophy of language. Nevertheless, the implications revealimportant features of epistemological theorizing, features that

    83 See, e.g., Gary L. Wells, Elizabeth A. Olson and Steve D. Charman,Distorted Retrospective Eyewitness Reports as Functions of Feedback andDelay, Journal of Experimental Psychology 9 (2003), pp. 4252. A strand inthe Courts Sixth Amendment jurisprudence attempts to deal with thisproblem by providing defense counsel to observe and challenge these po-tential external factors. United States v. Wade, 388 U.S. 218 (1967); Gilbertv. California, 388 U.S. 263 (1967).

    84 In these situations, the reliabilist should win, at least from our per-spective. In the legal context, when objective reliability and internal justi-cation conict, the fact that a conclusion is more likely to be true is a betterreason than whether any one particular person or persons is internallyjustied or believes themselves to be.

    85 Brandom, Articulating, supra note 51, at pp. 97122;Making it Explicit,supra note 50, at pp. 199229.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 351

  • will prove useful in the next section in discussing laws epistemicpractices. The following points concentrate on the relationbetween reason-based internal theories and reliability-basedexternal theories. Brandoms primary conclusion is that reli-ability cannot replace reason-giving (or evidence-citing) as theprimary epistemological question. Indeed, reliability assess-ments must be understood as a type of reason-based activity. Ifollow his labeling.

    1. The Founding Insight

    The founding insight of reliabilism, as discussed above, is therecognition that one can have justied (i.e., in the sense of non-accidental) true beliefs without having gone through a justica-tory process of citing the evidence or reasons sucient for justi-cation.86 The insight allows us to see two dierent senses ofjustication: (1) a process of justifying, in which a belief has beenjustied when it is shown to come about or to be accounted forthrough reasoning or appeal to evidence; (2) a belief having apositive justicatory status, that is, that it could be justied viareasoning, that the agent is entitled to the belief. This secondsense arises from an observer external to the agent, with moreinformation, appealing to a reliability relation of which the agentmay or may not be aware in order to confer positive justicatorystatus on the belief. The point of the founding insight: an agentsjustifying a belief or judgment in sense one (appeals to reasons,evidence) is just one way of achieving sense-two justication.

    For example, consider a police ocer who has such an un-canny intuitive ability that she can almost always tell whichsuspects are guilty therst time she interviews them.Her intuitionis, by hypothesis, reliable in sorting guilty from innocent sus-pects. Yet she doubts these intuitions, does not put much weightin them, and always allows further evidence to be gathered beforeshe forms any kind of judgment. She thus does not rely on the

    86Brandom,Articulating, supranote 51, at pp. 97102. The founding insightthus is the ipside of theGettier situations described supra. In those situationsan agent having a justied (in an internal, having-reasons sense) true belief isnot sucient for knowledge. The founding insight shows that having a justi-ed (in an internal, having-reasons sense) belief is not necessary either.

    MICHAEL S. PARDO352

  • reliability of her intuitions as a reason for her judgments. But shecommunicates these intuitions to her partner, who does takethese intuitions as reliable evidence for whether particular sus-pects are guilty or innocent (shes almost always right, after all;she is reliable). Thus, from thepartners perspective, at the timeofthe rst interview the ocer already knew whether a particularsuspect was guilty (even if the ocer wants to wait for conrmingevidence).

    2. The Conceptual Blindspot

    The concept of reliability provides a usefulway to think about thenature of justication andhence knowledge, but it cannot explainanother of knowledges primary components: beliefs.87 This isthe conceptual blindspot of reliabilism. Consider again theintuitive ocer. Shehas knowledge because she is a reliable non-inferential indicator of guilty suspects; it is just this ability thatmakes her intuitive beliefs justied. But the concept of reliabilitycannot explain why what she has are beliefs or even judgments(or, if you prefer, the propositional content of those beliefs orjudgments). Like the intuitive ocer, such things as rusted iron,thermometers, radar detectors, breathalyzers, parrots, and drug-sning dogs may also be reliable non-inferential indicators ofsuch things as wet environments, temperature, speed, blood-alcohol level, recently uttered sentences, and narcotics.88 Yetnone of the latter possesses a propositionally contentful belief orjudgment, something that x that could even serve as a potential

    87 Brandom, Articulating, supra note 51, at pp. 106110. An epistemologistmay object that this criticism is o the mark because it is simply not the job ofepistemology to provide a theory of beliefs; rather, this a job for philosophy ofmind or philosophy of language. Thismay be correct, but Brandoms criticismnevertheless reveals potential problems for reliabilist accounts if one takesbeliefs to be a necessary component of knowledge, these accounts ascribeknowledge to objects like thermometers that lack this necessary component.

    88 Rust (a particularly nice one) and parrots are examples given byBrandom. Id. On the essential normativity of belief, see Kim, supra note 41;Donald Davidson, Thought and Talk, Inquiries into Truth & Interpretation(Oxford: Oxford University Press, 1984), pp. 15570.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 353

  • candidate for knowing that.89 All of these, including theocer, are responding dierentially to the environment, but onlythe ocer is applying concepts. That is, she possesses a propo-sition that occurs in the space of reasons;90 it is something thatcan be seen as following inferentially from other premises andleading inferentially to further conclusions, something consistentwith some propositions and incompatible with others.91 Anypropositionally contentful beliefs or judgments in the case of theother examples are those of an external observer only not of apotentially knowing agent. Thus, the reliability of the breatha-lyzer provides propositionally contentful reasons, but for anexternal observer only not for the breathalyzer. In other words,the ocers occupying a position in this inferential network ofreasons makes her a potentially knowing (in the propositionalsense) agent to begin with.92

    89 Of course, some may be said to know how to do things, for instance,detect narcotics. But the relevant knowledge under discussion, and the kindrelevant in legal-proof context, as discussed in the next section, is of thepropositional, knowing-that kind. And this requires the agent to possess abelief or judgment with propositional content, something the intuitive ocerpossesses and the others do not. This does not mean that knowing that isnot a particular kind of knowing how. Indeed, this is Brandoms view.

    90 See Sellars, supra note 21 and accompanying text. See also JohnMcDowell, Knowledge and the Internal, Meaning, Knowledge and Reality(Cambridge, MA.: Harvard University Press, 1998), pp. 395413.

    91 Brandoms primary project explains semantic content in terms of suchinferential relations. So far as I can tell, nothing Im arguing for in this articlenecessarily turns on whether one accepts Brandoms larger, semantic projectas correct. See also Williams, Problems of Knowledge, supra note 36, at p. 36(This understanding consists in the human reporters grasp of what followsfrom his reports, what is evidence for them, how they might be challenged,how challengesmight bemet, and so on. Propositionally contentful utterancesand genuine beliefs are, in Brandoms phrase, inferentially articulated. Theyare essentially the sorts of things that can function as reasons and for whichreasons can be demanded and given. If this is even roughly right, theremust besomething misleading about reliabilist intuitions.).

    92 In a similar vein, Williamson argues that evidence is propositional, anda proposition can be evidence for one only if one grasps the proposition.Supra note 24, at pp. 184208.

    MICHAEL S. PARDO354

  • 3. Goldmans Insight

    This insight ows from the fake-barn example discussedabove.93 This example demonstrates that one can have both (1)good reasons and evidence for a true proposition, and (2) anunproblematic causal connection between the propositionstruth and an agents coming to believe the proposition (in thiscase a direct, non-inferential perception in good conditions) and still not have knowledge.94 These conditions are not su-cient because features of the environment, external to theagents awareness, can undermine the agents belief and thusdisqualify it as knowledge. In others words, these conditionsshow that, even though the agent has good reasons and evi-dence and the belief was formed with ideal causal antecedents,it is still just accidentally true given the circumstances in whichit was formed (i.e., the prevalence of fake barns). The potentialdefeasibility is Goldmans insight; however, Brandom arguesthat it reveals a more signicant implication.

    4. The Naturalistic Blindspot

    Goldmans example, Brandom argues, shows that the naturalworld (regardless of whether the facts are described in common-sense or scientic terms) cannot determine reliability (hencejustication or knowledge) because objective probability can bespecied only relative to a reference class and nothing in theway the world is privileges one of those reference classes.95

    Rather it is our interests and attitudes that do so.96 Returning tothe fake barns, recall that the unfortunate agent was deniedknowledge because of the prevalence of fake barns in the county;

    93 See supra at pp. 3132.94 Brandom, Articulating, supra note 51, at p. 115; Making it Explicit,

    supra note 50, at pp. 209210.95 Brandom, Articulating, supra note 51, at pp. 112, 116.96 This is so in the full range of real-world situations to which episte-

    mological theories apply, as opposed to the carefully idealized situationsdescribed in articially restricted vocabularies to which concepts of objectiveprobability are applied in the special sciences. Id. at pp. 112113. And, itprobably goes without saying, the epistemological issues in the law concernthe full range of real-world situations.

    THE FIELD OF EVIDENCE AND THE FIELD OF KNOWLEDGE 355

  • he was an unreliable reporter of barns in the county. But supposethat in the state in which he and his passenger were traveling realbarns outnumber fake barns by a large number; he is thus now areliable reporter in the state. But suppose that fake barns outnumber real ones in the country; he is now unreliable. Back andforth as the relevant jurisdiction expands, or collapses.97 (And allalong its the very same belief and report.) Now, of course thisexample is extreme, but it is not meant to support an empiricalclaim that we very often have these neatly nested situations withalternating assessments of reliability. Rather, the articialexample is making the conceptual point that there are alwayssome reference classes with respect to which one is reliable. [ori-ginal emphasis]98 And unreliable. What is going on as thejurisdiction expands and the agent ip-ops between being reli-able and unreliable is not some natural fact about the worldregarding the reliability of the agents report; what is going on isthat as the boundary conditions (which can be gerrymandered inall sorts of bizarre ways) of the reference class change so do theinferences drawn from the original report. Because we (oftentacitly via our interests and attitudes) provide the boundaryconditions:

    [a]n argument place remains to be lled in, and the way the worldobjectively is does not, by itself, ll it in. Put another way, the reliabilityof the belief-forming mechanism (and hence the status of its true productsas states of knowledge) varies depending on how we describe the mech-anism and the believer.99

    If Brandom is right here (and I think he is), then a purely natu-ralistic, descriptive account of objective reliability (therefore ofjustication and knowledge) cannot replace conceptual, nor-mative epistemological theorizing in accounting for knowledge.It is this conceptual, normative dimension of knowledge that

    97 But recall that trying to dene the reference class can collapse to thevery belief or report under consideration, and thus would reduce the reli-ability inquiry to a truth one. See supra pp. 3031.

    98 Brandom, Making it Explicit, supra note 50, at p. 11.99 Brandom, Articulating, supra note 51, at p. 116. Kornblith points out,

    however, that sometimes the interests pick the animals. Kornblith, supranote 22, at pp. 6369.

    MICHAEL S. PARDO356

  • provides non-empirical work for the epistemologist to do. And itdoes so as well for those theorizing about the unavoidable epis-temological aspects of evidence law. Indeed, Brandom goes fur-ther and suggests that reliability itself must be understood fromwithin this dimension, namely, as a particular type of social,inferential practice. This is the implicit insight of reliabilism.

    5. The Implicit Insight

    The implicit insight synthesizes the founding and Goldmansinsights and avoids the conceptual and naturalist blindspots byfocusing on the external observer, that is, the third-partyassessing reliability, justication, and knowledge.100 In otherwords, us. The ascription of knowledge involves a three-partsocial, inferential structure, corresponding to the three com-ponents in the traditional justied-true-belief framework. First,the assessor attributes to another a propositionally contentfuldoxastic101 commitment, a belief or judgment or reportthat. Second, in assessing whether the commitment wasreliably generated and whether it is justied the assessor isexamining whether the agent is entitled to the commitment.Third, in conferring entitlement and taking the commitment tobe a genuine case of knowledge, the assessor is endorsing thecommitment oneself, that is, taking it to be true. So what theassessor is doing is using the reliability assessment (entitlement)as evidence or a reason for the truth of the commitment. Inother words, when an assessor concludes that an agent hasknowledge, or at least is epistemically justied, the assessor can

    100 Brandom, Articulating, supra note 51, at pp. 117122. In Making itExplicit, supra note 50, at p. 217, the subheading speaks for it