Well-Founded Belief: New Essays on the Epistemic Basing Relation (Routledge Studies in Epistemology) [1 ed.] 1138503754, 9781138503755

Citation preview

Well-Founded Belief

Epistemological theories of knowledge and justification draw a crucial distinction between one’s simply having good reasons for some belief and one’s actually basing one’s belief on good reasons. While the most natural kind of account of basing is causal in nature—a belief is based on a reason if and only if the belief is properly caused by the reason—there is hardly any widely accepted, counterexample-free account of the basing relation among contemporary epistemologists. Further inquiry into the nature of the basing relation is therefore of paramount importance for epistemology. Without an acceptable account of the basing relation, epistemological theories remain both crucially incomplete and vulnerable to errors that can arise when authors assume an implausible view of what it takes for beliefs to be held on the basis of reasons. Well-Founded Belief brings together 16 essays written by leading epistemologists to explore this important topic in greater detail. The chapters in this collection are divided into two broad categories: (i) the nature of the basing relation; and (ii) basing and its applications. The chapters in the first section are concerned, principally, with positively characterizing the epistemic basing relation and criticizing extant accounts of it, including extant accounts of the relationship between epistemic basing and propositional and doxastic justification. The latter chapters connect epistemic basing with other topics of interest in epistemology as well as ethics, including: epistemic disjunctivism, epistemic injustice, agency, epistemic conservativism, epistemic grounding, epistemic genealogy, practical reasoning, and practical knowledge. J. Adam Carter is a lecturer in philosophy at the University of Glasgow, UK. His work has appeared in Noûs, Philosophy and Phenomenological Research, Analysis, Philosophical Studies, and the Australasian Journal of Philosophy. He is the author of Metaepistemology and Relativism (2016). Patrick Bondy is Assistant Professor of Philosophy at Wichita State University, USA. His work has appeared in Pacific Philosophical Quarterly, Dialogue, Synthese, and Episteme. He is the author of Epistemic Rationality and Epistemic Normativity (2018).

Routledge Studies in Epistemology Edited by Kevin McCain University of Alabama at Birmingham, USA

Scott Stapleford St. Thomas University, Canada

Pragmatic Encroachment in Epistemology Edited by Brian Kim and Matthew McGrath New Issues in Epistemological Disjunctivism Edited by Casey Doyle, Joseph Milburn, and Duncan Pritchard Knowing and Checking An Epistemological Investigation Guido Melchior Well-Founded Belief New Essays on the Epistemic Basing Relation Edited by J. Adam Carter and Patrick Bondy For more information about this series, please visit: www.routledge. com/Routledge-Studies-in-Epistemology/book-series/RSIE

Well-Founded Belief New Essays on the Epistemic Basing Relation

Edited by J. Adam Carter and Patrick Bondy

First published 2019 by Routledge 52 Vanderbilt Avenue, New York, NY 10017 and by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 Taylor & Francis The right of the editors to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilized in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. Library of Congress Cataloging-in-Publication Data Names: Carter, J. Adam, 1980– editor. | Bondy, Patrick, editor. Title: Well-founded belief : new essays on the epistemic basing relation / edited by J. Adam Carter and Patrick Bondy. Description: New York : Routledge, 2019. | Series: Routledge studies in epistemology; 4 | Includes bibliographical references and index. Identifiers: LCCN 2019044402 (print) | LCCN 2019044403 (ebook) | ISBN 9781138503755 (hbk) | ISBN 9781315145518 (ebk) Subjects: LCSH: Epistemics. | Evidence. | Knowledge, Theory of. Classification: LCC B820.3 .W45 2019 (print) | LCC B820.3 (ebook) | DDC 121/.4—dc23 LC record available at https://lccn.loc.gov/2019044402 LC ebook record available at https://lccn.loc.gov/2019044403 ISBN: 978-1-138-50375-5 (hbk) ISBN: 978-1-315-14551-8 (ebk) Typeset in Sabon by Apex CoVantage, LLC

Contents

Introduction

1

PATR I C K B O NDY A N D J. A DA M CARTE R

PART I

The Nature of the Basing Relation 1 A Doxastic-Causal Theory of Epistemic Basing

13 15

RU YE

2 All Evidential Basing Is Phenomenal Basing

34

ANDREW MOON

3 Dispositions and the Problem of the Basing Relation

53

H A M I D VA H ID

4 The Many Ways of the Basing Relation

74

L U CA M O R E T TI A N D TO MMASO P IA ZZA

5 Reasons and Basing in Commonsense Epistemology: Evidence From Two Experiments

92

J O H N TU R R I

6 Inference and the Basing Relation

113

K E I TH A L L E N KO RCZ

7 The Superstitious Lawyer’s Inference

125

PATR I C K B O NDY A N D J. A DA M CARTE R

8 Prime Time (for the Basing Relation) E R RO L L O R D A N D KURT SYLVAN

141

vi

Contents

PART II

Basing and Its Applications 9 Hermeneutical Injustice as Basing Failure

175 177

M O N A S I M I ON

10 Agency and the Basing Relation

190

R A M N E TA

11 Epistemic Conservatism and the Basing Relation

201

K E V I N M c CAIN

12 Can Beliefs Be Based on Practical Reasons?

215

M I R I A M S C HL E IFE R McCO RMICK

13 Epistemological Disjunctivism and Factive Bases for Belief

235

DU N CA N P R ITCH A RD

14 From Epistemic Basing to Epistemic Grounding

251

J E S P E R K A L L E STRUP

15 Well-Founded Belief and the Contingencies of Epistemic Location

275

G U Y A X TE L L

16 The Epistemic Basing Relation, and Knowledge-That as Knowledge-How

305

S TE P H E N H ETH E RIN GTO N

List of Contributors

325

Index

328

Introduction Patrick Bondy and J. Adam Carter

I The Basing Relation: A Brief Overview Well-founded belief is belief that is properly held on the basis of good, justifying reasons.1 Just as it often happens that people do the right thing for the wrong reasons, or fail to act when they have good reason for acting, it also often happens that people have good reasons for holding a belief but don’t hold it, or they do hold it but on the basis of some other bad reasons instead. Of course holding a belief on the basis of bad reasons does not make the belief false, just as performing an action on the basis of bad reasons does not make the action itself wrong. But in order to be fully justified, the beliefs that we hold for reasons must be held on the basis of good reasons. Epistemologists standardly distinguish between propositional justification and doxastic justification. Propositional justification is the justification a person can have for holding a belief, even if she does not hold it, or even if she holds the belief on the basis of some other bad reasons instead. Doxastic justification is full justification, the justification a belief has when it is held in the right kind of way. “Well-founded belief” and “doxastically justified belief” are not quite synonymous, but in many contexts they are interchangeable. The former is often the more natural expression to use, though it is more theoretically loaded, as it suggests a foundationalist account of the structure of justification.2 In any case, the latter is the more widely used expression in contemporary epistemology. Getting clear on doxastic justification is important in its own right, but because doxastic justification is plausibly also necessary for other epistemic statuses such as knowledge and understanding, getting clear on doxastic justification is of paramount importance for epistemology. A key ingredient in doxastic justification is the epistemic basing relation, the relation between beliefs and the reasons on the basis of which they are held. In order to understand doxastic justification, then, we need to understand the basing relation. And, with a satisfactory account of the basing relation in hand, we can go on and use that account to shed light on a variety of other epistemological issues.

2

Patrick Bondy and J. Adam Carter

The basing relation is an explanatory relation: it provides an explanation of why a subject holds a belief. But it’s not just any kind of explanatory relation. It explains why a person has a belief specifically by reference to what are often called “motivating” or “operative” reasons. To illustrate: suppose we want to explain why Jane brews herself a cup of coffee every morning. Because the caffeine in coffee is mildly addictive, in some contexts we can explain why Jane brews coffee every morning by citing the addictive nature of the drink. But the addictive nature of coffee is not normally a reason on the basis of which Jane acts when she pours her coffee. It is an explanatory reason but not a motivating reason. Jane’s (motivating) reasons would be more like, “coffee helps me wake up,” or “I like it,” or “it makes me feel ready for the day,” or something along these lines. These are the reasons which prompt Jane’s action, and which Jane sees as making the action of brewing coffee worthwhile. Of course, Jane will often not explicitly think of these reasons before she brews her coffee. We often perform actions on the basis of reasons even without explicitly calling our reasons to mind. Still, Jane has her reasons, and without those reasons she would likely not brew coffee every morning. So the basing relation is an explanatory relation which holds between beliefs and the reasons for which the beliefs are held. The task for us lies in specifying the content of that relation. Two broad distinctions will help to orient the approaches to, and the arguments about, the basing relation. The first distinction is between the activity of justifying—that is, of providing or at least being able to provide the reasons on the basis of which one’s belief is held, and which justify one’s belief—and the state of having beliefs that are based on, and are justified by, the reasons one possesses. Some epistemologists have held that having doxastically justified beliefs is a matter of being able to provide reasons which justify those beliefs. For example, Stephen Toulmin has written that We “know” something (in the full and strict sense of the term) ifand-only-if we have a well-founded belief in it; our belief in it is wellfounded if-and-only-if we can produce good reasons in its support. (1976, p. 89, italics in original) Toulmin thought that well-founded belief is sufficient for knowledge, that it requires conclusive reasons, and that it requires that one be able to produce those reasons. Similarly, Keith Lehrer has written that if a person has evidence adequate to completely justify his belief, he may still fail to be completely justified in believing what he does because his belief is not based on that evidence. What I mean by

Introduction

3

saying that a person’s belief is not based on certain evidence is that he would not appeal to that evidence to justify his belief. (1965, p. 169, italics in original) It is now widely accepted, however, that the state of having beliefs which are based on, and justified by, one’s reasons, is entirely distinct from the activity of providing or of being able to provide the reasons on which one’s beliefs are based, and which justify one’s beliefs. Keith Allen Korcz identifies several important, widely accepted aspects of this distinction:3 First, and most apparently, being justified in believing p is a state whereas showing that one is justified in believing p is an action. Second, it seems clear, for instance, that one may be justified in holding a belief even if one lacks the epistemic concepts needed to show that it is justified. Thus, one need not be able to justify one’s belief in order to be justified in holding it. Third, I could attempt to show that my belief that p is justified even if I am not justified in believing p. . . . Similarly, I might not be justified in believing p but nonetheless believe p. Under pressure to justify my belief, I could discover that I do have good reasons to believe p and perhaps become justified in believing p on the basis of those reasons. (2000, p. 533, italics in original) Two of Korcz’s points are particularly important for us here. The first is that one can have fully justified beliefs, which one holds on the basis of good reasons, even if one is unable to provide those reasons in defense of one’s beliefs. (Maybe one lacks the necessary concepts to formulate the reasons as reasons, or maybe one is simply too nervous and one tends to forget one’s reasons when asked for them, or maybe some other mechanism intervenes and prevents a person from providing her reasons.) The second point is that one might hold a belief on the basis of bad reasons at a time t1, and not even realize that there are good reasons for holding it— but then, once one is pressed to provide reasons, one immediately comes to realize at t2 that there are other good reasons available for the belief. So, at t2 one comes to base one’s belief on the good reasons, and one’s belief thereby becomes justified. The point is that at t1 one has this ability to provide good reasons for the belief, but at t1 one’s belief is not held on the basis of the good available reasons. So, Korcz argues, being justified is entirely distinct from being able to provide a justification. That is the current orthodoxy in epistemology,4 and that is the first broad distinction to help orient the debates about the epistemic basing relation. The second important distinction to draw is between causal and doxastic accounts of basing. We’ve noted earlier that the basing relation is an explanatory relation. Causal accounts of basing provide a causal

4

Patrick Bondy and J. Adam Carter

interpretation of that relation. The basic idea of a causal account is that a belief is based on a reason when the reason causes the belief. But that basic idea needs, at minimum, to be bolstered with a way to rule out deviant causal chains, because beliefs can be caused by reasons in so-called “deviant” ways, where the belief is clearly not held on the basis of the reason. For example: Suddenly seeing Silvia, I form the belief that I see her; as a result, I become rattled and drop my cup of tea, scalding my leg. I then form the belief that my leg hurts; but though the former belief is a (part) cause of the latter, it is not the case that I accept the latter on the evidential basis of the former. (Plantinga 1993, p. 69n8) Ruling out causal deviance is a challenge, and it is key to giving a satisfactory account of basing in causal terms. Causal accounts of one sort or another have been widely defended or assumed.5 An alternative to the causal approach is to give an account of basing in doxastic terms. Doxastic accounts hold that having an appropriate “meta-belief,” to the effect that a reason R is a good reason for holding a belief B, is the key to holding B on the basis of R. Believing, of some reason that you possess, that it is a good reason for a belief that you hold, seems like it’s at least sufficient—and possibly even necessary—for you to count as holding your belief on the basis of that reason.6 One worry for doxastic accounts is that they over-intellectualize the basing relation, with the result that conceptually unsophisticated agents cannot count as basing beliefs on reasons. Another worry for some epistemologists is that our beliefs can be based on reasons of which we are unaware, or which we have forgotten. If that is correct, it’s a problem for doxastic accounts of basing, because we clearly cannot have appropriate meta-beliefs regarding the quality of our reasons if we are unaware of those reasons. So we have purely causal accounts of basing, as well as purely doxastic accounts. Hybrid accounts containing both causal and doxastic conditions have also been proposed, as have other alternative approaches.7 The chapters in this volume propose novel analyses of the basing relation, new lines of argument for and objections against various analyses of the basing relation, arguments regarding what sorts of things can stand in basing relations, and interesting and important connections between the basing relation and various other issues in epistemology.

II Overview of Chapters We turn now to an overview of the volume’s chapters—16 in total—which we’ve organized into two broad categories: (i) the nature of the basing relation; and (ii) basing and its applications. The former chapters are

Introduction

5

concerned, principally, with positively characterizing the epistemic basing relation and criticizing extant accounts of it, including extant accounts of the relationship between epistemic basing and propositional and doxastic justification. The latter chapters are unified in that they connect epistemic basing with other topics of interest in epistemology as well as ethics, including: epistemic disjunctivism, epistemic injustice, agency, epistemic conservativism, epistemic grounding, epistemic genealogy, practical reasoning, and practical knowledge. This division between the nature of the basing relation and its applications is of course an imperfect one (as there will be some overlap in places), though we hope it will be helpful nonetheless as a way to navigate the volume. II.1 Part I: The Nature of the Basing Relation The book begins with a brand-new account of basing defended by Ru Ye in her chapter “A Doxastic-Causal Theory of Epistemic Basing.” The key idea of Ye’s account is that a belief is based on a reason just when two conditions hold: first, the reason must cause the belief; in this respect, the view lines up with causal accounts. But, second, and here is Ye’s theoretical novelty, that the reason causes the belief must itself be because the subject believes that the reason supports the belief. In a bit more detail, the idea is that epistemic basing is a matter of “causation caused by taking” and the taking must be a belief about evidential support. This proposal, which she calls “Causation Caused by Believing (CCB),” is then argued to both avoid deviant causation and also fit snugly with a plausible view of proper basing. In “All Evidential Basing is Phenomenal Basing,” Andrew Moon’s objective is to defend a novel necessary condition on evidential basing, that is, on what it is for beliefs to be based on evidence. On Moon’s proposal, the evidential basing relation obtains between someone’s belief and the evidence she has only if the mental state associated with that evidence has phenomenal character, where the phenomenal character of a mental state is the experiential “what it is like” to be in that mental state. Moon’s argument for this thesis—the phenomenal basing thesis—is inductive: across a wide range of cases considered, either the beliefs are not based on evidence or the mental state associated with the relevant evidence has phenomenal character. So, the phenomenal basing thesis is probably true. Hamid Vahid, in “Dispositions and the Basing Relation,” aims to carve out an alternative to causal and doxastic accounts of the basing relation, what he calls the dispositional account of the basing relation. A motivating idea in the chapter is that an adequate account of basing must avoid the kind of causal deviance objections that notoriously plague standard causal accounts. Vahid’s key move for getting around the problem is to defend a dispositional analysis of how propositional and doxastic justification are related to one another; in particular, the suggestion is that

6

Patrick Bondy and J. Adam Carter

propositional justification is an (epistemic) dispositional property that a subject can have with doxastic justification as its manifestation. Vahid then argues that this account provides the basis of an account of the basing relation that avoids deviant causal chain objections. Luca Moretti and Tommaso Piazza, in “The Many Ways of the Basing Relation,” set out to expand traditional thinking about how well-grounded beliefs must be based on reasons that give the subject propositional justification for those beliefs. They note, as a starting point, that what the basing process involves can be different depending on the kind of reason one has. For example, non-doxastic reasons (e.g., experiences) require a basing process that is immediate in a way that doxastic reasons, which require inference, do not. Moretti and Piazza’s novelty is to show that these ways of basing in cases of well-grounded beliefs are not exhaustive, and to accommodate outlier cases, they introduce what they call enthymematic inference, which corresponds with a way of basing that stands apart from the more traditional varieties. John Turri, in his chapter “Reasons and Basing in Commonsense Epistemology: Evidence from Two Experiments,” combines traditional thinking about the nature of the basing relation with non-traditional methodology. Turri begins by arguing for the importance of experimental evidence in our theorizing about the closely related notions of basing and epistemic reasons. He then reports the results of two new experiments about our concepts of both: the first experiment lends support to the causal theory of the basing relation, and the second suggests that reasons include both psychological and non-psychological items. In “Inference and the Basing Relation,” Keith Allen Korcz’s main objective is to resolve some mistaken ideas about the relationship between epistemic basing and inference, and in particular as regards the latter, whether inferences can occur only among beliefs, and whether an inference must be stated as a premise within an argument. Once these confusions about the nature of inference and its relation to basing are sharpened, Korcz argues, it will help us to appreciate among other things how basing works in cases of analytic truths, and also how to better understand the relationship between propositional and doxastic justification. We (that is, Pat Bondy and Adam Carter) in “The Superstitious Lawyer’s Inference,” offer a new diagnosis of what is the most well-known— as well as perhaps the most divisive—case in the basing relation literature: Keith Lehrer’s (1971) case of the superstitious lawyer, which Lehrer poses as a counterexample to the causal theory of the basing relation. On our view, and contra Lehrer, the superstitious lawyer case plausibly features both doxastic justification and well-founded basing, even though there are independent reasons (though not those Lehrer or others have adverted to) to think that the target belief falls short of knowledge. We round out Part I of the volume with Errol Lord and Kurt Sylvan’s chapter “Prime Time (for the Basing Relation),” which challenges

Introduction

7

received thinking about what it takes to believe something for sufficient normative reasons. Their target is what they call the “Composite View,” according to which believing something for a sufficient normative reason involves nothing more than (i) believing on the basis of a motivating reason, and (ii) that motivating reason’s corresponding to a sufficient normative reason for that belief, where (i) and (ii) are conditions that could obtain independently of each other. Seeing why the composite view is false and a prime view true, they argue, has important ramifications for our theorizing about propositional and doxastic justification, the place of reasons in epistemology and of competence in theories of doxastic justification that appeal to reasons. II.2 Part II: Basing and Its Applications Part II of the book, which connects basing with other topics of interest in epistemology and ethics, begins with Mona Simion’s “Hermeneutical Injustice as Basing Failure.” Hermeneutical injustice, a key species of epistemic injustice, occurs, according to Fricker’s (2007) influential account, when the interpretive resources available to a community render a person’s experiences unintelligible to her. Moreover, as Fricker maintains, this unintelligibility must itself be due to the epistemic marginalization of that person or members of her social group. Simion’s chapter has two main aims, one negative and the other positive. The negative aim is to show that Fricker’s account is too restrictive; it rules out genuine cases of hermeneutical epistemic injustice. Building from this criticism of Fricker, Simion then advances and defends the positive thesis that hermeneutical injustice is unjustly brought about basing failure. An important implication of unpacking hermeneutical injustice in terms of epistemic basing, Simion maintains, is that hermeneutical epistemic injustice can be appreciated as a form of distributive injustice, a point that is elided on Fricker’s more restrictive characterization. In “Agency and the Basing Relation,” Ram Neta connects epistemic basing with agency, by criticizing a particular way of thinking about the relationship between epistemic and practical agency defended by Kieran Setiya (2013). Setiya’s account of epistemic agency and its relationship to practical agency is predicated on his acceptance of a particular view of basing, according to which to believe that p on the ground that q is to believe that p and that the fact that q is evidence that p. Setiya maintains that if this is right, then, epistemic agency is unlike practical agency in that it does not involve our exercise of a capacity to cause anything—it involves nothing over and above our having certain kinds of belief. Neta argues that even if Setiya’s preferred way of thinking of basing were true, this difference between practical and epistemic agency wouldn’t follow, though he also argues that basing is in fact not what Setiya assumes it is. An upshot, Neta maintains, is that the very same

8

Patrick Bondy and J. Adam Carter

kind of agency that we find in intentional action can also be found in beliefs and other attitudes. Kevin McCain’s chapter “Epistemic Conservatism and the Basing Relation” attempts to resolve an apparent tension between the causal theory of the basing relation and epistemic conservatism, the view that having a belief confers some positive epistemic status on the content of that belief. The prima facie tension between the two views is this: it looks as though, if the causal theory of the basing relation were true, then epistemic conservativism could be true only if a belief could cause itself, which it can’t. As McCain argues, this apparent incompatibility between the causal theory of the basing relation and epistemic conservativism boils down to a misunderstanding of epistemic conservatism which, once suitably clarified, suffices to dissipate the puzzle. Miriam Schleifer McCormick’s contribution “Can Beliefs be Based on Practical Reasons?” answers her title’s question in the affirmative. It’s uncontroversial that practical reasons can contribute to what one believes in the sense that they can make a difference to what one believes. Much more controversial, though, is whether beliefs can be based on practical reasons. Some philosophers, such as Nomy Arpaly (2019; see also Thomas Kelly 2002), have gone so far as to suggest that the very idea of practical reasons for belief is a “category mistake.” After arguing that beliefs can be, and commonly are, based on practical reasons, McCormick then proceeds to defend the further thesis that in at least some cases, practical reasons can justify the beliefs that are based on them. Duncan Pritchard, in his chapter “Epistemological Disjunctivism and Factive Bases for Belief,” shows how epistemic basing interfaces with a particular way of thinking about rational support in cases of perceptual knowledge. According to epistemological disjunctivism, one’s perceptually formed belief can enjoy rational support that is both factive and reflectively accessible (see, e.g., Pritchard 2012). An important commitment of epistemic disjunctivism, Pritchard shows, is that basing is itself distinctively factive. A benefit of the kind of factive basing that epistemological disjunctivism involves is that it can be used in the service of responding to traditional problems for epistemological disjunctivism, including what Pritchard has described elsewhere (e.g., 2012) as the basis problem and the access problem. In his chapter “From Epistemic Basing to Epistemic Grounding,” Jesper Kallestrup shows how epistemic basing differs, despite some similarities, from epistemic grounding, and gives a novel account of the latter. The key difference, according to Kallestrup, comes at the level of explanation: an epistemic basis, as a result of which you know, backs causal explanation of knowledge, while an epistemic ground, in virtue of which you know, backs metaphysical (rather than causal) explanation of knowledge. On the account of epistemic grounding Kallestrup proposes and defends, epistemic grounding is a non-primitive relation of asymmetric metaphysical dependence between knowledge and its epistemic ground.

Introduction

9

Guy Axtell, in “Well-Founded Belief and the Contingencies of Epistemic Location,” takes as a starting point that many of our beliefs in controversial areas (e.g., politics, religion, etc.) are culturally nurtured—that is, historical, temporal, geographical, cultural contingencies often play a significant role in determining how our opinions in these areas take shape. Suppose we reflect on this fact and then conclude that we would be very likely to see our own nurtured belief as both false and tainted by unrecognized bias, were we to have been nurtured in a different culture or epistemic community. Should this undermine the well-foundedness of the nurtured beliefs we hold? Axtell’s answer is “no,” but this negative answer comes with a range of important qualifications, including some that are in tension with how epistemic conservativists and dogmatists will be inclined to think about the epistemic status of these beliefs. Finally, in “The Epistemic Basing Relation and Knowledge-That as Knowledge-How,” Stephen Hetherington shows how the platitude that knowledge requires basing can be unpacked on Hetherington’s own practicalist conception of propositional knowledge, according to which knowledge-that is a species of knowledge-how (see, e.g., Hetherington 2011a, 2011b). Drawing inspiration from Plato’s Statues of Daedalus analogy in the Meno, Hetherington defends the view that proper basing—a kind of “tethering” relation—is best understood as a kind of knowledge-how, nestled within knowledge-that.

Notes 1. Feldman and Conee’s influential evidentialist definition of well-foundedness, for example, goes as follows: WF: S’s doxastic attitude D toward p is well-founded for S at t if and only if (i) having D toward p is justified for S at t; (ii) S has D toward p on the basis of some body of evidence e such that: (a) S has e as evidence for p at t; (b) having D toward p fits e; (c) there is no more inclusive body of evidence e’ had by S at t such that having D toward p does not fit e’. (1985, p. 24, italics added) 2. “Well-founded belief” is also theoretically loaded in that it suggests an account of justification in terms of beliefs that are based on reasons, ruling out an externalist account of justification like Goldman’s (1979, 1986) reliabilism, according to which, roughly, beliefs are fully justified iff they are produced by reliable processes, including processes that do not take reasons for belief as inputs. Goldman calls fully justified beliefs produced by reliable processes “ex post” justified, but we can view his account as an account of doxastically justified belief. Still, because many accounts of doxastic justification do make essential use of epistemic reasons for belief, and all accounts of doxastic justification allow that at least some beliefs are justified in virtue of being held on the basis of good reasons, and this volume is about beliefs held on the basis of reasons, we

10

3. 4. 5. 6.

7.

Patrick Bondy and J. Adam Carter

don’t need to worry about this sense in which “well-founded belief” is theoretically loaded. Beliefs that are doxastically justified in virtue of being held on the basis of good reasons are well-founded beliefs. See also Harman (1970) and Alston (1985) for similar arguments against the view that the activity of justifying is part of what constitutes the state of being justified. But see Leite (2004), Bondy and Carter (2018) and Hetherington (this volume) for some pushback against this orthodoxy. See Winters (1980), Swain (1981), Audi (1983), Turri (2011), McCain (2012), and Bondy (2016) for various causal accounts of the basing relation. See Tolliver (1982) for a defense of a doxastic account of basing. Lehrer (1971) argues against causal accounts of knowledge, and his central counterexample to the causal account of knowledge (the case of the superstitious lawyer) is also naturally interpreted as an attempt to counterexample causal accounts of the basing relation, in favour of a doxastic account of basing. Setiya (2013) proposes a doxastic account of inferential basing. For example, Audi (1986) holds that causal and doxastic conditions are both necessary for basing; Korcz (2000) holds that appropriate causal and doxastic conditions are each sufficient for basing; Evans (2013) argues against both causal and doxastic approaches, and in favour of an alternative dispositionbased account.

References Alston, William. 1985. “Concepts of Epistemic Justification.” The Monist 68 (1): 57–89. Arpaly, Nomy. 2019. “Epistemology and the Baffled Action Theorist” (ms). Audi, Robert. 1983. “The Causal Structure of Indirect Justification.” Journal of Philosophy 80 (7): 398–415. Audi, Robert. 1986. “Belief, Reason, and Inference.” Philosophical Topics 14 (1): 27–65. Bondy, Patrick. 2016. “Counterfactuals and Epistemic Basing Relations.” Pacific Philosophical Quarterly 97 (4): 542–569. Bondy, Patrick and J. Adam Carter. 2018. “The Basing Relation and the Impossibility of the Debasing Demon.” American Philosophical Quarterly 55 (3): 203–215. Evans, Ian. 2013. “The Problem of the Basing Relation.” Synthese 190: 2943–2957. Feldman, Richard and Earl Conee. 1985. “Evidentialism.” Philosophical Studies 48 (1): 15–34. Fricker, Miranda. 2007. Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press. Goldman, Alvin. 1979. “What Is Justified Belief?” In G. Pappas (ed.), Justification and Knowledge. Dordrecht: Springer, pp. 1–23. Goldman, Alvin. 1986. Epistemology and Cognition. Cambridge, MA: Harvard University Press. Harman, Gilbert. 1970. “Knowledge, Reasons, and Causes.” Journal of Philosophy 67 (21): 841–855. Hetherington, Stephen. 2011a. How to Know: A Practicalist Conception of Knowledge. Malden, MA: Wiley-Blackwell.

Introduction

11

Hetherington, Stephen. 2011b. “Knowledge and Knowing: Ability and Manifestation.” In S. Tolksdorf (ed.), Conceptions of Knowledge. Berlin: De Gruyter, pp. 73–100. Kelly, Thomas. 2002. “The Rationality of Belief and Some Other Propositional Attitudes.” Philosophical Studies 110: 163–196. Korcz, Keith A. 2000. “The Causal-Doxastic Theory of the Basing Relation.” Canadian Journal of Philosophy 30 (4): 525–550. Lehrer, Keith. 1965. “Knowledge, Truth, and Evidence.” Analysis 25 (5): 168–175. Lehrer, Keith. 1971. “How Reasons Give Us Knowledge, or the Case of the Gypsy Lawyer.” Journal of Philosophy 68 (10): 311–313. Leite, Adam. 2004. “On Justifying and Being Justified.” Philosophical Issues 14 (1): 219–253. McCain, Kevin. 2012. “The Interventionist Account of Causation and the Basing Relation.” Philosophical Studies 159: 357–382. Plantinga, Alvin. 1993. Warrant and Proper Function. Oxford: Oxford University Press. Pritchard, Duncan. 2012. Epistemological Disjunctivism. Oxford: Oxford University Press. Setiya, Kieran. 2013. “Epistemic Agency: Some Doubts.” Philosophical Issues 23: 179–198. Swain, Marshall. 1981. Reasons and Knowledge. Ithaca, NY: Cornell University Press. Tolliver, Joseph. 1982. “Basing Beliefs on Reasons.” Grazer Philosophische Studien 15: 149–161. Toulmin, Stephen. 1976. Knowing and Acting: An Invitation to Philosophy. New York: MacMillan and London: Collier MacMillan. Turri, John. 2011. “Believing for a Reason.” Erkenntnis 74: 383–397. Winters, Barbara. 1980. “Reasonable Believing.” Dialectica 34 (1): 3–15.

Part I

The Nature of the Basing Relation

1

A Doxastic-Causal Theory of Epistemic Basing Ru Ye

1 Introduction Suppose that two students, Anna and Barry, both have strong evidence for believing that they will get an A for the epistemology course they are taking, because both have done exceptionally well in their papers and exams. However, while Anna believes that she will get an A on the basis of that evidence, Barry comes to this belief because he hopes it is true. Intuitively, Anna’s belief is epistemically better than Barry’s, even though her evidence for the belief is no better than Barry’s. What makes Anna’s belief better? Nowadays, a common answer is to distinguish doxastic justification from propositional justification. A popular way to characterize the distinction says that the former demands more than the latter: one’s belief is propositionally justified when one has good reason for the belief, and the belief is doxastically justified when it’s based on good reasons. So, Anna’s belief is better than Barry’s because it is not only propositionally justified but also doxastically justified. Given its important role in the characterization of how doxastic justification differs from propositional justification, we need to get clear on what exactly the relation of epistemic basing involves. And this is not a task faced only by those who treat propositional justification as more fundamental. Those who think doxastic justification is more fundamental also have vested interest in getting clear on basing. For they will say that what makes a set of considerations R something that provides propositional justification to believe p is that, if one were to believe p on the basis of R, the belief would thereby be doxastically justified.1 The importance of getting clear on basing goes beyond understanding doxastic justification. It’s also important for understanding reasoning or inference. For intuitively, one’s belief is a result of reasoning or inference from some premises only if one comes to hold the belief on the basis of the premises.2 I offer a new doxastic-causal theory of basing in this chapter, a theory that I call ‘Causation Caused by Believing’ (CCB). It says that one’s belief that p is based on reason R just in case R causes the belief and the causation happens because one believes that R supports p. I argue that

16

Ru Ye

this theory has two main virtues. First, it avoids the problem of deviantcausation, a problem plaguing simple causal theories of basing. Second, it implies a plausible theory of proper basing that can nicely account for the heatedly debated phenomenon of higher-order defeat. Here is the plan. In Section 2, I defend a theory that I call ‘Causation Caused by Taking’ (CCT) as a stepping stone of CCB. CCT says that one’s belief is based on R when it is caused by R and when the causation happens because one takes R to support p. In Section 3, I argue that the taking condition should be understood as a belief about evidential connection. These two steps together constitute my argument for CCB. In Section 4, I answer some objections.

2 Avoiding Deviant Causation: Causation Caused by Taking Although I am offering a doxastic-causal account of basing, I won’t say much to defend the ‘causal’ element in the account. That causation must be involved in basing is both intuitive and widely accepted.3 For basing is an explanatory relation. For me to hold a belief on the basis of some reason, that reason must explain why I hold the belief, and the most natural candidate of that explanation is a causal one. (See Turri (2011) for a more thorough defense of the causal element.) Of course, mere causation is not enough for basing. There are many ways in which one’s belief can be caused by one’s reasons, and not all of them qualify as ways of basing. So, it becomes a vexed problem for causal theorists of basing to say which kind of causation qualifies as basing and which kind doesn’t, a problem known as ‘deviant causation.’ In Section 2.1, I criticize some current solutions to the problem in a way that would motivate my causal theory CCT. In Section 2.2, I explain CCT in greater detail. 2.1 Deviant Causation Let’s consider the following two paradigm cases of deviant causation discussed in the literature of epistemic basing relation. Late and Birds I believe that I am going to be late to my class, and that causes me to run on a slippery sidewalk, lose my footing, and fall down, whereupon I find myself flat on my back looking up at the birds in the tree above me. I thereby believe that there are birds in the tree. (Pollock and Cruz 1999, p. 36.) Seeing and Hurting Suddenly seeing Sylvia, I form the belief that I see her; as a result, I become rattled and drop my cup of tea, scalding my leg. I then form the belief that my leg hurts. (Plantinga 1993, p. 69, fn. 8.)

A Doxastic-Causal Theory of Epistemic Basing

17

In the first case, my belief that I am late causes my belief that there are birds in the tree, but intuitively I don’t hold the latter belief on the basis of the former. In the second case, my belief that I see Sylvia causes my belief that my leg hurts, but intuitively the latter belief is not based on the former. It’s natural to think that the deviancy in causation has to do with how the causal path proceeds. In both cases, the causation is not direct but mediated through several stages of causation. And in both cases, the causal path involves elements that are entirely external to the subject’s cognitive system (running, losing footing, rattling, etc.). This may suggest that deviant causation happens when the causation is not direct, or when the causal path doesn’t happen entirely within one’s cognitive system (see Korcz 2000, p. 540 for the latter suggestion). Both suggestions are problematic. They are simply too strong, given that the relevant conditions also obtain in intuitively non-deviant causation. Suppose I infer ‘p’ from ‘p and q,’ which is turn inferred from ‘p and q and r.’ Then my belief p is intuitively based on my belief ‘p and q and r,’ even though the causation is not direct, and even though the causation might involve purely physical activities happening in my brain, elements that are outside of my cognitive system: my belief ‘p and q’ might first cause some neurons firing, which in turn causes my belief p. The thought that deviant causation must have to do with how the causal path proceeds also underlies the ‘causal-manifestation theory’ of basing recently proposed by Turri (2011, p. 393). According to the theory, deviant causation leading to your belief is causation that doesn’t manifest your cognitive disposition. And Turri takes a cognitive disposition to be a habit to form doxastic attitudes in certain circumstances, such as habitually taking experience at face value, habitually trusting testimony of others, or habitually reasoning in certain patterns. (In what follows, I will use ‘cognitive disposition’ and ‘cognitive habit’ interchangeably; it’s in line with Turri’s [2011, p. 391] own usage.) This theory seems able to explain why the causation is deviant in Late and Birds and in Seeing and Hurting. In both cases, the causation is entirely accidental. It doesn’t manifest my cognitive habits: I simply don’t have the habit of believing that there are birds when I believe that I am late or the habit of believing that my leg hurts when I see Sylvia. However, the causal-manifestation theory still doesn’t get to the heart of the problem. Consider the following variant of Late and Birds. Late and Running Joe believes that he is late for class. This causes him to run, which causes him to believe that he is running. And the causation chain is by no means accidental: He has a cognitive disposition to believe that he is running when he believes that he is late for class. This is because he has a habit to run when he believes that he is late for class, and he has a habit to believe that he is running when he is in fact running.

18

Ru Ye

In this case, Joe’s belief that he is running is intuitively not based on his belief that he is late for class, even though the causation between the two beliefs is not accidental but is a manifestation of his cognitive disposition. (You might doubt whether Joe’s disposition to believe that he is running when he believes that he is late for class counts as a ‘cognitive’ disposition. But according to Turri, a cognitive disposition is just a habit to form beliefs in certain patterns, and by stipulation Joe does have a habit to form beliefs in the relevant pattern—he would believe that he is running whenever he believes that he is late for class. Or perhaps Turri will object that Joe’s disposition is not a cognitive one because it’s mediated through Joe’s disposition of running, a disposition outside of Joe’s cognitive system. But this objection will rule out too much. As I have explained earlier, mediation through purely physical activities in one’s brain can be involved in normal, non-deviant causation.) So, whether an instance of causation leading to a belief is deviant doesn’t seem to depend on whether the causation is accidental or is a manifestation of cognitive dispositions.4 2.2 How to Avoid Deviant Causation What the previous diagnoses of deviant causation have in common is that they all focus on how the causation happens. They all focus on the causal paths and try to identify some features shared by those deviant causal paths. These attempts would fail because, as I see it, deviancy in causation doesn’t lie in how the causation happens, but in why it happens.5 Let’s consider some paradigm cases of non-deviant causation. I come to believe that I am mortal by inferring it from ‘All humans are mortal.’ Observing that the sun has risen every day in the past, I conclude that the sun will rise tomorrow. Receiving tons of witness testimony placing Jack at the murder scene, I come to believe that Jack is the murderer. Given clear visual experience of Tom in the library, I believe that Tom is in the library. In these cases, the causal chains leading to my belief happen for a reason: they happen because I take the putative basis to support the proposition in question. When I believe that p on the basis of R, the causation between R and my belief is ‘sanctioned’ by my taking R to support p, and the causation wouldn’t happen without the taking. For instance, the reason why my belief that all humans are mortal causes my belief that I am mortal involves, at least partly, my taking ‘all humans are mortal’ to support ‘I am mortal.’ Note that the causation in the previous examples might be indirect because it might involve mediate steps of causation, and it might involve elements external to my cognitive systems such as pure physical brain activities, and it might be accidental in a sense because it doesn’t manifest my habits. But these factors are harmless, if the causation happens

A Doxastic-Causal Theory of Epistemic Basing

19

against the backdrop of my cognitive taking about the evidential support in question. In contrast, imagine such a case: I don’t take ‘all humans are mortal’ to support ‘I am mortal’ because I just don’t see the connection, and yet the causation still happens because of some glitches in my brain. In this case, the causation would be deviant given that it doesn’t happen because of my take on the evidential connection. Also, in Late and Birds, Seeing and Hurting, and Late and Running, the causation doesn’t happen because I take the putative basis to support the believed proposition. To put the point another way, deviancy of a causal path isn’t located in how the causal path proceeds. Nothing about the causal path itself— whether it is direct, involves external elements, or is habitual—can suggest whether it’s deviant or not. When a reason R causes a belief p, the causation might be mediated by some purely physical brain activities both in deviant causation and in non-deviant causation. Nothing about those brain activities themselves marks the difference between deviant causation and non-deviant causation. What makes a stream of brain activities ‘a glitch’ instead of normal activities involved in basing is that the former doesn’t happen because of one’s take on the evidential support. So, to locate deviancy in causation, we have to examine why the causal path proceeds in the way it does. Particularly, we have to see whether it happens because of one’s taking on the evidential support in question. So, my initial conclusion is this: non-deviant causation from reasons to belief is the kind of causation that happens because of one’s take on the evidential support relation. Now, I will clarify the term ‘the causation happens because of the taking.’ There are many kinds of ‘because’ relations since there are many kinds of explanatory relations (e.g., causal, metaphysical, conceptual, etc.) Here, I intend to express a familiar causal relation: when I say that the causation from R to belief p happens because I take R to support p, what I mean is that the causation is caused by the taking. So, an initial form of the doxastic-causal theory of basing I defend is this: Causation Caused by Taking (CCT) One’s belief that p is based on reason R just in case R causes the belief and the causation is caused by one’s taking R to support p. Note that, in claiming that the causal chain from R to belief p is caused by the taking, I am not committed to claiming that the taking is always already in place before the causal chain gets started. What I am committed to is that the taking must be in place before the causal chain is completed. Suppose that I come to believe C on the basis of A in this way: at t1, I infer B from A; at an immediately later time t2, I infer C from B. The causal chain from A to C might start at t1, even though at t1 I am not taking A to support C because I haven’t recognized the connection

20

Ru Ye

between the two. But if I really come to believe C on the basis of A, then the taking must be in place at t2 (and in this case I might come to appreciate the connection between A and C because I take A to support B and I take B to support C.) You might find it odd to say that an event can be the cause or the effect of a complex event that is itself an instance of causation. But there is nothing mysterious in this claim. Second-order causation is commonly invoked in evolutionary science. For example, it makes perfect sense to say that the fact ‘human hearts cause blood-circulation’ causes human hearts to remain in the evolutionary process of human being. To further demystify the talk of ‘causation caused by taking’ involved in CCT, I will appeal to the distinction between triggering cause and structuring cause. The distinction is first noted by Dretske (2010, pp.  139– 144) in discussing a different problem. Here is a typical example. When you press the key of the mouse of a computer, the cursor on the screen moves. Your pressing the key is the triggering cause of the movement of the cursor. And whatever causes the hardware and software condition of the computer (e.g. engineers’ building the computer in a certain way) is the structuring cause of the movement of the cursor. Those causes structure the computer in such a way that when you press the mouse, the cursor will move. In general, when an event C triggers another event E, the structuring cause is the cause of the standing conditions that enable C to cause E. It causally explains why the causation between C and E can happen. Similarly, my taking R to support p can be the structuring cause of why R triggers me to believe that p. The taking structures my brain in a certain way; it sets up a backdrop that enables R to cause my belief p. But you might think that, even if the talk of structuring cause as a second-order cause makes perfect sense, why bother? Why not simply claim that, in non-deviant causation, the taking is also normal triggering cause just as the putative basis is? So, instead of endorsing CCT, consider the following simpler picture: Joint Causation One’s belief that p is based on one’s reason R just in case: one takes R to support p and this taking, joined with R, causes one’s belief that p (regardless of whether the causal contribution of R depends on the taking). This picture admits the causal role of one’s taking on the evidential connection, but it doesn’t require that the taking be a second-order cause: it doesn’t require that the taking is whatever that enables R to make its causal contribution.6 Joint Causation is indeed simpler than CCT. And presumably, it can account for cases like Late and Birds or Seeing and Hurting to the extent

A Doxastic-Causal Theory of Epistemic Basing

21

that there is no taking in these cases. However, it just doesn’t accord with our intuition to say that, in non-deviant causation involved in basing, the causal contribution of one’s putative basis R is entirely independent of the causal contribution of one’s taking R to support p. Moreover, Joint Causation is inferior to CCT for another two reasons. First, CCT can help clarify the notoriously vague distinction between bases and enablers (or between premises and background beliefs in an inference), but Joint Causation cannot. Here is an example to explain the distinction: when I infer ‘q’ from ‘p’ and ‘if p then q,’ it’s often said that the latter two beliefs are my bases (premises), whereas my acceptance of the modus ponens rule is my enabler (background belief). Although the distinction is intuitive, those authors who use it typically fail to say precisely what it amounts to. CCT can capture the distinction because it implies that the causal roles played by the bases and enablers are different: as explained earlier, bases can be understood as triggering causes, while enablers (i.e., the taking on the evidential relation) can be understood as those structuring causes that enable the triggers to make their causal contribution. In contrast, Joint Causation cannot capture this distinction because it doesn’t explain how the putative bases and the taking on the evidential relation play different causal roles in leading to my belief. If the causal contribution made by my reason R is independent from that of my taking R to support p, then we will lack resource to say that it is only R, not my taking, that constitutes the basis of my belief that p. Instead, we will have to say that both R and the taking are my basis. Second, it seems that Joint Causation cannot entirely rule out deviant causation. Even if my belief p is jointly caused by reason R and my taking R to support p, the causation can still be intuitively deviant. For example, the thought that it’s hot in the room and the thought that this supports that I am perspiring cause me to be excited, which causes me to perspire, which then causes me to believe that I am perspiring. Intuitively, I don’t hold this belief on the basis of the thought that it’s hot in the room, so the causation here must be deviant. In order to rule out this case, a proponent of Joint Causation will need to introduce a higher-order taking: I take R and ‘R supports p’ to support p. But doing so will lead to the infamous regress discussed by Lewis Carroll.7 In conclusion, in order to rule out deviant causation, both the putative reason and my taking must have made causal contributions, and the causal contribution of the reason must be causally explained by that of the taking. This completes my defense of CCT. In the next section, I argue that the taking condition involved in CCT must be understood as a belief.

3 Believing That R Supports p CCT says that one’s belief that p is based on reason R just in case R causes the belief and the causation is caused by one’s taking R to support

22

Ru Ye

p. Now the question is how to understand the taking. While it is tempting to understand the taking as a disposition to believe p given R, I argue that taking should just be a meta-belief, a belief whose content is something like ‘R supports p’. (I will say more on what exactly the crucial difference is and why it matters.) I will defend this understanding by arguing for the following two claims. (1) Understanding the taking as a meta-belief enables us to develop a very plausible theory of proper basing from CCT, but understanding the taking as certain disposition cannot enable us to do so. (2) Contrary to common impression, a doxastic theory of taking won’t over-intellectualize proper basing and it won’t lead to the Carrollian regress or vicious circularity. I will defend (1) in this section and will leave (2) to the next section. 3.1 CCT and Proper Basing To understand why CCT implies a plausible theory of proper basing, let me introduce why we need this notion on top of the notion of basing. Traditionally, the orthodox view on the relationship between propositional justification and doxastic justification says that one has the latter when one’s belief is based on whatever that provides the former. This view has recently been challenged. Turri (2010) argues that there are cases in which one’s belief p is based on a good reason and yet it’s not doxastically justified. Consider the following two cases. Tea Leaf Reader A detective is investigating whether John is the murderer. The evidence he gains (fingerprints, witnesses’ testimonies, etc.) strongly supports that John is the murderer. Then the detective comes to believe that John is the murderer on the basis of his evidence, because he takes that this is what’s supported by his evidence. However, he takes the evidence to support his belief not because he appreciates the connection between the two, but because he does some tea-leaf reading and it says that the evidence supports his belief. (Adapted from Turri 2010, p. 316) Sloppy Reasoner Susan observes that it’s raining outside. She also knows that either it’s not raining outside or the street must be wet. Based on these two beliefs, she comes to believe that the street must be wet. However, she thinks so only because she thinks that anything will follow from ‘it’s raining’ and ‘either it’s not raining or the street must bet wet.’8 In both cases, the subject bases his belief on something that provides propositional justification and yet his belief is not doxastically justified. So the orthodox view about doxastic justification faces a problem.

A Doxastic-Causal Theory of Epistemic Basing

23

An easy response is available to proponents of the orthodox view: doxastic justification requires that one’s belief be properly based, not simply based, on the propositional justifier. In fact, Turri himself (2010, p.  315) suggests that what’s going on in Tea Leaf Reader is that the subject bases his belief on his evidence in a bad way. Surprisingly, Turri doesn’t consider this easy answer; instead, he goes on to directly reject the orthodox view— perhaps he thinks that it’s just too difficult to say what kind of basing is bad. But it’s not all that difficult, if we avail ourselves of CCT as a theory of basing and if the taking condition in CCT is a belief. For if basing is causation caused by the taking and if the taking is a belief about evidential connection, then it’s natural to say that the basing is proper when the belief about the evidential connection is proper, namely, justified. So, the theory of proper basing we are looking for is as follows: Causation Caused by Justified Believing (CCJB) One’s belief that p is properly based on R just in case R causes the belief p, the causation is caused by one’s believing that R supports p, and this belief about evidential support is justified.9 CCJB explains why the basing in these two cases is bad. Both agents believe that the putative basis supports the belief in question, and yet for both of them this belief about evidential support is not justified. In Tea-Leaf Reader, the subject believes that his evidence supports that Jack is the murderer by consulting tea-leaf reading. In Sloppy Reasoner, the subject holds the belief about evidential support because he thinks that his reasons support everything. In fact, we don’t need to rely on specific examples to see the power of CCJB in accounting for proper basing. If one is justified in believing that R supports p, and if because of this belief the reason R causes one to believe p, then it’s hard to see how the way one comes to believe p could be defective. So, satisfying CCJB should be sufficient for proper basing. The only problem with CCJB, if any, is that it might be too strong for proper basing. I will address this problem in Section 4. 3.2 CCJB and Higher-Order Defeat Now, it might seem that, in order to explain the previous two cases of improper basing, what we need is just that it’s ‘unproblematic’ for one to take the putative basis R to support p, and whether the taking is a belief or a disposition is not important. For dispositional understanding of taking could also do the job: in the previous two cases, the subjects take R to support p in the sense that they are disposed to believe p given R, and what makes their basing improper is that the dispositions are epistemically problematic—for example, both dispositions are unreliable because they don’t lead to true beliefs in general.

24

Ru Ye

I admit that both the doxastic and the dispositional understanding of taking could explain why the basing in these two cases is improper. However, the doxastic understanding is better because it can cover what I see as an important type of cases of improper basing—the improper basing resulted by gaining higher-order evidence. Recent development of epistemology has witnessed a booming interest in the so-called phenomenon of higher-order defeat—defeat that is the result of higher-order evidence. This is a type of evidence not directly about the content of one’s belief, but about the epistemic status of one’s belief. Specifically, it’s evidence that one’s belief is a result of cognitive malfunction, malfunctions that would imply that the belief is unjustified. Typical examples of higher-order evidence include: evidence provided by peer disagreement; evidence that the coffee you just had was slipped some drug that undetectably harms one’s logical reasoning ability; and evidence that you (as a pilot) suffer from a condition known as ‘hypoxia,’ a condition that often undetectably harms pilots’ reasoning. (See Christensen 2007, p. 10, 2010, pp. 186-187 for more details of these cases.) It seems that when you get such evidence, the justification you have for your belief is thereby defeated. Higher-order defeat is a fascinating phenomenon partly because of the puzzles that arise out it. On the one hand, our intuition is strong that this phenomenon is real—higher-order evidence does defeat justification. On the other hand, it’s hard to explain how such evidence could defeat justification. As Christensen (2010, p. 197) notes, when one gains higherorder evidence, one’s total evidence might still support p. For example, suppose that a detective believes that Jack is the murderer when recognizing that his first order evidence supports this proposition. But when he gets evidence that he had drugged coffee that results in malfunction in assessing evidence, it seems that his total evidence still supports that Jack is the murderer—whether Jack is the murderer is related to his first order evidence like finger prints or witness’ testimony, but it has nothing to do with whether the detective is drugged. The detective’s total evidence still reliably indicates or is best explained by the truth of the proposition. This difficulty in accounting for how exactly higher-order evidence defeats justification has forced scholars into making some radical claims. Some simply deny our intuition for higher-order defeat (see, for instance, Titelbaum 2015). They think that, if your belief is originally justified, then no evidence that suggests otherwise could defeat it. Others think that we should retain the intuition for higher-order defeat but insist that there are epistemic dilemmas—we are rationally required to give up beliefs as response to higher-order evidence and we are also required to follow the evidentialist norm of believing what our total evidence supports (see Christensen 2010). However, there is an attractive account of higher-order defeat—an underappreciated one—that won’t lead to these two radical claims. It

A Doxastic-Causal Theory of Epistemic Basing

25

says that when one gains higher-order evidence, what is defeated is not propositional justification but doxastic justification, particularly, the proper basing condition involved in doxastic justification (see Smithies 2015; van Wietmarschen 2013). One’s propositional justification is not defeated because one’s total evidence still supports p. But one can no longer hold the belief because one’s believing p can no longer be properly based on one’s evidence. This explanation of the defeating mechanism of higher-order evidence is attractive. First, it is intuitively plausible. In fact, Christensen (2010) comes very close to endorsing it—he suggests that higher-order evidence defeats justification by requiring us to ‘bracket,’ or to set aside, one’s first order evidence. An intuitive way to cash out this idea of bracketing is to say that higher-order evidence defeats proper basing, that is, it makes it the case that one can no longer rely on one’s evidence. Second, this solution allows us to both respect higher-order evidence and the spirit of evidential norms. For evidentialism is most fundamentally a thesis about propositional justification. So, an evidentialist can accept that, even if one’s evidence supports p, one should not believe p if there is no way for one’s belief p to be properly based on the evidence. So, it’s an attractive account of the defeating mechanism of higherorder evidence to say that it defeats justification by making one’s belief improperly based. But in order to cash out this account, we need to understand what proper basing is. And it’s here that the defenders of the proper-basing account of higher-order defeat run into problems. Both Smithies and Wietmarschen admit that it’s difficult to say exactly what proper basing involves, although they do offer a tentative account: one’s belief is properly based just in case it’s a result of good reasoning, or when the process through which one comes to hold the belief resembles a good argument. But this is not of much help: It’s hard to see how exactly higher-order evidence renders one’s reasoning bad or renders one’s belief the result of a bad argument.10 Here we can see that my CCJB fits the bill perfectly. CCJB says that one’s belief p is properly based on one’s reason R only if one justifiably believes that R supports p. This condition is no longer met when one gets higher-order evidence, because one can no longer justifiably believe that R supports p. For example, when you get evidence that the coffee you just had was slipped some drug that undetectably harms one’s logical reasoning ability, you get evidence that your evidence-assessing abilities are damaged, so you cannot continue to hold your belief about the evidential connection. In contrast, if we understand the taking in CCT as a disposition to believe p given R, then it would be hard to see how higher-evidence defeats proper basing. While we can easily see how higher-order evidence makes it the case that one shouldn’t hold the belief that R supports p, it’s difficult to see how the evidence makes it the case that one shouldn’t have

26

Ru Ye

the disposition to believe p given R.11 When you get evidence that you are unable to assess whether R supports p, this substantially raises the chance that your meta-belief that R supports p is false, and thus it makes the meta-belief unjustified. However, to the extent the higher-order evidence doesn’t substantially raise the chance that p is false (for example, evidence that you are affected by a certain drug that damages one’s reasoning abilities substantially raises the chance that your proof of a mathematical proposition T is flawed, but it doesn’t substantially raise the chance that T itself is false), it’s unclear how the disposition to believe p given R becomes problematic when one gains higher-order evidence. The most natural way to problematize a cognitive disposition is to render it unreliable. But the disposition to believe p given R doesn’t need to be rendered unreliable by one’s higher-order evidence about p. To see this, recall the crucial point mentioned earlier that, in some cases, gaining higherorder evidence leaves the evidential support relation between R and p intact. In those cases, one’s disposition of believing p given R would be as reliable as it originally is—for instance, R might still reliably indicate p even in the presence of evidence that you are unable to appreciate the reliable indication.12 Another way to problematize a cognitive disposition is having justification to believe that this disposition is unreliable. But one’s higher-order evidence doesn’t need to give you such justification. Evidence that you are unable to assess whether R supports p doesn’t give you justification to believe that if you believe p on the basis of R then most times this belief would be false. Or you might say that one’s disposition can become problematic in another sense: even though one is allowed to have the disposition given its reliability, one might not manifest this disposition. So, higher-order evidence might render the disposition of believing p given R problematic in the sense that it doesn’t allow it to be manifested. In evaluating whether one is allowed to have a certain disposition, what matters is what the disposition does for one in the long run—that’s why reliability matters. But in evaluating whether one is allowed to manifest the disposition in a specific case, what matters is what the manifestation does in that specific case. For example, I can have the moral disposition of helping those in need. But when I know that the person in need is a terrorist I shouldn’t manifest the disposition, even if it’s still good to have it. This suggestion makes sense, but it wouldn’t give us an explanation of the defeating mechanism of higher-order evidence. For why exactly shouldn’t I manifest the disposition of believing p given R when I gain higher-order evidence? The most natural answer is that manifesting the disposition would give me an unjustified belief (namely, my belief p). But remember that we want to invoke the problematic disposition to explain how exactly my belief p becomes unjustified due to higher-order evidence. If so, we cannot say that the disposition is problematic because it leads to an unjustified belief that p.

A Doxastic-Causal Theory of Epistemic Basing

27

To sum up this section, we should understand the taking in CCT as a belief, because doing so gives us a theory of proper basing CCJB, a theory that can nicely explain typical cases of improper basing as well as the improper basing resulted by higher-order evidence. And we won’t have this benefit if we understand the taking in CCT as a disposition. Given the importance of understanding higher-order defeat, this counts as a good reason to prefer the doxastic understanding to the dispositional one. So, we have arrived at CCB as a specific version of CCT: Causation Caused by Believing (CCB) One’s belief that p is based on reason R just in case R causes the belief and the causation is caused by one’s believing that R supports p.

4 Objections and Replies Given that CCB has a doxastic element, we can expect that it will face some of the problems with simpler doxastic theories. The most prominent ones are that of over-intellectualization and infinite regress. In this section, I argue that both problems can be avoided. 4.1 The Problem of Regress As we have seen, part of my motivation for CCB comes from my defense of CCJB—a theory of proper basing. My thought is that we should accept CCB as a theory of basing because it gives us CCJB as a plausible theory of proper basing. The first objection to this strategy is that CCJB seems problematic, because it would lead to an infinite regress. For CCJB says that one’s belief that p is properly based on reason R only if one justifiably believes that R supports p. But one justifiably believes that R supports p only if this belief itself is properly based on some reason R*. According to CCJB, this requires that one justifiably believe that R* supports ‘R support p,’ which would require further proper basing and thus some further beliefs about evidential support. So, an infinite regress results. This argument is problematic. The regress will end at some point, because it’s not true that every belief about evidential support is justified only if it is properly based on some reason. Ultimately we will get to some basic beliefs about evidential support that are not based on reasons. If you ask me why I believe that the sun will rise tomorrow, I will say that the sun has risen everyday in the past. Then you ask me why I believe that ‘that the sun has risen everyday in the past supports that it will rise tomorrow,’ I will say ‘because that something has been true consistently in the past in general supports it will be true in the future.’ When you ask me why I believe that, I will run out of reasons. Another example. If you ask me why I believe that I am mortal, I will answer ‘because all humans

28

Ru Ye

are mortal, and if all humans are mortal then I am mortal.’ You ask me why I believe that, and I will answer ‘because Modus Ponens is a correct rule of inference.’ If you ask me why I hold this belief, I will run out of reasons. The point is familiar: some of our beliefs about evidential support encode our most basic epistemic rules, such as Modus Ponens and basic induction. Beliefs in these basic rules must be justified, but they are not based on further reasons.13 You might think that this answer doesn’t solve the problem of regress. For we need to give some story to explain why beliefs in the basic rules are justified even if they are not based on reasons. And one prominent story is that they are justified because of the meaning of terms in the rules (Boghossian and Williamson 2003, pp.  239-241). For example, we are justified in believing Modus Ponens because having this belief is part of what we mean by the connective ‘if.’ But this suggests that these beliefs are based on something after all—they are based on what we mean by the terms. And for the basing condition to obtain I must have further beliefs about how the meaning supports those beliefs. So, beliefs of the basic rules are not the ending point of the regress chain. To this objection, my answer is as follows. Granting the story that our beliefs in basic rules are justified because they are constitutive of meanings of terms in the rules, it’s misleading to claim that our beliefs are based on the meaning. For note that basing is a causal relation. And yet if the relation between our beliefs and the meaning in question is a constitutive one, then it’s not a causal relation. Here is a further defense of this point. Recall that the need of basing arises because having good reasons is not enough for one’s belief to be justified. Even if you have reason R that supports p, you still need to do some extra work for your belief to be justified—you need to believe p on the basis of R. But no such extra work needs to be done in the cases of constitution. Consider this example. Suppose my belief p is constitutive of my belief ‘p and q’ and suppose that the later belief is a reason of mine because it is justified. Then my belief p would ipso facto be justified. Similarly, if I mean a certain thing by a term (assuming that the meaning is not defective—see Boghossian and Williamson [2003, pp. 241-244] on why this assumption is important), and if believing p is constitutive of the meaning, then I am ipso facto justified in believing p. There is no need for a basing relation to play a role here. Of course, you can still call it a basing relation if you want, but it would be a very different one from the basing relation that my account is designed to capture.14 To sum up, our basic beliefs about evidential support can be justified even if they are not based on reasons. So, the regress chain implied by CCJB will end when we reach these basic beliefs. This point can also help answer the charge that CCJB employs implicit conceptual circularity: CCJB says that one’s belief p is properly based on R only if one’s belief ‘R supports p’ is justified, but since this latter belief

A Doxastic-Causal Theory of Epistemic Basing

29

is justified only if it’s properly based, CCJB implicitly employs the notion ‘proper basing’ in the condition it gives for proper basing, and hence there is a threat of circularity. This threat is not serious, however. For if basic beliefs of evidential support can be justified without being based on reasons, then the term ‘proper basing’ will ultimately be eliminated within a finite number of steps: one’s belief p is properly based only if one’s belief ‘R supports p’ is properly based on some reason R*, which in turn obtains only if one’s belief ‘R* supports “R supports p”’ is properly based .  .  . which in turn obtains only if one’s basic belief of evidential support is justified. Since the last condition needs not involve proper basing, the term of ‘proper basing’ will eventually disappear from a complete statement about what it takes for one’s belief p to be properly based. 4.2 The Problem of Over-Intellectualization A second worry for CCB is that it over-intellectualizes proper basing. By including a belief about evidential support, I have made proper basing too hard to come by. Children’s beliefs about their surroundings are often justified and therefore properly based on their experience, but they typically lack the cognitive capacity to form the beliefs like ‘my experience as if there is a table here supports that there is a table.’ My response to this problem is as follows. First, let’s not overestimate the difficulty involved in having a belief about evidential support. To have such a belief, one doesn’t need to master the term ‘evidential support.’ There are other options: R ‘confirms’ p, R ‘makes it likely’ that p, ‘if R then p,’ etc. All these terms can capture something close to evidential support. Furthermore, to have a belief about evidential support, one doesn’t need an explicit representation of its content. As Paul Moser (1989, pp. 141–142) has noted in defending the doxastic theory of basing, one may merely have de re awareness of one’s basis (and presumably a de re awareness can be justified in the same way as a belief can). In addition, a belief about evidential support doesn’t need to be an occurrent mental state but could be dispositional. So, having a belief about evidential support is not as hard as it seems to be. They might be available for those children who are mature enough to form justified beliefs about their surroundings. Of course, even if it’s easy to have a belief about evidential support, surely there are still some young kids who are not able to form it. But it would also be less obvious that those kids’ beliefs are really properly based on their experience. (So it’s less obvious that their beliefs enjoy doxastic justification instead of only some inferior status such as blamelessness.) Consider a two-year-old kid who believes that there is a tree in front. Suppose he doesn’t have a belief that his experience in general supports his belief. His brain doesn’t register any general connection between experience and the world. All that is going on when he forms the

30

Ru Ye

tree belief is some brute causation from his experience to his belief. Then it is hard to see how exactly this brute causal relation differs from the one we have in deviant causation. In Late and Running, the subject’s belief that he is running is caused by his belief that he is late for class. And the causal mechanism is as reliable as the one underlying the kid’s transition from his experience of tree to his belief of tree, given that the subject does typically run when he believes that he is late for class.15 So, if proper basing is absent in Late and Running, it’s also absent in the two year-old kid’s belief.

5 Conclusion Let’s take stock. In this chapter, I have argued that epistemic basing is a matter of ‘causation caused by taking’ and the taking must be a belief about evidential support. This view has two main benefits: it nicely explains what kind of causation from reasons to belief is deviant, and it gives us a plausible theory of proper basing, a theory that allows us to understand the defeating mechanism of higher-order evidence. I have also argued that this view doesn’t over-intellectualize epistemic basing and it doesn’t lead to infinite regress. I want to conclude this chapter by noting that the view of epistemic basing offered here can be easily extended to moral basing and practical basing (namely, intending to do something on the basis of moral reasons or practical reasons). So, one’s intention to perform some action is based on a moral (or practical) reason R when one’s internalization of R causes the intention and the causation happens because one believes that R supports the action.16 For example, consider my intention to pay back my friend his money on the basis of my promise to do so. I perform the action on the basis of the promise when believing (or realizing, remembering, etc.) that I have made the promise causes the intention and the causation is itself caused by my believing that a promise to do something is a moral reason to do it. So, to conclude, we have a unifying theory of basing: basing is causation caused by the belief about how the basis supports the relevant belief or intention.17

Notes 1. Comesaña (2010)’s evidential reliabilism is a theory of doxastic justification along this line. As a version of reliabilism, it treats doxastic justification as more fundamental. But it crucially relies on the notion of basing in individuating the relevant belief-forming processes: It holds that the relevant process (the availability of which would give one propositional justification according to reliabilism) is the process that corresponds to one’s forming the belief on the basis of one’s evidence. 2. Boghossian (2014, p.  8), for example, thinks that the notion of basing is crucial for understanding inference.

A Doxastic-Causal Theory of Epistemic Basing

31

3. See Plantinga (1993, p. 69). Besides, Goldman (2012, p. 85) even claims that ‘there is no hope for elucidating a suitable basing relation without giving it a causal interpretation.’ An exception is Leite (2008), who holds that having certain meta-belief is sufficient for basing. 4. I am grateful to Patrick Bondy and Matthew Lutz for the discussion here. Bondy (2016) also proposes another counterexample to Turri’s account, which involves manifesting a problematic kind of cognitive disposition such as forgetfulness. 5. Perhaps Turri would respond by saying that his manifestation account is meant to concern both the how and the why. Whether he could say this depends on what exactly the ‘manifestation’ relation is. But Turri (2011, p. 391) treats this term as primitive; his only explanation of the term is that a manifestation relation should be contrasted with a ‘merely because’ relation. Moreover, even if Turri’s manifestation account addresses the why question, it doesn’t address it in the correct way, since (as I will argue) there are cases where one’s belief manifests one’s cognitive dispositions but there is no proper basing. 6. Joint Causation is similar to a theory of basing suggested by Paul Moser (1989, p. 157), who thinks that S’s belief that P is based on a reason Q just in case the belief is causally sustained in a nondeviant manner by Q and by his associating P and Q. 7. Although CCT is better than Joint Causation for these two reasons, both seem to face this problem: They preclude cases in which I base my belief on factors that I think have nothing to do with the truth of p, but such cases seem possible. For example, I might base my belief on wishful thinking, even though I explicitly claim that my wishing that p has nothing to do with the truth of p. In reply, I don’t think such cases are possible. For if the previous case were possible, then the following kind of cases would also be possible: I base my belief that p on a factor F, even though I think that F makes it likely that p is false. But the latter kind of cases doesn’t seem possible. In believing p, I take p to be true; so, if I think F makes p likely to be false, it’s hard to see in which sense I am holding the belief p on the basis of F. 8. This case is attributable to Goldman (2012, p. 7), where he discusses a case in which the subject comes to the belief in question by employing the rule of ‘overgeneralized disjunctive syllogism.’ 9. I intend to be neutral on what specific theory of justification is involved here. As far as I can see, CCJB is compatible with both the internalist and the externalist theories of justification. 10. van Wietmarschen (2013, p. 415) says that one’s belief will be based on a bad argument because the argument would ignore higher-order evidence as a potential defeater. This explanation is unsatisfying: If the higher-order evidence is merely a ‘potential’ defeater, not a real one, then why couldn’t we ignore it? If it’s a real defeater, then the explanation is circular: our aim is exactly to explain why higher-order evidence is a real defeater. 11. I should note that, even when one’s meta-belief ‘R supports p’ is merely a dispositional one, it still dramatically differs from a disposition to believe p given R. The dispositional meta-belief might also differ from the disposition to hold this meta-belief occurrently. See Audi (1994) for the distinction between dispositional belief and disposition to believe. 12. In diagnosing what goes wrong in Sloppy Reasoner, Goldman (2012, p. 7) suggests that the basing is not proper because the subject’s belief-forming mechanism that produces the belief is not reliable. In my view, Goldman’s diagnosis doesn’t apply to those cases of improper basing resulted by gaining higher-order evidence.

32

Ru Ye

13. Or if you don’t want to call those beliefs ‘justified,’ you could say they are ‘entitled’ as Crispin Wright (2004) does. Then the last clause of CCJB becomes: the belief about evidential support is justified or entitled. 14. A consequence of saying that beliefs about basic evidential support relation are not based on reasons is that they don’t admit higher-order defeat. For I have suggested that higher-order evidence defeats proper basing and yet those beliefs are not based on reasons. This is a consequence I am happy to accept: it’s hard to imagine how my belief in modus ponens could be defeated by, say, someone’s testimony that this rule is incorrect. 15. You might think that there is a difference: the two-year-old kid’s experience of a tree is a good reason for believing that there is a tree, but ‘I am late for class’ is not a good reason for believing ‘I am running.’ Although this might be true, it shouldn’t be relevant here: whether the relation between a belief and a factor is one of basing should be independent from whether the factor is indeed a good reason for the belief. 16. I speak of ‘internalization’ of R rather than R itself because, unlike epistemic reasons, moral or practical reasons are typically understood to be external facts rather than one’s mental states. 17. For comments and discussion, I am grateful to Patrick Bondy, Adam Carter, and Matthew Lutz.

References Audi, R. (1994). Dispositional beliefs and dispositions to believe. Noûs, 28(4), 419–434. Boghossian, P. (2014). What is inference? Philosophical Studies, 169(1), 1–18. Boghossian, P., & Williamson, T. (2003). Blind reasoning. Proceedings of the Aristotelian Society, Supplementary Volumes, 77, 225–293. Bondy, P. (2016). Counterfactuals and epistemic basing relations. Pacific Philosophical Quarterly, 97(4), 542–569. Christensen, D. (2007). Does murphy’s law apply in epistemology? Self-doubt and rational ideals. Oxford Studies in Epistemology, 2, 3–31. Christensen, D. (2010). Higher-order evidence. Philosophy and Phenomenological Research, 81(1), 185–215. Comesaña, J. (2010). Evidentialist reliabilism. Noûs, 44(4), 571–600. Dretske, F. (2010). Triggering and structuring causes. In T. O’Connor and C. Sandis (Eds.), A Companion to the Philosophy of Action. Chichester, West Sussex; Malden, MA: Wiley-Blackwell. Goldman, A. I. (2012). Reliabilism and Contemporary Epistemology: Essays. New York: Oxford University Press. Korcz, K. A. (2000). The causal-doxastic theory of the basing relation. Canadian Journal of Philosophy, 30(4), 525–550. Leite, A. (2008). Believing one’s reasons are good. Synthese, 161(3), 419–441. Moser, P. (1989). Knowledge and Evidence. Cambridge, UK: Cambridge University Press. Plantinga, A. (1993). Warrant: The Current Debate. Oxford: Oxford University Press. Pollock, J. L., & Cruz, J. (1999). Contemporary Theories of Knowledge, 2nd edition. Lanham, MD: Rowman & Littlefield. Smithies, D. (2015). Ideal rationality and logical omniscience. Synthese, 192(9), 2769–2793.

A Doxastic-Causal Theory of Epistemic Basing

33

Titelbaum, M. (2015). Rationality’s fixed point (or: In defense of right reason). In J. Hawthorne and T. Gendler (Eds.), Oxford Studies in Epistemology (Vol. 5, pp. 253–294). Oxford: Oxford University Press. Turri, J. (2010). On the relationship between propositional and doxastic justification. Philosophy and Phenomenological Research, 80(2), 312–326. Turri, J. (2011). Believing for a reason. Erkenntnis, 74(3), 383–397. van Wietmarschen, H. (2013). Peer disagreement, evidence, and well-groundedness. Philosophical Review, 122(3), 395–425. Wright, C. (2004). Warrant for nothing (and foundations for free)? Aristotelian Society Supplementary Volume, 78(1), 167–212.

2

All Evidential Basing Is Phenomenal Basing Andrew Moon

1 Introduction I will defend the following: Phenomenal Basing Thesis: The evidential basing relation obtains between someone’s belief and evidence E only if the mental state associated with E has phenomenal character. This thesis highlights the important role that states with phenomenal character have to epistemology. In §2, I explain the phenomenal basing thesis and provide background. In §3—§6, I defend it.

2 Background 2.1 Understanding the Thesis What do I mean by ‘evidential basing relation’? Let’s start with some paradigm examples. Suppose Riley opens her eyes and sees something moving. On the basis of her visual evidence, she comes to believe that something is moving. Or suppose Holmes learns that the murderer wore a corsage at the party and also that only Greta wore a corsage at the party; on the basis of this evidence, he comes to believe that Greta is the murderer. In both cases, our characters believe something on the basis of their evidence; this relation between belief and evidence is the evidential basing relation. The Holmes story could have gone differently. Despite his good evidence that Greta is the murderer, suppose that Holmes had resisted believing it because of his emotional ties to her. One can have good evidence for p without believing p. Furthermore, suppose we add to the story that Holmes does come to believe that Greta is the murderer, not due to the excellent evidence at his disposal, but due to a brain lesion. Then the evidential basing relation is not instantiated. I make the following distinction:

All Evidential Basing Is Phenomenal Basing

35

Distinction: There is the evidence you have, and there is the evidence you actually use to form your belief. The latter evidence is what your belief is based on. (I will appeal to this distinction in §5.) Some philosophers are interested in some unified relation that holds not only between belief and evidence, but also between actions and nonevidential reasons.1 If the reason I hug you is because I like you, then I am also instantiating an important basing relation, one that does not relate belief and evidence. In this chapter, I am only concerned with evidential basing, which only holds between belief and evidence. It might be that what I say in this chapter will apply to other sorts of basing, but for focus, I will only address evidential basing. What do I mean by ‘evidence’? Some think that one’s evidence is one’s beliefs and experiences; others think it is propositions; others think it is only facts.2 So, in the previous example, is Holmes’ evidence the fact that only Greta wore a corsage, the proposition that only Greta wore a corsage, or Holmes’s belief (or knowledge) that only Greta wore a corsage? I’ll not take a stance on this hard question. However, I can now explain what I mean by ‘mental state associated with evidence E’. Notice that Holmes must still believe the proposition (or fact) that only Greta wore a corsage in order to have the evidence. So, even if the evidence is not the belief itself, in order to have the evidence, Holmes must still believe the relevant proposition (or fact). Similarly, Riley believed on the basis of her visual evidence. She might not need to believe her visual evidence, but she must at least be in some sort of mental state to have her visual evidence. So, either evidence is itself a mental state or one must be in a mental state to have the evidence. I intend for ‘mental state associated with evidence E’ to pick out either E itself (if E is a mental state) or the mental state required to have E (if E is a fact or proposition). And the phenomenal basing thesis affirms that evidential basing obtains only if that mental state has phenomenal character. For ease of discussion, I will assume that one’s evidence just is the mental state associated with the evidence. Those who disagree can imagine that whenever I am talking about the mental state that is the evidence, I am actually talking about the mental state required to have the evidence. The phenomenal character of a mental state is the experiential what it is like to be in that mental state. For example, there is something it is like to be in pain. Plausibly, some mental states have no phenomenal character. When I dreamlessly sleep, none of my beliefs (nor any of my mental states) have phenomenal character. Just as there is nothing it is like to be a rock, there is nothing it is like to dreamlessly sleep. Furthermore, most of one’s beliefs have no phenomenal character, even when one is awake;

36 Andrew Moon my belief that snow is white has no phenomenal character when, say, I am thinking about philosophy while eating supper. Still, I can bring my belief that snow is white to mind; then it will have phenomenal character. There is something it is like to consciously believe that snow is white.3 Not everybody agrees with this view. Many will say that it’s not the belief that has phenomenal character; rather, it is the conscious judgment that snow is white that has phenomenal character, and a conscious judgment is distinct from a belief.4 Whether the bearer of phenomenal character is a conscious judgment or a conscious belief (or both) is neither here nor there for my chapter. In our current example, there is some sort of conscious judging or conscious belief-like mental state I stand in to the proposition that snow is white when that belief “comes to mind”. It does have phenomenal character, and I will refer to that mental state with the term ‘conscious belief’. Those who think that what I am calling ‘conscious belief’ is not a real belief can imagine that I am saying ‘conscious judgment’. Some think the direct opposite of the view just discussed. They think that beliefs cannot exist unconsciously or without phenomenal character: one believes that p only if there is something it is like to believe p.5 If they are right, then I have no beliefs while dreamlessly asleep. If that is true, then nearly all the troublesome potential counterexamples I must deal with in the following sections are not actually counterexamples. So, in order to not make things too easy for me, I will assume that this view is false and that some beliefs can exist unconsciously (and without phenomenal character).6 2.2 Some Background Epistemology First, I will explain propositional justification and doxastic justification. The former is justification for believing some proposition. To understand propositional justification, it will be useful to consider one popular theory of it: EvidentialismPJ: what S is propositionally justified in believing is determined solely by S’s evidence. Note that one might have evidence for believing p (be propositionally justified in believing p) even if one does not actually believe p. Competitors of evidentialismPJ will say that evidence is not the only factor that determines whether one is propositionally justified in believing something; perhaps other pragmatic or moral conditions must be met.7 ‘Doxastic justification’ refers to a belief’s justification; the belief is formed or maintained in an epistemically proper way, perhaps in the way that is necessary (but probably not sufficient) for knowledge. Evidential

All Evidential Basing Is Phenomenal Basing

37

basing can be a link between propositional and doxastic justification: it can be that relation between a belief and evidence which converts a merely propositionally justified belief into a doxastically justified one. So, when Riley believes that something is moving on the basis of her propositional justification (or her evidence), her belief is doxastically justified. An evidentialist theory of doxastic justification will say at least this: EvidentialismDJ: S’s belief that p is doxastically justified only if S believes that p on the basis of S’s evidence. This does not give us sufficient conditions for doxastic justification.8 However, it gives us a necessary condition that nearly all who identify by the term ‘evidentialist’ would accept.9 An important question that will later arise is whether doxastic justification entails propositional justification. Many think that it does. Those who accept evidentialismDJ will naturally think that having doxastically justified belief that p entails having propositional justification (evidence) for p. Furthermore, notice that one could accept evidentialismDJ without accepting evidentialismPJ; one might think that doxastic justification has believing on the basis of evidence as a necessary condition, but then think that other factors are also necessary for propositional justification.10 So, many will think that doxastic justification entails propositional justification. However, many will also disagree. For example, Michael Bergmann (2006, 132–142) explicitly denies such a connection. On his view, roughly, what makes a belief doxastically justified is its production by properly functioning, truth-aimed, cognitive faculties. The belief needn’t be based on evidence or propositional justification. Similarly, a reliabilist might hold that whether a belief is doxastically justified is determined by its being reliably formed.11 Again, no appeal to the belief’s being based on evidence or propositional justification is needed. I’ll call such views nonevidentialist since they reject evidentialismDJ. They will be important for my later arguments. Second, I will distinguish between two species of evidentialismPJ: timeslice evidentialism and historical evidentialism. Both theories affirm evidentialismPJ. However, the time-slice evidentialist says that evidentialismPJ is true, and one’s propositional justification at a time t is determined only by one’s evidence at t. The historical evidentialist thinks that evidentialismPJ is true, and one’s propositional justification at t can also be determined by one’s evidence at times other than t. For example, evidence you had in the past can be relevant to what you are justified in believing now, even if you no longer have that evidence now.12 Third, an inferential belief is a belief that is arrived at by way of conscious inference from another belief. A noninferential belief is a

38 Andrew Moon belief that is not an inferential belief. I will use the terms ‘basic belief’ and ‘noninferential belief’ synonymously and use neither to imply anything about a belief’s justificational status.13 Notice, however, that a belief’s being basic does not exclude its being based on evidence, such as when Riley noninferentially believes on the basis of her visual evidence. She just doesn’t make a conscious inference from her evidence. With these background theories and points in mind, I will now defend my thesis.

3 Defending the Phenomenal Basing Thesis: Simple Basic Beliefs Defending the phenomenal basing thesis requires checking various beliefs and confirming that either they are based on evidence with phenomenal character or they are not actually based on evidence. In §3, I will focus on simple basic beliefs. In §4, I will focus on inferential beliefs. In §5, I will focus on more complex basic beliefs and respond to objections. In §6, I will argue for the phenomenal basing thesis. 3.1 Simple Basic Beliefs When Formed When we consider paradigm, doxastically justified, basic beliefs, they are based on evidence with phenomenal character. Consider again Riley’s perceptual belief. When she believes that something is moving, on the basis of her visual evidence, her belief is related to a mental state with phenomenal character, whether that mental state is its seeming that something is moving or some other part of her visual experience.14 Or suppose Tim believes the law of noncontradiction (LN) on the basis of its seeming true. This seeming has phenomenal character; there is something it is like for it to seem to Tim that LN is true.15 Note that I do not need to hold that the seeming is the evidence; about Tim and LN, for example, some might think that the evidence is a seeing the truth (Conee 1998) or a rational insight (BonJour 1998, 103). These mental states, though they might not be seemings, still have phenomenal character. Here are three more examples: the moral belief that suffering is bad, the memorial belief that something moved, and the introspective belief that I feel pain. In ordinary cases, these basic beliefs will be based on evidence—probably a seeming or intuition—that has phenomenal character. Some nonevidentialists might argue that these justified, basic beliefs aren’t formed on the basis of any evidence at all. Perhaps, in some of the previous examples, there is not an independent seeming that is the evidence upon which the belief is based. For example, perhaps the belief that suffering is bad is not evidentially based on its seeming that suffering is bad; rather, the seeming is just part of the conscious experience of having that belief.16

All Evidential Basing Is Phenomenal Basing

39

Fortunately, the phenomenal basing thesis only says that the beliefs that are based on evidence must be based on evidence with phenomenal character. It does not make the stronger claim that all doxastically justified beliefs are based on evidence that has phenomenal character. If the belief is not based on any evidence at all, then there is no problem for the phenomenal basing thesis.17 The phenomenal basing thesis is compatible with nonevidentialist views. I have so far discussed only relatively simple basic beliefs. By ‘simple’ basic belief, I mean a basic belief with no other background, unconscious beliefs playing a causal role in its formation. There are more complex perceptual, basic beliefs, such as the belief that the sunset is beautiful and the belief that there’s an avocado, which do have background, unconscious beliefs playing a causal role in their formation. I will discuss such beliefs in §5. 3.2 Simple Basic Beliefs When Sustained I’ve so far only considered simple basic beliefs when they are formed. I should also examine the evidential bases of beliefs after they’re formed. William Alston wrote, “[T]he role of post-origination bases in justification is a complex matter, one not at all adequately dealt with in the epistemological literature” (1989, 229).18 I believe that, over three decades later, this matter is still not given the attention it deserves, so I will try to do justice to consideration of post-origination bases in this chapter. Suppose Tim falls into dreamless sleep from all of the excitement of learning that the law of noncontradiction is true. As he sleeps at time t, presumably, his belief in LN still exists at t. I argue, 1. Either Tim’s belief in LN at t can only be based on evidence that exists at t or it can also be based on evidence that existed at a time earlier than t. 2. If the former, then it is not based on evidence. 3. If the latter, then either it is not based on evidence or it is based on the same evidence it was based on when it was initially formed. 4. Therefore, either the belief at t is not based on evidence or it is based on the same evidence it was based on when it was initially formed. Premise 1 exhausts the times at which the evidence must exist. The support for premise 2 is that, while Tim is asleep, there simply is no good candidate to count as the evidential basis for Tim’s belief, if we are only restricting the evidence to what exists at that time. In Moon (2012b), I examine a number of candidates for what the evidence might be, and I argue that they all fail. For example, at t, the earlier conscious seeming that LN no longer exists. And at t, there aren’t other unconscious beliefs or memories that serve as the evidential basis for LN. Tim might

40 Andrew Moon have a memory of having come to believe LN or its having seemed to him that LN, but Tim never comes to believe LN on the basis of those memories. The belief in LN just exists without any evidence sustaining it.19 The ones who are most likely to accept the antecedent of (2) are the time-slice evidentialist and the nonevidentialist. The nonevidentialist can think that Tim’s belief continues to be doxastically justified in virtue of being sustained by properly functioning faculties or a reliable process (or some other external factor). The time-slice evidentialist, if she accepts my claims in the previous paragraph, must conclude that Tim’s belief is not justified.20 The antecedent of (3) could be accepted by either the nonevidentialist or the historical evidentialist.21 As I just mentioned, the nonevidentialist can think that Tim’s belief’s doxastic justification is determined by external factors. The historical evidentialist will likely think the belief is based on the same evidence it was based on when it was formed. Once we are not restricting the evidence to the present moment, the earlier evidence upon which Tim’s belief was based—the seeming that LN—is a strong candidate for being the evidence upon which Tim’s belief is based.22 There really is no other good candidate. So, the consequent of (3) includes both of the likely implications of its antecedent and so is plausible. We can then conclude (4). And what is true of Tim’s simple basic belief in LN is true of other simple basic beliefs. Hence, the phenomenal basing thesis is consistent not only with simple basic beliefs when they’re formed but also when they’re sustained. Either they’re based on nothing or they are based on evidence with phenomenal character (e.g., the earlier seeming). So, the phenomenal basing thesis remains defended for simple basic beliefs.

4 Inferential Beliefs Let us return to Holmes’s inferential belief. When he forms his belief that Greta is the murderer on the basis of his evidence, he reasons as follows: “Okay, the murderer wore a corsage at the party. Only Greta wore a corsage at the party. Gasp! Greta is the murderer!” When Holmes reasons, he is inferring his belief from other conscious beliefs (or judgments) that have phenomenal character. Conscious inferences are no problem for the phenomenal basing thesis.23 Let us now consider the inferential belief when it is sustained. Suppose that Holmes, from all his excitement, falls into dreamless sleep. It is natural to think that at this time, call it ‘t’, Holmes’ belief that [G] Greta is the murderer is based on his currently unconscious beliefs that [M] the murderer wore a corsage at the party and that [O] only Greta wore a corsage at the party. If we were to ask, at t, “On what basis does Holmes believe G?” it would be intuitive and natural to cite his knowing (or believing) M and O. But at t, these beliefs have no phenomenal character. This is a potential counterexample to the phenomenal basing thesis.

All Evidential Basing Is Phenomenal Basing

41

I’ll argue that this is not a genuine case of basing because the causal condition on basing is not met. Although all three beliefs exist at t, the beliefs in M and O do not, at t, causally sustain the belief in G. Now, nobody holds that mere causation is sufficient for basing; counterexamples in the literature are numerous.24 For sufficiency, one must either specify a kind of causal relation or add a meta-belief condition.25 But as a necessary condition, the causal condition is quite plausible and widely held. So, why think it is not met in this case? First, drawing from Matthew Boyle (2011, 12), it seems that Holmes’ present belief in G can be explained by Holmes’ having believed M and O. Put another way, the earlier time-slices of Holmes’ beliefs that M and O, back when the initial inference was made, explain why he believes G now. There’s no need to appeal to the beliefs in M and O now to explain the existence of the belief in G. Analogously, one’s parents at an earlier time can explain why their adult son exists now, but the parents need not currently causally sustain the existence of the adult son. (I am imagining that he has moved out of the house and is relatively independent now, both financially and emotionally.) I agree that it’s natural to cite Holmes’ beliefs (or knowledge) that M and O as the basis for his belief in G. This can be explained by his past believing of M and O. We need not appeal to Holmes’ currently believing M and O. Second, I will argue by analogy. Suppose that while I am staying at a hotel, I leave a “Do not disturb” sign hanging outside my room and leave it there during the entire length of my stay. The maid sees it right when I put it up, and so she does not clean my room. Furthermore, she sees it once every morning until I check out and so she never cleans my room during my stay. Now, compare two claims: i) The sign is part of the cause of my room’s not getting cleaned. ii) At a particular time t, when the maid is not around, the sign is part of the cause of my room’s not getting cleaned. Claim (i) seems true. What makes it true is the fact that there were times during my stay when the maid saw the sign and did not clean up because of the sign. Claim (ii), on the other hand, seems false. At t, when the maid is not around, it seems that the sign is causally inert with respect to the room’s getting cleaned. If I were to take down the sign without the maid around, my room wouldn’t get immediately cleaned; it might be a while. Now, if the maid were to stop by and notice the absent sign, then at that moment, the room would get cleaned. But that does not imply that the sign was part of the cause of the room not getting cleaned at the other moments. Let us apply the analogy. We should distinguish these claims: iii) The beliefs in M and O are part of the cause of the belief in G. iv) At a particular time t, when Holmes is not consciously entertaining M and O, the beliefs in M and O are part of the cause of the belief in G.

42 Andrew Moon Claim (iii) seems true. What makes it true is the fact that there was a time when Holmes consciously entertained his beliefs in M and O and believed G because of those beliefs. Claim (iv), on the other hand, seems false. During those times, when Holmes’ mind is on other things, it seems that the beliefs in M and O are causally inert with respect to belief G. If Holmes were to lose the beliefs in M and O without noticing that he lost them (say, by forgetting them or by the scheming of a demon), the belief in G would not immediately disappear; it might be a while. Now, if Holmes noticed that he no longer believed M and O, then at that moment, he would drop his belief in G. But that does not imply that the beliefs in M and O were part of the cause of the belief in G at the moments before then. I spelled out the scenarios to show how they are analogous in relevant ways. I’ll now state the argument. The sign is causally inert with respect to the room’s not getting cleaned when the maid is not around to see the sign. This state of affairs is relevantly analogous to Holmes’ doxastic state. So, Holmes’ beliefs in M and O are causally inert with respect to his belief in G when Holmes is not consciously entertaining his beliefs in M and O. And, given the causal condition on basing, it follows that Holmes’ belief in G is not evidentially based on his beliefs in M and O when Holmes is not consciously entertaining those beliefs.26

5 Complex Basic Beliefs 5.1 The Challenge of Complex Basic Beliefs What makes a basic belief complex is that unconscious background beliefs play a causal role in its formation. (So, strictly speaking, the beliefs themselves aren’t more complex; it’s that they have an additional type of cause.) One might think that if a belief is evidentially based on such unconscious beliefs, then the phenomenal basing thesis is false. Here are three ways to spell out the potential counterexample. First, distinguish between pre-theoretic intuitions and theory-driven intuitions. Suppose an ardent utilitarian, Eunice, is asked whether it’s right to kill someone and harvest his organs in order to save five people. She says, “It seems right to me!” Eunice believes that it’s right to harvest the organs on the basis of the intuition that it’s right. However, plausibly, her intuition is theory-driven. Someone might conclude that her unconscious belief in utilitarianism is also part of the evidential basis of her belief.27 The second states that perceptual object identification requires background beliefs as part of one’s evidential basis. Alvin Plantinga notes, “In order to be able to see that something is an orange . . . I must also know or take it for granted that things that look like that are oranges” (1993, 100). He goes on,

All Evidential Basing Is Phenomenal Basing

43

Perhaps the thing to say is that such judgments as That tree is at least 100 feet tall are partially basic; they aren’t formed solely on the evidential basis of other beliefs, but are formed partly on the basis of present perception and partly on the basis of beliefs about what trees at least 100 feet tall look like. (1993, 100–101) Matthew McGrath (2017, 12–27) argues that to know by perception, say, that those are avocados, one must know what avocados look like. On a theory McGrath finds plausible, this entails having knowledge that avocados look way W (where W is the way avocados look).28 So, for the perceptual object identification of oranges and avocados, it seems that unconscious, background beliefs (or knowledge) serve as an evidential basis. The third draws from Thomas Senor (2005), who gives the following example: [Y]ou look at the sky and come to believe that the sunset is beautiful. This is a newly formed belief. . . . Nevertheless, its justification is no doubt dependent on other beliefs that you hold. For example, if you didn’t at least tacitly believe that you were looking west or that it is evening and not morning, the belief wouldn’t be justified (I assume that the phenomenology of sunsets and sunrises is indistinguishable) . . . many relatively simple beliefs we form about the external world typically depend for their justification on background beliefs; and background beliefs are memory beliefs. Your belief that the sunset is beautiful appears to be based not only on your sensory evidence, but also, say, on the unconscious belief that it is evening. Senor’s example has been cited by McCain (2014, 38–43), Lyons (2016, 247), and McGrath (2017, 15) as a reason to think that unconscious background beliefs are part of the evidential basis for one’s beliefs. These three cases are not meant to be exhaustive. They are just a representative sample of cases in which it seems that unconscious beliefs are part of the evidential basis for a doxastically justified belief.29 What I will say about these three cases should apply to the other examples.30 5.2. First Reply The phenomenal basing thesis requires for evidential basing that the mental state associated with one’s evidence has phenomenal character. Suppose that part of the mental state associated with one’s evidence in these cases is an unconscious belief. Still, another part of that mental state will be an intuition or a sensory experience, each of which has phenomenal

44 Andrew Moon character. Hence, the total mental state associated with one’s evidence will include something with phenomenal character. So, the phenomenal basing thesis still holds. The reply strikes me as satisfactory. However, it would be interesting to see whether a case can be made that the unconscious beliefs are not actually a part of the mental state associated with one’s evidence. I explore this in the next section. 5.3 Second Reply The intended audience of my argument will be those with internalist intuitions. According to these intuitions, those who are internally identical to me—those who have the same phenomenal states, experiences, beliefs, and so forth—are justificationally identical to me. Externalists will think that external factors—such as whether my belief was formed by a reliable process—can also make a difference to my belief’s being justified. Now, consider any belief that p that is purported to be evidentially based on some unconscious belief that q. I argue, 5. Either i) the belief that p is partly caused by a seeming that p, which is in turn partly caused by the belief that q, or ii) there is no such seeming. 6. If (i), then there is no counterexample to the phenomenal basing thesis. 7. If (ii), then there is no counterexample to the phenomenal basing thesis. 8. So, there is no counterexample to the phenomenal basing thesis. I take (5) to be plausible and predict that (6) and (7) will be the more controversial premises. The antecedent of (6) describes a natural way to understand the cases. The unconscious belief in utilitarianism makes it seem as if it is okay to harvest the organs, which causes the belief that it is okay to harvest the organs. The unconscious knowledge that it is evening makes it seem as if the sunset is beautiful, which causes the belief that the sunset is beautiful. And the unconscious knowledge that avocados look way W makes it seem as if those are avocados, which causes the belief that those are avocados.31 (Those who are skeptical that such seemings exist can move on to my discussion of premise 7.) I will give two arguments that the unconscious beliefs are not part of the evidential basis in these cases and that the seemings are doing all the work. The first one appeals to an analogy. Suppose you have a machine that performs calculations and flashes true equations on a screen. Inside the machine are all the steps in a proof that lead up to (and causally bring about) the flashing on the screen, although you won’t see the steps

All Evidential Basing Is Phenomenal Basing

45

unless you press a few buttons. “46×37 = 1702” flashes on the screen, and you form the belief that 46×37 = 1702. Even though the steps inside the machine played a causal role in flashing “46×37 = 1702” on the screen, the evidential basis for your belief was not those earlier steps. Return to Eunice. She uses the intuition (or seeming) that it’s right to harvest the organs to form her belief. The unconscious belief that utilitarianism is true might have played a causal role in producing this seeming, just as the calculations inside the machine play a causal role in producing what is flashed on the screen. However, just as your belief is not evidentially based on the calculations inside the machine, Eunice’s belief is not evidentially based on her unconscious beliefs. Recall what I earlier called the Distinction: “There is the evidence you have, and there is the evidence you actually use to form your belief.” What is Eunice actually using to form her belief that it is right to organ harvest? She is using the intuition. Just as you “read off” the correct answer from the screen and believe 46×37 = 1702, so does Eunice “read off” from her intuition (how things seem) to form her belief that it’s right to organ harvest. Similarly, what is actually used in the other two cases is the seeming that those are avocados and the seeming that the sunset is beautiful. The seemings are what you actually use, what you “read off” from, just as what’s on the screen is what you actually use or “read off” from. The second argument will appeal to internalist intuitions. Consider the unconscious belief snatching demon.32 Returning to McGrath’s avocado case, suppose that, at 5 p.m. on his 40th birthday, Matt walks up to his desk and sees and believes that those are avocados in the usual way. As mentioned before, it initially seems that Matt’s belief is evidentially based on his unconscious belief that avocados look way W (where W is the way avocados look). Now imagine another possible individual, Matt*, who is nearly identical to Matt with respect to their history and experiences. In fact, Matt* even came to know what avocados look like in the same way Matt did. The only difference is that at 4:59:59 p.m. on Matt*’s 40th birthday, a demon deletes Matt*’s unconscious belief that avocados look way W. The demon also makes it so that the phenomenal character of Matt*’s experiences are exactly like Matt’s at 5 p.m.; it seems to Matt* that those are avocados and he comes to believe it on the basis of the seeming. At 5:00:01, the demon reinserts the unconscious belief that avocados look way W, and the rest of Matt*’s life is exactly like the rest of Matt’s life. Throughout this process, Matt and Matt* are phenomenologically identical. It seems that, at 5 p.m., Matt and Matt*’s belief that those are avocados are equally doxastically justified. For those who do not share this intuition, consider that it seems that Matt* shouldn’t withhold belief or believe less confidently than Matt at 5 p.m. Put another way, it seems that it would be wrong for Matt* to stop believing those are avocados for those two seconds. But this should not

46 Andrew Moon be if Matt’s belief has a firmer evidential basis than Matt*’s since Matt’s evidential base is supposed to also include knowledge of what avocados look like. I conclude that Matt’s evidential base does not include the belief that avocados look way W after all. For comparison, suppose that the demon had altered the phenomenal character of Matt*’s experience instead of his background beliefs. If the demon altered Matt*’s experience between 4:59:59 and 5:00:01 so that it appeared to him, in that time interval, as if there is a salmon (rather than avocados), then Matt* wouldn’t be justified in believing those are avocados during those two seconds. It would seem that Matt* should stop believing that there are avocados. But if Matt* loses only the unconscious belief, remaining phenomenally the same, it seems that Matt* shouldn’t withhold belief. Now, some might be unpersuaded by this case because they are externalists. They might think, for theoretical or other reasons, that Matt*’s belief forming or sustaining process is less reliable because of the interference of the demon; hence, his belief is not justified during the two seconds. So, I do not expect externalists to be persuaded by this case.33 This argument is meant for those with strong internalist intuitions, which lead some to think that external factors are justificationally irrelevant. My case takes us to the next step and shows that that type of intuition should also lead us to think that unconscious beliefs are justificationally irrelevant. What I say about the avocados case applies to the utilitarian case and the sunset case. Even if a demon were to quietly delete and re-create the relevant unconscious beliefs, the target beliefs would seem equally justified. So, premise (6) is defended. Now to premise (7). Some philosophers are skeptical that there are such things as seemings that exist independently of beliefs and sensory experiences. They will think that there are better ways of interpreting the cases than the ways mentioned in my discussion of premise (6).34 Imagine that there is no intervening seeming in our three scenarios; there is only belief. When Eunice hears the organ-stealing scenario, she just believes, without any intervening seeming, that it is permissible to harvest the organs. When Matt has the sensory experience as of avocados, he just believes, without any intervening seeming, that those are avocados.35 When you see the sun over the horizon, you just believe that the sunset is beautiful, but there is no intervening seeming that the sunset is beautiful. My defense of (7) will employ the unconscious belief snatching demon that I used in defense of (6). Imagine Eunice hearing the story and believing, at 5 p.m., that it is permissible to harvest the organs. Imagine our Eunice’s phenomenological twin, Eunice*, whose demon does three things: deletes her background belief in utilitarianism at 4:59:59 p.m., causes the belief that it is permissible to harvest the organs at 5 p.m., and

All Evidential Basing Is Phenomenal Basing

47

reinserts the belief in utilitarianism at 5:00:01 p.m. Eunice and Eunice* are phenomenologically identical throughout this time. As with our earlier case of Matt and Matt*, it seems that they are justificationally identical. Insofar as we think that Eunice shouldn’t withhold her belief, it seems that we should think the same about Eunice*. So, we should think that Eunice and Eunice*’s evidential bases are the same. What I’ve said here applies to versions of the avocado case and the sunset case where there is a sensory experience but no independently existing seemings. In both cases, we can imagine an unconscious belief snatching demon, with the relevant beliefs remaining justificationally the same. With (6) and (7) defended, it follows that (8) is defended. The phenomenal basing thesis remains unscathed from the challenge of complex basic beliefs.

6 Arguing for the Phenomenal Basing Thesis My first argument for the phenomenal basing thesis uses a generalized version of the case of Matt and Matt*.36 Suppose that the demon deleted many of Matt*’s unconscious beliefs throughout his life, including those that epistemologists are inclined to think are part of Matt*’s evidential basis for his beliefs. The demon also ensured that Matt and Matt* were phenomenologically identical. If one agreed with my original case (where the demon only deletes Matt*’s belief for two seconds), then one should agree that Matt and Matt* are equally justified in their conscious beliefs in this more generalized case. Furthermore, it seems that the case of Matt and Matt* can generalize to other believers and their justified beliefs. This indicates that mental states without phenomenal character are irrelevant to justification. A complaint about this argument some might have is that it depends on certain internalist intuitions. My main argument for the thesis is as follows. Throughout the course of this chapter, I’ve assessed a number of paradigm cases of belief formation and sustenance. In every case, the belief was either evidentially based on nothing or based on evidence with phenomenal character, whether that evidence existed in the past or at the present moment. This is inductive evidence that every case of belief is either not based on evidence or based on evidence with phenomenal character. So, the phenomenal basing thesis is probably true.37

Notes 1. For example, see Sylvan’s (2016, 377) introductory remarks on the topic. 2. For discussions of these views, see Turri (2009), LittleJohn (2012, 92–109), McCain (2014, 9–30), and Sylvan (2016). Similar to the first view, Smithies (2019, ch. 5) argues that your evidence consists of facts about your phenomenally individuated mental states.

48 Andrew Moon 3. Notice that, in holding this view, I do not need to hold anything nearly as strong as David Pitt’s (2004, 30) view that a conscious thought’s phenomenal character is proprietary (different from any other conscious mental state), distinctive (different from consciously believing any other proposition), and individuative (constitutes its representational content). I only hold that consciously believing that snow is white has a phenomenal character, not that that phenomenal character is proprietary, distinctive, or individuative. Thanks to Adam Carter for pushing me to clarify. 4. For example, see Crane (2013, 162–167). 5. For example, see Pitt (2016). 6. The distinction between ‘conscious belief’ and ‘unconscious belief’ roughly follows the distinction epistemologists commonly make between ‘occurrent belief’ and ‘dispositional belief’ (e.g., Senor [1993, 461] and Moon [2012a, 349]). I prefer the former terms because that way, 1) there is no temptation to confuse ‘dispositional belief’ and ‘disposition to believe’ and 2) a belief’s being conscious doesn’t imply that it is occurring any more than when it is not conscious. Cf. footnote 1 of Moon (2015, 108). 7. On pragmatic encroachment, see Fantl and McGrath (2002, 2009) and Kim (2017). On moral encroachment, see Gardiner (2018). 8. For an example of a sufficient condition, see Conee and Feldman’s (2004, 93) lengthy definition of doxastic justification (or ‘well-foundedness’). 9. I say ‘nearly all’ because Smithies (2019, ch. 6, sect. 2) endorses evidentialism but also argues that in cases of a priori justified belief, the belief need not be based on one’s evidence. Thanks to Declan Smithies for helpful conversation. 10. See footnote 7. 11. For the canonical view, see Goldman (1979). For a recent defense, see Lyons (2009). 12. Here is a comment on my terminology. Conee and Feldman (2004, 101) define ‘evidentialism’, or what they call ‘ES’, as what I am calling here ‘timeslice evidentialism’. According to their definition, then, the view I call ‘historical evidentialism’ is not, strictly speaking, a type of evidentialism. However, it is natural to categorize both views as types of evidentialism, so I will go with my new terminology in this paper. 13. This non-normative use of ‘basic belief’ goes back at least to Plantinga (1981, 41). He and other epistemologists reserve the term ‘properly basic belief’ for basic beliefs that do have the relevant positive epistemic status. I’ll note that some epistemologists use ‘basic belief’ to denote only beliefs that are justified, e.g., Lyons (2009, 3). 14. The view that seemings can be an evidential basis for beliefs is discussed in Alston (1989, 106–107), Tolhurst (1998), Swinburne (2001), Huemer (2001), Tucker (2010), and the essays in Tucker (2013). 15. For argument that seemings have phenomenal character essentially, see Moon (2012b, 313–315). 16. For arguments to this effect, see Plantinga (1993, 185–193) and especially Lyons (2013, 23–25). 17. Smithies (2019, ch. 6, sect. 2) argues that a priori justified beliefs are not based on evidence. I’ve analytically defined ‘evidential basing relation’ as a relation that obtains between one’s belief and one’s evidence. So, at least as I am using the terms, Smithies’ view entails that the evidential basing relation is not instantiated by a priori justified beliefs. So, such beliefs are no problem for the phenomenal basing thesis. 18. This is from his essay, “An Internalist Externalism”.

All Evidential Basing Is Phenomenal Basing

49

19. See Moon (2012b) for more defense of these claims. For criticisms, see Madison (2014, 56), McCain (2014, 148), and McCain (2015, 371–372). For replies to these criticisms, see footnotes 21 and 22 of Moon (2018). 20. This, in my view, is a strike against time-slice evidentialism, since it seems that Tim’s belief is justified. For argument and discussion of this claim, see the references in the previous footnote. 21. A time-slice mentalist could accept the antecedent of (3) if she thought that although what grounds one’s propositional justification at t must exist at t, what grounds one’s doxastic justification at t can still exist before t. This contradicts a thesis I call ‘Unity’, which claims that what grounds one’s propositional justification at t also grounds the evidence that determines one’s doxastic justification at t. I call it ‘Unity’, since it claims a unity to what grounds propositional and doxastic justification at a time. However, since I suspect that most people will accept Unity, I assume it in the main text. Thanks to Declan Smithies for helpful discussion, and see footnotes 12 and 23 of Moon (2018) for more discussion of Unity and its relevance to evidentialism. 22. I examine in more detail this sort of theory in §2.3 of Moon (2012b), §4 of Moon (2012a), and §14.4.4 of Moon (2018). 23. Huemer (2016) argues that seemings or appearances could be part of the evidential base of an inferential belief. If so, this is consistent with the phenomenal basing thesis. 24. Here’s one from Plantinga (1993, 69): “Suddenly seeing Sylvia, I form the belief that I see her; as a result, I become rattled and drop my cup of tea, scalding my leg. I then form the belief that my leg hurts; but though the former belief is a (part) cause of the latter, it is not the case that I accept the latter on the evidential basis of the former.” 25. See a survey of these theories in Korcz (2015). 26. Some might not accept a causal condition. If so, then I will rely on the unconscious belief snatching demon argument in Section 6 to help us see that the beliefs in M and O are not part of the evidential basis of the belief in G. 27. Some might be hesitant to call Eunice’s utilitarianism as part of her evidence, either because they think that nobody could be justified in believing utilitarianism or because they think that utilitarianism is false and cannot be known (and further, only justified beliefs or knowledge can be part of one’s evidence). However, so long as it’s possible to know that some localized, moral theory is true, we could then just replace the example with a case in which someone knows a moral theory. 28. This is based on what he calls the simple intellectualist theory. 29. For other examples, see Fales (2004, 377–379), Koons (2011), and Huemer (2016, 144–145). Huemer gives cases of complex inference, where unconscious beliefs appear to serve as part of an evidential basis for an inferred belief. What I say later should apply to these cases. 30. What about subpersonal representational states? Might they be part of the evidential basis of our beliefs since they are often part of the cause of our beliefs? See Lyons (2016, 249–255) for good reasons to think that they aren’t part of the evidential basis of our beliefs. 31. I need not take a stance on how the background belief causes the seeming. It might make the person’s sensory experience have a rich content (see Siegel 2010) or the sensory experience might remain the same and the person has a felt inclination to believe, and this felt inclination is identical to the seeming. Or perhaps there is another way. 32. I draw from my argument in Moon (2012a, 350–352), which is basically my new evil demon problem for internalism. The main difference between this

50 Andrew Moon

33. 34. 35. 36. 37.

case and my earlier Melissa case is that the demon intervenes after belief formation in the Melissa case whereas in this case, the demon intervenes before belief formation. However, they should be persuaded. Clearly, Matt* shouldn’t be withholding belief for those two seconds. See Chudnoff and DiDomenico (2015). For further discussion and references, see footnote 16. This assumes that the seeming can exist independently of the sensory experience itself. This view is held by Plantinga (1993), Huemer (2001), and Tucker (2010). For dissenters, see previous footnote. For further defense of the argument I give here, see Moon (2012a, 2015). Thanks to Pat Bondy, J. Adam Carter, David DiDomenico, and Mona Simion for helpful comments on previous drafts. Thanks to Miriam Schleifer McCormick, Matthew McGrath, and Chris Tucker for helpful conversation. Thanks to the audience at Dalhousie University colloquium (5/2018). Lastly, thanks to Declan Smithies for many helpful conversations. Unfortunately, there was not space and time to incorporate some relevant issues from his forthcoming book (Smithies [2019, ch. 5]), including his objection to the sort of argument I give in this paper; that will have to wait for a future paper.

References Alston, William 1989: Epistemic Justification. Ithaca: Cornell University Press. Bergmann, Michael 2006: Justification without Awareness. New York: Oxford University Press. BonJour, Laurence 1998: In Defense of Pure Reason. Cambridge: Cambridge University Press. Boyle, Matthew 2011: ‘“Making Up Your Mind” and the Activity of Reason’. Philosopher’s Imprint, 11, pp. 1–24. Chudnoff, Elijah and David DiDomenico 2015: ‘The Epistemic Unity of Perception’. Pacific Philosophical Quarterly, 96, pp. 535–549. Conee, Earl 1998: ‘Seeing the Truth’. Philosophy and Phenomenological Research, 58, pp. 847–857. Conee, Earl and Richard Feldman 2004: Evidentialism: Essays in Epistemology. Oxford: Oxford University Press. Crane, Tim 2013: ‘Unconscious Belief and Conscious Thought’. In Phenomenal Intentionality, ed. Uriah Kriegel. New York: Oxford University Press, pp. 156–173. Fales, Evan 2004: ‘Proper Basicality’. Philosophy and Phenomenological Research, 68, pp. 373–383. Fantl, Jeremy and Matthew McGrath 2002: ‘Evidence, Pragmatics, and Justification’. The Philosophical Review, 111, pp. 67–94. Fantl, Jeremy and Matthew McGrath 2009: Knowledge in an Uncertain World. Oxford: Oxford University Press. Gardiner, Georgi 2018: ‘Evidentialism and Moral Encroachment’. In Believing in Accordance with the Evidence: New Essays on Evidentialism, ed. Kevin McCain. Cham: Springer, pp. 169–195. Goldman, Alvin 1979: ‘What Is Justified Belief?’. In Justification and Knowledge, ed. George Pappas. Dordrecht: Reidel, pp. 1–25. Huemer, Michael 2001: Skepticism and the Veil of Perception. Lanham, MD: Rowman and Littlefield Publishers, Inc.

All Evidential Basing Is Phenomenal Basing

51

Huemer, Michael 2016: ‘Inferential Appearances’. In Intellectual Assurance: Essays on Traditional Epistemic Internalism, ed. Brett Coppenger and Michael Bergmann. Oxford: Oxford University Press, pp. 144–163. Kim, Brian 2017: ‘Pragmatic Encroachment in Epistemology’. Philosophy Compass, 12, pp. 1–14. Koons, Jeremy 2011: ‘Plantinga on Properly Basic Belief in God: Lessons from the Epistemology of Perception’. Philosophical Quarterly, 61, pp. 839–850. Korcz, Keith A. 2015: ‘The Epistemic Basing Relation’. In Stanford Encyclopedia of Philosophy, https://plato.stanford.edu/entries/basing-epistemic/#DoxTheBasRel, August 13, 2015 version. Littlejohn, Clayton 2012: Justification and the Truth-Connection. Cambridge: Cambridge University Press. Lyons, Jack 2009: Perception and Basic Beliefs. New York: Oxford University Press. Lyons, Jack 2013: ‘Should Reliabilists Be Worried about Demon Worlds?’. Philosophy and Phenomenological Research, 86, pp. 1–40. Lyons, Jack 2016: ‘Unconscious Evidence’. Philosophical Issues, 26, pp. 243–262. Madison, Brent 2014: ‘Epistemic Internalism, Justification, and Memory’. Logos & Episteme, 5, pp. 33–62. McCain, Kevin 2014: Evidentialism and Epistemic Justification. New York: Routledge. McCain, Kevin 2015: ‘No Knowledge without Evidence’. Journal of Philosophical Research, 40, pp. 369–376. McGrath, Matthew 2017: ‘Knowing What Things Look Like’. The Philosophical Review, 126, pp. 1–41. Moon, Andrew 2012a: ‘Three Forms of Internalism and the New Evil Demon Problem’. Episteme, 9, pp. 345–360. Moon, Andrew 2012b: ‘Knowing without Evidence’. Mind, 121, pp. 309–331. Moon, Andrew 2015: ‘The New Evil Demon, a Frankfurt-style Counterfactual Intervener, and a Subject’s Perspective Objection: Reply to McCain’. Acta Analytica, 30, pp. 107–116. Moon, Andrew 2018: ‘Evidentialism, Time-Slice Mentalism, and Dreamless Sleep’. In Believing in Accordance with the Evidence: New Essays on Evidentialism, ed. Kevin McCain. Cham: Springer, pp. 245–259. Pitt, David 2004: ‘The Phenomenology of Cognition, or, What Is It Like to Think That P?’. Philosophy and Phenomenological Research, 69, pp. 1–36. Pitt, David 2016: ‘Conscious Belief’. Rivista Internazionale Di Filosofia E Psicologia, 7, pp. 121–126. Plantinga, Alvin 1981: ‘Is Belief in God Properly Basic?’. Nous, 15, pp. 41–52. Plantinga, Alvin 1993: Warrant and Proper Function. New York: Oxford University Press. Senor, Thomas D. 1993: ‘Internalistic Foundationalism and the Justification of Memory Belief’. Synthese, 94, pp. 453–476. Senor, Thomas D. 2005: ‘Epistemological Problems of Memory’. In Stanford Encyclopedia of Philosophy, http://plato.stanford.edu/entries/memory-episprob/, January 3, version. Siegel, Susannah 2010: The Contents of Visual Experience. New York: Oxford University Press. Smithies, Declan 2019: The Epistemic Role of Consciousness. New York: Oxford University Press.

52 Andrew Moon Swinburne, Richard 2001: Epistemic Justification. New York: Oxford University Press. Sylvan, Kurt 2016: ‘Epistemic Reasons II: Basing’. Philosophy Compass, 11, pp. 377–389. Tolhurst, William 1998: ‘Seemings’. American Philosophical Quarterly, 35, pp. 293–302. Tucker, Christopher 2010: ‘Why Open-Minded People Should Endorse Dogmatism’. Philosophical Perspectives, 24, pp. 529–545. Tucker, Christopher 2013: Seemings and Justification: New Essays on Phenomenal Conservativism and Dogmatism. Oxford: Oxford University Press. Turri, John 2009: ‘The Ontology of Reasons’. Nous, 43, pp. 490–512.

3

Dispositions and the Problem of the Basing Relation1 Hamid Vahid

It is widely believed that what distinguishes between propositional and doxastic justification is the obtaining of an epistemic relation, the basing relation, whose nature and character has long been a controversial issue in epistemology. A proposition is propositionally justified for a subject if she has adequate grounds or evidence for that proposition. If the subject then comes to believe that proposition on the basis of the evidence in question, her belief is said to be doxastically justified. The epistemic basing relation is thus a relation that obtains between a belief and the evidence or reason for which it is held. There are currently two major approaches to the problem of the basing relation, namely, the causal and doxastic theories. Doxastic theories come in different varieties and strengths. Some hold that the necessary and sufficient conditions for a belief’s being based on evidence consist in having a connecting or meta-belief to the effect that the evidence provides adequate support for the belief in question.2 Other versions of the theory postulate such meta-beliefs only in connection with the beliefs that are based on doxastic states.3 The doxastic view faces, however, a number of problems mostly originating from its requirement of the meta-belief. The idea is that subjects can have based beliefs despite lacking the relevant epistemic concepts. It is also possible for beliefs to be unconsciously formed on the basis of the relevant grounds.4 Moreover, a subject who is mistaken about the basis of her belief would fail to form the appropriate meta-belief despite her belief’s being adequately grounded. Problems such as these have pushed philosophers toward a causal account of the basing relation with almost every epistemologist believing that causation must play a role in articulating the notion of the basing relation. Alvin Goldman, for example, claims that “there is no hope of elucidating a suitable basing relation without giving it a causal interpretation”.5 The causal account of the basing relation, however, faces the challenge of the deviant causal chains which seems to show that causing the formation of a belief is not sufficient for basing.

54

Hamid Vahid

In this chapter, I will offer a dispositional account of the basing relation that is not threatened by the problem of causal deviance. In Section 1, I will look at how a similar problem in the theory of action has been treated by philosophers and then critically evaluate some of the recent attempts that have sought to adopt similar strategies in the epistemology debate. In Section 2, I shall defend a dispositional account of propositional and doxastic justification. It will be claimed that such a view has the resources to resolve the problem of causal deviance for the causal theory of the basing relation. This account is then extended to cover cases where beliefs are based on inadequate grounds.

1 The Problem of the Basing Relation: Some Responses According to the causal theory of the basing relation, for something to be a reason for a belief it must cause that belief. For example, when I come to form the belief that it will rain on the basis of my belief that the barometer has dropped, I hold the former belief because I hold the latter belief.6 My reasons are causes because they make a difference to what I believe. It is, however, well-known that the causal account is bedeviled by the notorious problem of the deviant causal chains.7 The problem is that not every case of a reason’s causing a belief constitutes the obtaining of a basing relation. Here are some illustrations. Consider the following scenario (due to Pollock)8 where a subject, Joe, believes that he is going to be late to his class which causes him to slip, falling down on his back whereupon he finds himself looking at the birds in the tree above him. Joe’s belief that he is going to be late to his class causes him to believe that there are birds in the tree, but he does not believe the latter on the basis of the former. Or, consider a case in which, due to a neural assembly malfunctioning, the subject’s belief that the oranges in the fridge are rotten causes him to believe that it will rain. It is obviously implausible to take the belief about the rotten oranges as a reason for believing that it will rain. It is also well known that a similar problem arises for the causalist theories of action. It would thus be interesting to see if we could emulate some of the responses that have been made to the problem of causal deviance in the theory of action for its counterpart in the epistemology debate. Davidson famously argued that reasons have to be causes if they are to explain our actions.9 Accordingly, he sought to explain acting for a reason in terms of a belief-desire complex (embodying that reason) that causes the action. Such an account is, however, derailed by the evidence of deviant causal chains. One can conceive of pairs of cases in which an agent has the same reason to do something but only in one of those cases she will be acting for that reason. Despite the intense interest that this problem generated among philosophers of action, none of its purported early responses has survived critical scrutiny.

Dispositions

55

One particular strategy in dealing with this problem that involves an appeal to dispositions has, however, come to increasing prominence in recent years.10 The idea is that, instead of imposing structural constraints on the causal chain leading from antecedent mental states to behavior, we can take the underlying system which produces the behavior as having causal powers and understand such powers on the model of dispositions. Indeed, Davidson himself has often described pro-attitudes such as desires as special dispositions to act or behave in a certain way.11 So, on this approach, intentional actions consist in the manifestation of the agent’s dispositions. This seems to provide a neat solution to the problem of causal deviance in the case of actions since the problem becomes just an instance of a more general claim in the dispositions debate where it is universally acknowledged that since the manifestation of a disposition involves a process running from a stimulus to a response, the response can come about without the disposition being manifested. This introduces the crucial distinction between a response obtaining merely because of a disposition and a response manifesting or realizing that disposition.12 Only when an agent’s behavior is a manifestation of such powers can we say that the behavior is the result of the agent’s intentions. Let us now see if a similar strategy can be adopted in response to the problem of causal deviance in the case of belief formation. An early attempt to imitate this strategy in the case of belief formation was made by Ralph Wedgwood whose main concern was to articulate the nature of reasoning.13 According to Wedgwood, reasoning is the process of revising one’s beliefs or intentions for a reason where the output of reasoning (belief) is said to be based on the mental antecedents (one’s other beliefs) that have caused it. This would however allow for the possibility that the causal chain will proceed deviantly as illustrated in the following example by Wedgwood. Suppose, on discovering the murder victim’s handkerchief in the professor’s laundry basket, an inspector fails to make the connection and instead thinks that the handkerchief has got its way into the basket by accident. However, the discovery forces the professor to confess to her crime prompting the inspector to believe that the professor is the murderer. In this case, the discovery of the handkerchief is not the reason on the basis of which the inspector forms his belief. In response, Wedgwood takes the dispositions route and assumes that the causal process that leads from the relevant mental antecedents to the output belief must consist purely of the manifestations of the reasoner’s dispositions. These are mental dispositions like the disposition to form a certain mental state (like forming a belief in the conclusion of an argument) in response to being in certain antecedent mental states (such as considering an argument). In reasoning the causal chain leading from the rationalizing antecedent states to one’s forming a new mental state (belief or intention) consists only of the manifestation of such dispositions. When such conditions are met, we can say that the formation of the

56

Hamid Vahid

output belief in reasoning is a direct response to the rationalizing antecedents. For example, when a subject, who is reasoning in accordance with modus ponens, ends up with the belief q, formed in response to her holding the beliefs that p and that if p then q, the belief q is based on the aforementioned mental antecedents because it results from the manifestation of her rationalizing disposition to make inferences in accordance with that rule. Similar, dispositions-oriented, assaults on the problem of causal deviance in the case of the epistemic basing relation have also appeared recently in the literature. Consider one such account (DT), due to Ian Evans,14 who takes his cue from an observation of Tom Kelly’s15 according to which if a reason is really a basis on which we form a particular belief, then we should be inclined to revise the belief upon losing the reason in question. (DT) S’s belief that p is based on m if and only if S is disposed to revise her belief that p when she loses m. (DT) seems to be able to explain our problematic cases involving causal deviance. It can, for instance, easily explain why Joe’s belief, in Pollock’s example, is not based on his belief that he will be late to his class. Suppose Joe realizes that his class starts at a much later time than he had initially thought. This discovery would not however incline him to revise his belief that there are birds in the tree even though he would be inclined to do so if he were to think that he was hallucinating seeing the birds in the tree. It seems to me however that (DT) is still vulnerable to Pollock-style counterexamples such as the following. Suppose I am in my office on the university campus. A colleague of mine, who also happens to be an expert bird watcher, informs me that he has just seen a number of white ravens in the garden on the campus. I have only a rudimentary knowledge of birds and, being brought up on such puzzles as the Ravens Paradox, believe that all ravens are black. Although I normally regard my colleague’s pronouncements about birds as reasons to believe what he says, this time I find his testimony so incredible that I hesitate to believe him. So to see it for myself, I rush toward the garden. While in there, I slip, falling down on my back whereupon I find myself looking up at what seem to be white ravens in the tree above me. This time, however, I form the belief that there are white ravens in the garden. We can further stipulate that although I have been reluctant to accept my colleague’s testimony as a basis for my belief, I am perfectly inclined to revise my belief in case he retracts what he said earlier. For example, he might tell me that he has made an error of judgment mistaking a rare species of pigeons for white ravens. Under such circumstances, I would certainly be disposed to revise my inexpertly formed belief about the existence of white ravens

Dispositions

57

in the garden. So here we have a case where I am disposed to revise my belief about white ravens upon losing a particular reason (my colleague’s testimony of having seen white ravens) while that belief is, by hypothesis, based on a different reason (seeing raven-like white birds in the tree). This shows that being disposed to revise one’s belief in response to losing a particular reason is not sufficient for the belief to be based on that reason. (DT) is inadequate as it stands.16 Another dispositions-oriented account of the basing relation has been recently championed by John Turri. Turri rejects the doxastic theories and claims that a purely causal account of the basing relation that can get around the problem of causal deviance is still possible.17 To show this, he seeks, following Wedgwood, to analyze the notion of “non-deviance” in terms of the manifestation of the cognitive traits of the subjects where a cognitive trait “is a disposition or habit to form (or sustain) a doxastic attitude in certain circumstances”.18 These dispositions include, for example, our habitually taking experience at face value, habitually trusting the testimony of others, reasoning in patterns corresponding to such formal rules as modus ponens, modus tollens, and so on. The idea then is that a subject’s reasons always manifest his or her cognitive traits. Accordingly, Turri offers the following account of the basing relation. (CMA) R is among your reasons for believing p if and only if R’s causing your belief manifests (at least some of) your cognitive traits. (CMA) seems to take care of the cases of causal deviancy discussed so far. For instance, in Pollock’s example, if Joe’s belief that he will be late for his class is to count as a reason for believing that there are birds in the tree, not only must it cause the latter belief but also its causing the latter belief must manifest Joe’s appropriate cognitive traits. But, says Turri, Joe has no disposition to trust his being late for class as indication that birds are in the tree. Likewise, although the belief of the malfunctioning subject about rotten oranges causes his belief that it will rain, it does not do so by manifesting his cognitive traits. What these and the other cases that Turri discusses in his paper have in common is that the subject lacks the required dispositions. Let us now see how his account fares in more complicated cases such as the following.19 Millicent has perfectly reliable visual powers. She is, however, falsely told by her (otherwise reliable) neurosurgeon that her visual powers are impaired. But she ignores this information and continues to form perceptual beliefs on the basis of her experiences. Suppose she sees what seems to her to be a red table (e) and forms the belief that there is a red table before her. This belief is caused by her perceptual experience e and is also the manifestation of her disposition to take perceptual experience at face value. But e cannot function as a reason for her belief because her doctor’s claim undermines e’s ability to do so which is why Millicent’s

58

Hamid Vahid

belief lacks doxastic justification. Thus, Millicent’s predicament undermines (CMA). It might, however, be objected that, despite her possession of the undermining information, the defeater does not encroach upon e’s ability to form the basis of that belief. But this invites the question of why, despite being also disposed to trust the testimony of others, Millicent should ignore and discard her doctor’s information. In other words, why of the two dispositions she possesses only one of them, namely, her disposition to take experience at face value, is manifested? After all, given the availability of their relevant stimuli, both of those dispositions can be manifested in her cognitive behavior. What is it that tips the balance in favor of the manifestation of one disposition rather than another? The obvious answer, presumably, is that, despite her doctor’s testimony to the contrary, she still regards her perceptual experience e, perhaps on the strength of the perfect track record of her visual perception, as a good reason for holding the belief in question. This would apparently save (CMA) from the counterexample. But to go along this route is to impute a meta-belief to the subject about the character of her reasons. Thus, Millicent’s reasons for believing that there is a red table before her not only consist of her perceptual experience e but also include her meta-belief about the character of e. This would lend more support to a causal-doxastic (or even a pure doxastic) theory of the basing relation than a purely causal one which (CMA) is intended to be. Moreover, there is, I think, a more serious problem with Turri’s account which concerns the exact ground on which a subject’s belief is supposed to be based. To attribute a disposition to an object is to say that the object has an intrinsic causal basis (C) in virtue of which it manifests the response (r) in any normal condition in which it undergoes a stimulus (s). It is obvious that in such cases both the stimulus s and the categorical property C play a causal role in bringing about the response r.20 That both the stimulus and the categorical properties play a causal role in bringing about r also follows from Turri’s acceptance of the thesis that causes are difference-makers requiring that causes make a difference to their effects. It follows then that, if a belief is based on what non-deviantly causes it, both the stimulus and the causal basis of the subject’s disposition (whatever it is) would compete for being the reason for which the belief is held. So (CMA) cannot uniquely identify the basis of our beliefs. So far, we have seen that none of the attempts to articulate the notion of the basing relation in terms of dispositions and their manifestations succeeds. Yet, it seems that appealing to the manifestation of a subject’s dispositions to rule out deviant causal chains is on the right track (as noted in the case of actions). So perhaps we need to look to different types of dispositions than those suggested by Turri and Evans in order to solve our problem. This is what I shall argue for in the rest of the chapter.

Dispositions

59

2 Propositional and Doxastic Justification: A Dispositional Account of the Basing Relation To believe for a reason, the belief must be based on that reason. When reasons are adequate, the obtaining of the basing relation is what distinguishes doxastic from propositional justification (I will discuss cases involving inadequate reasons later).21 This seems to suggest that we will have a better understanding of the basing relation once we are in possession of a correct account of how propositional and doxastic justification are related to one another. My proposal is to view propositional justification as an (epistemic) dispositional property that a subject can have with doxastic justification as its manifestation and then utilize it to arrive at an adequate account of the basing relation. To explain, I shall first present an intuitive case for my proposed dispositional gloss on propositional and doxastic justification and then proceed to show why the extant accounts of how these two species of justification are related to one another actually lend support to the dispositional thesis. Consider a subject, S1, who is at a certain crime scene where the butler of a large household has been murdered. S1 learns from the police that the gardener’s handkerchief has been recovered and that his fingerprints are also all over the place. While these pieces of evidence strongly support the proposition (p) that the gardener is the culprit, S1 fails to form the appropriate belief. Consider now another member of the household, S2, who has not left her room for days and is therefore not in possession of such evidence. There is a clear sense in which S1 possesses an epistemic property (propositional justification for p) that S2 lacks. This epistemic difference fully manifests itself when both S1 and S2 ‘decide’ to believe that the gardener is the murderer. However, when S1 forms this belief in response to the body of evidence at his disposal, we are inclined to give him credit for his cognitive achievement whereas when S2 forms the same belief, without any shred of evidence, she would only receive epistemic scorn. While, prior to believing, there was only a merely potential (but real) epistemic difference between S1 and S2, this difference fully manifests itself (after the beliefs are formed) in our epistemic assessment of S1’s and S2’s cognitive behavior. This scenario is not unlike a case where a lump of sugar is contrasted with a piece of paper in respect of their molecular structures. The paper clearly lacks sugar’s molecular structure which gives it the potential to dissolve in water. This potential difference (solubility) then fully manifests itself when they are both placed in water. Another way of describing the difference is to say that the piece of paper lacks the dispositional property, “solubility”, which the lump of sugar possesses. Likewise, we can view propositional justification as an (epistemic) dispositional property that a subject can possess. This dispositional property is then fully manifested when the subject forms the relevant belief. Prior to the belief

60

Hamid Vahid

formation, there is the mere epistemic potential (having propositional justification). After the belief is formed, the potential is realized in the belief’s being doxastically justified. The dispositional gloss can explain why a subject can have propositional justification for a proposition p even if he may never form the belief that p just as it makes sense to say of a lump of sugar that it is soluble even if it is never placed in water. Dispositions are generally understood to show their characteristic manifestations under certain stimuli or test conditions. In the case of epistemic dispositions, the property of having propositional justification, the stimulus conditions would consist of believing the relevant propositions with the beliefs’ doxastic justification as the manifestation of those dispositions. I shall now proceed to show that, despite contrary appearances, some of the widely held views of how propositional and doxastic justification are related to one another can be seen as actually subscribing to this dispositional picture. Let us start with the standard view (SV) which, as we have seen, takes a doxastically justified belief p to be one that is based on the reasons that constitute propositional justification for the subject. (SV) If (i) S has reason(s) R in virtue of which p is propositionally justified for S, and (ii) S believes p on the basis of R, then S’s belief that p is doxastically justified. John Turri has, however, objected that (SV) suffers from a serious defect, namely, that it fails to take into account the fact that adequate reasons can confer justification only when they are properly utilized.22 To explain, consider again the example in which our subject, S1, in the crime scene, is in possession of various pieces of evidence (E) implicating the gardener. Suppose, however, that instead of taking E to support his belief that the gardener is the culprit, S1 forms his belief on the basis of E because he thinks that the tea leaves say that E makes it highly likely that the gardener did it. Here, says Turri, (SV)’s antecedent is satisfied but S1’s belief is not doxastically justified. The point is that it is not enough that one has good reasons for forming a belief, the manner of utilizing those reasons is also important. But this concern can easily be accommodated by a slight modification of (SV) requiring the subject to form her beliefs properly.23 (SV*) If (i) S has reason(s) R in virtue of which p is propositionally justified for S, and (ii) S properly forms her belief p on the basis of R, then S’s belief that p is doxastically justified. Turri does not, however, consider this option and instead goes on to suggest an alternative account which effectively reverses the direction of the explanation of doxastic justification in terms of propositional

Dispositions

61

justification. On his account, propositional justification depends on having the ability to acquire doxastic justification. (PJ) Necessarily, for all S, p, and t, if p is propositionally justified for S at t, then p is propositionally justified for S at t because S currently possesses at least one means of coming to believe p such that, were S to believe p in one of those ways, S’s belief would thereby be doxastically justified.24 (PJ), says Turri, is no longer vulnerable to the sort of problems discussed earlier. Before proceeding to assess (PJ), it is important to see what it actually says. To start with, (PJ) states a necessary condition for having propositional justification, namely, the subject’s having an ability to acquire doxastic justification (expressed by the clause after ‘because’). It may, however, appear that it also makes a sufficiency claim, if we are willing to take the occurrence of ‘because’ in (PJ) as being intended to provide an explanation for the obtaining of facts about propositional justification. But this reading of (PJ) is rejected by Turri himself. In response to a counterexample to (PJ), he notes that (PJ) does not claim that “[if] S currently possesses at least one means of coming to believe p such that were S to believe p in one of those ways, S’s belief would thereby be doxastically justified, then p is propositionally justified for S”.25 But if (PJ) is not intended to make a sufficiency claim, it can be simplified, without any loss, in the form of the following schema highlighting its role in stating a necessary condition for having propositional justification (with ‘N’ denoting “necessary condition”).26 (NPJ) S has propositional justification for p at time t only if (and because) S currently possesses at least one means of coming to believe p such that were S to form his belief p properly on the basis of R, S’s belief would be doxastically justified. Turri also considers a possible challenge to (PJ) which concerns cases where, despite having excellent reasons and, thus, having propositional justification for p, the subject is, on account of brainwashing, drugs and so on, incapable of believing that p and, ipso facto, incapable of forming a doxastically justified belief. In response, he qualifies (PJ) by idealizing our judgments about propositional and doxastic justification which involves abstracting away from the abilities and powers of the relevant subjects. It has been argued that, even with such modification, (PJ) is vulnerable to counterexamples.27 There is, however, a simpler way of accommodating the aforementioned objection. The challenge poses a problem because the subject is assumed to be incapable of belief. So, to get around

62

Hamid Vahid

this problem, all we need to do is to eradicate the reference to ‘belief’ in (PJ), or (NPJ), and instead highlight the subject’s reasons for his belief. (NPJ*) S has propositional justification for p at time t only if (and because) S has adequate reasons R such that were S to form his belief p properly on the basis of R, S’s belief would be doxastically justified. So we now have two competing accounts of how propositional and doxastic justification are related to each other: (SV*) and (NPJ*). To adjudicate between these views, we need to recall the dispositional gloss that I earlier suggested we put on these two notions of justification. I claim that (SV*) and (NPJ*) are best seen as reflecting this dispositional character. To show this, I need to say a few words about the debate over the question of how dispositional statements should be analyzed. Despite the prevalent use of dispositional expressions in science, dispositions have always seemed to some philosophers, particularly with an empiricist bent of mind, to be mysterious entities, not lending themselves to direct observation. An early attempt to render them respectable was made by Carnap who sought to construe disposition ascriptions in terms of (material) conditionals involving observational terms.28 (D) x has disposition to manifest M in response to s iff (x is put in stimulus condition s → x manifests M) Carnap, however, noticed that, as it is stands, (D) is inadequate because it ascribes dispositions to objects which they clearly lack. Consider, for example, a piece of paper that is never placed in water and then is burned. It certainly lacks the disposition to dissolve in water despite the right-hand side of ‘iff’ in (D) being trivially satisfied. Two things were, however, highlighted by the failure of the Carnapian definitions: (1) the relevance of the causal bases of dispositions to their analyses as well as (2) the need for a stronger conditional than material conditional. Kripke’s work on modal logic together with the subsequent development of the semantics for counterfactuals, however, paved the way for, what is known as, the simple counterfactual analysis of dispositional statements. (SCA) x has the disposition to yield manifestation M iff (were x put in stimulus condition s, x would manifest M in response to s). It turned out, however, that (SCA) is vulnerable to counterexamples involving the so-called ‘finkish’ dispositions.29 The thought was that dispositions come and go and that they take time to manifest themselves. So it is quite possible that when x undergoes a stimulus s, s would tamper

Dispositions

63

with x’s intrinsic causal basis resulting in the disappearance of the disposition, thus, preventing its manifestation. In response, David Lewis suggested that an adequate account of dispositions should require that their causal bases remain intact for sufficient time after they undergo the stimulus.30 This led to the following revision of (SCA). (RCA) x is disposed at t to manifest M in response to s iff x has some intrinsic property B at t and for some time t’ after t, such that were x to undergo s at t and retain B until t’, s and B would jointly cause x to manifest M. There is a lot more to be said about (RCA) and its viability as an analysis of dispositional predicates. But we have said enough to see the connections between the dispositions debate and the controversy over the relationship between propositional and doxastic justification. To explain, consider Carnap’s conditional analysis (D) of dispositional statements. Since (D) is a biconditional, it entails the following necessary condition on an object having a dispositional property. (ND) If x has a disposition to manifest M (in response to s), then (x is put in stimulus condition s → x manifests M). Now, let us recall (SV) which we can also write thus: (SV) If S has propositional justification for p (on the basis of reasons R), then (S believes that p on the basis of R → S’s belief that p is doxastically justified). It is easy to see that (ND) and (SV) (or (SV*)) share the same a structure. If, as I have suggested, we take propositional justification to be an (epistemic) dispositional property that a subject can have, then (SV) can best be seen as offering a Carnapian analysis of the dispositional statement ‘S has propositional justification for p’. Consider now the transition from (D) to the simple counterfactual analysis (SCA) in the dispositions debate whose primary motivation was to highlight the relevance of the causal bases (intrinsic properties) of dispositions (and their associated modal character) to their analyses. This resulted in the simple counterfactual analysis of dispositional statements (SCA). Given our claim that the metaphysical debate closely parallels the debate in epistemology, we might wonder what an analogue of (SCA) would be like in the corresponding epistemology debate. We may call it (JCA) (‘J’ denotes “justification”). (JCA) S has propositional justification for p at t only if (were S to believe p at t, S’s belief that p would be doxastically justified).

64

Hamid Vahid

We saw, however, that the case of finkish dispositions caused serious problems for (SCA) that culminated in Lewis’s (RCA). Is (JCA) also vulnerable to analogous threats? Finkish dispositions, we may recall, are those where the test itself takes away the dispositions. But such cases are possible because physical dispositions take time to manifest themselves, not so in the case of epistemic dispositions. Suppose I have propositional justification to believe a proposition p at t. When I form the belief p at t, my belief will be doxastically justified at t and not at any later time. This feature of epistemic dispositions renders (JCA) immune to finkishlike cases. But finkish dispositions were not the only reasons for Lewis’s dissatisfaction with (SCA). He also claimed that the intrinsic properties (causal bases) of dispositions are relevant to their analysis which led him to (RCA). I have already stressed that, in the case of epistemic dispositions, reasons play the role of those intrinsic properties or causal bases. Just as the dispositional properties of an object obtain in virtue of its intrinsic properties, a subject’s propositional justification for a proposition p also obtains in virtue of (or because of) the reasons that the subject possesses. So if we follow Lewis’s insight, we can either enrich (JCA) with a clause referring to the subject’s reasons, or delete the time gap in Lewis’s (RCA) in order to construct an analogous statement in the epistemology debate. Either way, we arrive at the following. (JCA*) S has propositional justification for p at t only if S has adequate reasons R at t such that were S to believe p at t (properly) on the basis of R, S’s belief that p would be doxastically justified. But (JCA*) is precisely the kind of view that our improved version of Turri’s (PJ), namely, (NPJ*) expresses. This confirms my claim that the extant theories of propositional and doxastic justification can be glossed without loss within a disposition framework thereby lending support to the dispositional account of propositional and doxastic justification defended here. However, while both (SV) (or (SV*)) and (NPJ*) can be understood as supporting a reductive account of (epistemic) dispositions, I take dispositions in general (and epistemic dispositions in particular) to be actual properties that exist independently of their manifestations. They can therefore still be there even if they are never manifested. This general view about the nature of dispositions is actually in conformity with the subsequent developments in the dispositions debate. Certain problem cases, known as ‘masks’ and ‘antidotes’, showed that, pace Lewis’s analysis (RCA), the presence of the dispositions plus the stimuli may still fail to bring about the expected manifestations. And it is fair to say that the subsequent patching ups of (RCA) have been unsuccessful. Although, for

Dispositions

65

lack of space, I cannot go into the details here, I think (NPJ*) is also vulnerable to similar problems. My own view is that a successful resolution of the epistemology debate should acknowledge that ‘S has propositional justification for p’ is itself an irreducible dispositional concept. I do not however need to argue for that here.31 Having shown that the standard view and Turri’s account of how propositional and doxastic justification are related to one another closely parallel various analyses of how dispositions are related to their manifestations, all I need for the purposes of this chapter is the thesis that we should view propositional justification as an (epistemic) dispositional property with doxastic justification as its manifestation.32 Let us now see how such a view might bear on the problem of causal deviance for the causal theory of the basing relation. 2.1 Dispositions and the Problem of Deviant Causal Chains Consider a subject S who has propositional justification for a proposition p in virtue of having adequate reasons (R). We have been arguing that S has thereby an epistemic dispositional property that manifests itself as doxastic justification when S comes to believe that p. Moreover, this dispositional property is a genuine property such that S can have it even if he never manifests it because of failing to believe that p. However, given that our beliefs are typically prompted by our evidence (experiences, etc.), when S’s belief that p is thus formed in response to R, it will be based on R in virtue of manifesting S’s disposition. So it is the obtaining of the basing relation that marks the transition from S’s having an epistemic dispositional property to his manifesting it when forming the pertinent belief. Accordingly, on our account, the notion of the basing relation discharges the same function that it does on other accounts. What is distinctive of our account, however, is that the kind of disposition that is taken to be relevant to the characterization of the basing relation is provided by the evidence itself rather than, as in other views, by the habits that a subject may or may not have. Thus, as we shall see shortly, our account will be invulnerable to the kind of objections that were earlier shown to undermine those views. In any case, the idea is that, with the obtaining of the basing relation as marking the transition from having an epistemic dispositional property (propositional justification) to its manifestation (doxastic justification), our dispositional account of propositional and doxastic justification would provide us with a better understanding of what that relation consists in. Before I proceed to justify this claim, a few remarks are however in order. To begin with, we have been assuming that for any epistemic dispositional property possessed by a subject S, there is some property R (having certain reasons) in virtue of which S has that disposition. R functions

66

Hamid Vahid

like a categorical basis (b) for a physical disposition (d). There are various ways of cashing out this ‘basis’ relation. We may say that b explains why object x has the disposition d, or that b is the truth-maker of an ascription of d to x. In general, in the debate over physical dispositions, the categorical bases are either conceived as causing the relevant dispositions that, in turn, cause the object to bring about a particular manifestation in suitable conditions, or they are taken to be directly causing those manifestations. There is, however, a more plausible way of conceiving of how epistemic dispositions are related to their bases. To explain, recall that we are assuming that the reasons that a subject possesses are adequate (which is why he is said to have propositional justification).33 Given that epistemic properties are taken to supervene on non-epistemic properties, it is widely held that the property of being justified is supervenient on such properties as being adequately grounded, being caused by experience and so on.34 Accordingly, we may take epistemic dispositions to supervene on their evidential bases in the sense that there will be no change in an epistemic disposition possessed by a subject unless there is a change in its evidential basis and that two subjects who are alike with respect to their evidential grounds are also alike in their epistemic dispositional properties (having propositional justification).35,36 Second, we know that some physical dispositions can have different causal bases in different objects. Likewise for epistemic dispositions. A subject can have propositional justification for a proposition p in virtue of various kinds of reasons like testimony, perceptual experience, and so on. Since the identity of such dispositions turns out to be important for our account of the basing relation, we can individuate those dispositions by their evidential bases or reasons (R). We may thus call an epistemic disposition ‘R-related’ just in case the disposition obtains in virtue of R. The same disposition might be R’-related if R’ can also give rise to that disposition. We can now present our account of the basing relation along the following lines. (Basing) A subject (S)’s belief p is based on reason R iff S has an R-related epistemic disposition (having propositional justification for p) such that when S believes that p, the R-related disposition manifests itself as the belief’s being doxastically justified. (Basing) has the resources to explain how deviant causal chains might intrude in the course of the subject’s forming his beliefs. When a subject strives to form a belief on the basis of his reason R, the outcome is produced deviantly if and only if it fails to be the manifestation of the subject’s R-related disposition (propositional justification). Moreover, (Basing) seems to be immune to the sort of objections that were earlier leveled against the other accounts of the basing relation. For

Dispositions

67

example, Turri’s dispositional analysis was criticized on the ground that it failed to uniquely identify the basis of one’s beliefs. But, on our account, there will be no competition for what it is that forms the basis of one’s beliefs since it is the subject’s body of evidence that is responsible for both causing the beliefs as well as grounding the dispositions which are manifested when the beliefs are formed. (Basing) can also handle the sort of cases that were earlier shown to cause trouble for Evans’s brand of dispositional account. To see more clearly how (Basis) discharges these functions, let us consider such test cases one by one. Let us start with Pollock’s example where Joe’s belief that he is late for class causes him to believe that there are birds in the tree. Here, the relevant epistemic dispositional property is Joe’s having propositional justification for the proposition that there are birds in the tree. But the only evidential basis that can give rise to this disposition is Joe’s sensory experience of seeing the birds rather than his belief that he is late for class. The latter belief is a completely inadequate ground to provide justification for the proposition that there are birds in the tree. Consequently, there is only an experience of seeing birds-related epistemic disposition that obtains in virtue of Joe’s sensory experience. Since it is this disposition that is manifested when Joe comes to believe that there are birds in the tree, by (Basing), Joe’s belief is based on his experience of seeing the birds rather than on his belief that he is late for class. As for the rotten-oranges example, no appropriate epistemic disposition is being manifested in this case. Seeing the rotten oranges in the fridge does not provide justification for the proposition that it will rain. So no appropriate epistemic disposition obtains. Let us now consider some of the cases that caused trouble for other dispositional accounts of the basing relation. Consider Wedgwood’s case where an inspector forms the belief that the professor is the murderer on the ground of his confession rather than the inspector’s evidence of finding the professor’s handkerchief in the victim’s laundry. This is easily resolved. Both the handkerchief and the professor’s confession give rise to an appropriate epistemic disposition, namely, the inspector’s having propositional justification for believing that the professor is the murderer. So there is a handkerchief-related epistemic disposition as well as a confession-related disposition. But since it is only the latter disposition, given rise to by the professor’s confession, that is manifested, the inspector’s belief that the professor is the murderer is based on this evidence rather than the handkerchief. What about the white ravens example? Here, too, we have two evidential bases, namely, my colleague’s testimony and my own experience of seeing the white ravens, which give rise to the pertinent epistemic disposition, that is, having propositional justification for believing that there are white ravens in the garden. However, since it is only the seeing white ravens-related disposition that is manifested, my belief about the ravens in the garden is based on such an experience rather than the testimony of my colleague.37

68

Hamid Vahid

Finally, let me say a few words about extending such an account to cover cases where one’s beliefs are based on inadequate reasons that fail to ground appropriate epistemic dispositions. My suggestion is to see such cases as being parasitic on cases where one’s reasons are adequate. To see this, consider a similar strategy, adopted by Jim Pryor, in connection with a totally different problem. The problem concerns the contrast between rational commitment and what one has justification to believe. Rational commitment is a hypothetical relation between one’s beliefs. Suppose you believe that Fred can fly. Regardless of whether or not this belief is justified, it rationally commits you to the belief that someone can fly (even if you happen to have strong reasons against the latter belief). In other words, if you believe that Fred can fly but refrain from believing that someone can fly, that would constitute a rational failing on your part. The question is how one can determine what the rational commitments of a belief are. Pryor has suggested that, since we have a good grasp of what the consequences of having a justified belief are, we should understand the rational commitment of a belief p in terms of “the epistemic effects of [one’s] having (decisive) justification for that belief. . . . If one of the effects is that the subject has decisive justification to believe q, then his belief in p counts as rationally committing him to the belief in q—regardless of whether he really does have any justification to believe p.”38 I propose to do the same for the problem of the basing relation. I initially presented an account of the basing relation in regard to the reasons that are adequate and justification-conferring. We can now take the cases involving inadequate reasons as being parasitic on those cases. Consider the example of the flying-Fred again. Suppose my belief that (p) ‘Fred can fly’ is justified. On our account, this justified belief gives rise to an epistemic disposition to believe that (q) ‘someone can fly’ with doxastic justification. So there is a Fred can fly-related epistemic disposition that obtains in virtue of my justified belief that Fred can fly. Since it is this disposition that is manifested when I come to believe that someone can fly, the latter belief is based on my belief that Fred can fly. Now, just as the subtraction of ‘justification’ from the ‘justified belief p’ leaves us only with the rational commitment of the belief p, subtracting ‘justification’ from the ‘justified belief p’—as an adequate basis for believing q—leaves us only with the basing relation between the two beliefs. In both cases the required relations are identified by reference to the circumstances where the conditions are optimal. To conclude, I discussed various ways in which the notion of disposition can be employed in an analysis of the basing relation in order to make it immune to the problem of causal deviance. No such account, however, survived critical scrutiny. Instead, it was argued that a dispositional analysis of how propositional and doxastic justification are related to one another can provide the basis of an account of the basing relation that is no longer vulnerable to the problem of deviant causal chains.

Dispositions

69

Notes 1. I am very grateful to Patrick Bondy, Adam Carter, and Jesper Kallestrup for helpful comments and suggestions on earlier versions of this paper. 2. See, for example, Tolliver (1982). 3. Audi (1986). 4. See, for example, Alston (2005). 5. Goldman (2012, p. 84). 6. It is important to note how reasons interact with beliefs (and actions). Objective normative reasons are widely thought to be constituted by facts. It is, however, controversial how such reasons bear on the rationality of actions and beliefs especially in view of the thesis that rationality requires that one correctly respond to one’s reasons. For we can easily conceive of cases in which an agent’s actions and beliefs are rational despite the agent’s being non-culpably ignorant of the relevant normative reasons. This seems to suggest that reasons that justify actions and beliefs are those that are possessed by the agent, namely, those that fall within her ken. At a minimum, possession of reasons requires that they be cognitively accessible to the agent. One can understand this requirement either in normative terms as involving justified belief or perhaps knowledge or in non-normative terms involving such states as “seemings”, presentational states or merely doxastic states such as beliefs (see Schroeder 2011). 7. See, for example, Korcz (2000) and Vahid (2009). 8. Pollock (1986). 9. Davidson (1963). 10. See, for example, Stoecker (2003) and Stout (2005). See also Hyman (2014). 11. See, for example, Davidson (2004, p. 108). 12. To see the point of the distinction, consider the following example. To say that sugar has a disposition to dissolve in water is to say that it manifests this disposition when it is placed in water. Suppose now we insert a sensor into a lump of sugar such that whenever it is placed in water a signal is sent by the sensor with the effect that the sugar is immediately dissolved by a strong bombardment of X-rays before water has a chance to dissolve it. Here, although sugar’s disposition plays a causal role in bringing out the response (the sugar solution), the response is produced deviantly and so fails to be a manifestation of the sugar’s disposition. What is manifested in this process is not sugar’s solubility but a different disposition. 13. Wedgwood (2006). 14. Evans (2013). 15. Kelly (2002). 16. It might be suggested that a much simpler counterexample to (DT) is possible. Consider again the previous example of the rotten oranges where, due to the neural misfiring, S’s belief that oranges in the fridge are rotten causes him to believe that it will rain. All we need to do is to add that the misfiring also makes it the case that if S loses the orange-belief, then he will revise his belief that it is raining. Although I am sympathetic to this line of reasoning, a diehard (DT)-theorist might be inclined to bite the bullet and claim that, under these circumstances, the subject’s belief that it will rain is based on his belief that the oranges are rotten. He might say that any implausibility that might be associated with this claim has to do with, not the basing claim but, the fact that the basis of the rain-belief is wildly inadequate. My own counterexample has the advantage that while the belief about the white ravens is clearly based on something other than my colleague’s testimony, namely, my own experience, the belief is revised because of the testimony.

70 17. 18. 19. 20. 21.

22. 23. 24. 25. 26.

27. 28. 29. 30. 31. 32.

33.

34. 35.

Hamid Vahid Turri (2011). Ibid., 391. Adapted from Goldman (1986). In his account of dispositions (1997, to be discussed shortly), David Lewis requires that the stimulus s and the causal basis C jointly constitute, what he calls, an ‘x-complete’ cause of the response r. There are two ways in which a proposition that is propositionally justified for S can fail to be doxastically justified: it can either be held on the basis of an inadequate ground or it can be held on the basis of an adequate ground but in an epistemically inappropriate manner. I shall give examples of the latter when discussing the relationship between propositional and doxastic justification in this section. Turri (2010). See also Silva (2016). Turri (2010). Ibid., 322. Turri’s claim is of the form ‘if p then (p because q)’, so the primary necessary condition for p it provides is ‘p because q’, although since the latter entails q it does secondarily provide q as a weaker necessary condition. But since ‘p because q’ also entails p, ‘p because q’ is both a necessary and a sufficient condition for p (given that it is necessary). But since Turri holds that in such cases we can have q without p, he denies that q by itself is sufficient for p, so he is not using ‘because’ to mean something strong enough for strict implication. Silva (2016). Carnap (1936). Martin (1994). Counterexamples to the more sophisticated versions of the conditional analysis of dispositions involve the so-called masks and antidotes. For discussion see Vahid (2016). Lewis (1997). For this and many other related issues see Vahid (2016). In this paper, I assume a more or less general account of epistemic justification which is intended to be neutral in regard to the controversies involving that concept. The recent knowledge-first approach to epistemology is a case in point. On this approach, justification is subordinate to knowledge. Sometimes this is understood as claiming that it is knowledge that justifies belief, sometimes that justification is a kind of ‘would-be’ knowledge or simply that justified belief is nothing other than knowledge. The claim that forms the basis of my account of the basing relation is that we should think of propositional justification as an epistemic dispositional property with doxastic justification as its manifestation. If it turns out that justification is a kind of ‘would-be’ knowledge, then knowledge can be regarded as the manifestation of the dispositional property in question. The question of what it is to possess a reason is a large one that cannot be considered here. However, as pointed out before, reason-possession requires, at least, that one’s reason (r) for believing a particular proposition p be cognitively accessible to the subject. But this is not sufficient. The subject must also ‘treat’ r as a reason for believing that p. Only then can we say that the subject is guided by her reason or possesses that reason. The notion of ‘treating’ must, however, be construed in such a way as to avoid over-intellectualizing the whole process of belief formation (see Vahid (2018) for an account of reason-possession). See, for example, Kim (1988) and Goldman (1979). My concern here is only with the question of how dispositions are related to their categorical bases or grounds i.e., whether the grounds are the

Dispositions

71

truth-makers of the dispositions’ ascriptions, or that they have merely an explanatory function and so on. It seems that in the case of epistemic dispositional properties (having propositional justification) the supervenience relation can be aptly used to describe the relationship in question especially in view of the widely held thesis of epistemic supervenience. As mentioned earlier, I have, in this paper, tried to steer clear of the controversies regarding the concepts of epistemic justification and evidence. Evidence, and more generally the notion of an epistemic reason, are highly controversial issues. There are, for example, questions about the nature of epistemic reasons, whether they are of a normative or motivating variety, what it is to possess such reasons, etc., as there are questions about their ontological status (facts, mental states or propositions). On some views, a proper account of evidence should include elements from reliabilism (Goldman 2011) or virtue epistemology. I think it is best to discuss the question of the basing relation by trying to avoid these contentious issues as much as one can. 36. It may be worth reminding ourselves of the dialectical situation. We began with the truism that propositional justification obtains in virtue of the reasons at one’s disposal viz., the reasons one possesses. Since I have argued that propositional justification is best seen as an epistemic dispositional property with doxastic justification as its manifestation, those possessed reasons are what ground such dispositions functioning as their categorical bases. It was further suggested that the epistemic dispositional property (having propositional justification) supervenes on possessed reasons. Of course, sometimes those reasons involve further epistemic properties (such as further justified beliefs). But those justified beliefs must themselves supervene on further reasons. And, if the regress is to come to an end, the epistemic properties must eventually supervene on non-epistemic properties (such as experience, etc.). 37. Another controversial case is Lehrer’s Superstitious Lawyer (Lehrer 1971). The case involves a lawyer defending a man who is being accused of eight brutal murders. There is strong evidence that the man committed the first seven murders. The question is whether he is also guilty of the eighth murder which resembles the other murders. The lawyer happens to have absolute faith in the cards which tell him that his client is innocent of the eighth murder and he believes this. Upon reexamining the evidence, however, he finds a conclusive way of establishing his client’s innocence. But this fails to increase his conviction as he was already convinced by the cards. This is supposed to be a counterexample to the causal account of the basing relation. The idea is that while the lawyer is justified in believing that his client is innocent, his justification is not based on the evidence at his disposal. Many philosophers have found Lehrer’s example unconvincing. On my account, the lawyer’s possession of evidence provides him with an epistemic disposition (having propositional justification) to believe that his client is innocent. But since he chooses the testimony of the cards (whatever it is) as the basis of his belief, it is not that disposition which is manifested when he forms the belief that his client is innocent. So his belief fails to be based on his adequate evidence and it is therefore unjustified. 38. Pryor (2004, p. 364).

References Alston, W.: 2005, Beyond “Justification”: Dimensions of Epistemic Evaluation, Ithaca, NY: Cornell University Press. Audi, R.: 1986, ‘Belief, Reason and Inference’, repr. in Audi, 1993, The Structure of Justification, Cambridge: Cambridge University Press: 233–274.

72

Hamid Vahid

Carnap, R.: 1936, ‘Testability and Meaning’ (part I), Philosophy of Science 3: 420–471. Davidson, D.: 1963, ‘Actions, Reasons and Causes’, Journal of Philosophy 60/23: 685–700. Davidson, D.: 2004, Problems of Rationality, Oxford: Oxford University Press. Evans, I.: 2013, ‘The Problem of the Basing Relation’, Synthese 190: 2943–2957. Goldman, A.: 1979, ‘What Is Justified Belief’, in Epistemic Liaisons (1992), Cambridge: MIT Press. Goldman, A.: 1986, Epistemology and Cognition, Cambridge: Harvard University Press. Goldman, A.: 2011, ‘Towards a Synthesis of Reliabilism and Evidentialism? Or, Evidentialism’s Troubles, Reliabilism’s Rescue Package’, in T. Dougherty (ed.), Evidentialism and Discontents, Oxford: Oxford University Press: 254–281. Goldman, A.: 2012, Reliabilism and Contemporary Epistemology: Essays, Oxford: Oxford University Press. Hyman, J.: 2014, ‘Desires, Dispositions and Deviant Causal Chains’, Philosophy 89/1: 83–112. Kelly, T.: 2002, ‘The Rationality of Belief and Some Other Propositional Attitudes’, Philosophical Studies 110: 163–196. Kim, J.: 1988, ‘What Is Naturalized Epistemology’, in Supervenience and Mind (1993), Cambridge: Cambridge University Press. Korcz, K.: 2000, ‘The Causal-Doxastic Theory of the Basing Relation’, Canadian Journal of Philosophy 30/4: 525–550. Lehrer, K.: 1971, ‘How Reasons Give Us Knowledge: Or the Case of the Gypsy Lawyer’, The Journal of Philosophy 68/1): 311–313. Lewis, D.: 1997, ‘Finkish Dispositions’, The Philosophical Quarterly 47: 143–158. Martin, C.: 1994, ‘Dispositions and Conditionals’, The Philosophical Quarterly 44: 1–8. Pollock, J.: 1986, Contemporary Theories of Knowledge, Totowa, NJ: Rowman & Littlefield. Pryor, J.: 2004, ‘What’s Wrong with Moore’s Argument?’, Philosophical Issues 14, Epistemology: 349–378. Schroeder, M.: 2011, ‘What Does It Take to “Have” a Reason?’, in A. Reisner & A Steglich-Petersen (eds.), Reasons for Belief, Cambridge: Cambridge University Press: 201–222. Silva, P.: 2016, ‘On Doxastic Justification and Properly Basing One’s Beliefs’, Erkenntnis 80/5: 945–955. Stoecker, R.: 2003, ‘Climbers, Pigs and Wiggled Ears: The Problem of Waywardness in Action Theory’, in Sven Walter and Heinz-Dieter Heckmann (eds.), Physicalism and Mental Causation: The Metaphysics of Mind and Action, Exeter: Imprint Academic. Stout, R.: 2005, Action, Chesham: McGill-Queen’s University Press. Tolliver, J.: 1981, ‘Basing Beliefs on Reasons’, Grazer Philosophische Studien 15: 149–161. Turri, J.: 2010, ‘On the Relationship between Propositional and Doxastic Justification’, Philosophy and Phenomenological Research 80: 312–326. Turri, J.: 2011, ‘Believing for a Reason’, Erkenntnis 74: 383–397. Vahid, H.: 2009, ‘Triangulation, Content and Basing Relation’, Grazer Philosophische Studien 78: 231–250.

Dispositions

73

Vahid, H.: 2016, ‘A Dispositional Analysis of Propositional and Doxastic Justification’, Philosophical Studies 173: 3133–3152. Vahid, H.: 2018, ‘The Dispositional Architecture of Epistemic Reasons’, Philosophical Studies, https://doi.org/10.1007/s11098-018-1102-y Wedgwood, R.: 2006, ‘The Normative Force of Reasoning’, Nous 40/4: 660–686.

4

The Many Ways of the Basing Relation Luca Moretti and Tommaso Piazza

1 Introduction A subject S’s belief that Q is well-grounded if and only if it is based on a reason of S that gives S propositional justification for Q. Depending on the nature of S’s reason, the process whereby S bases her belief that Q on it can vary. If S’s reason is non-doxastic—like an experience that Q or a testimony that Q—S will need to form the belief that Q as a spontaneous non-inferential response to that reason. If S’s reason is doxastic—like a belief that P—S will need to infer her belief that Q from it. The distinction between these two ways in which S’s beliefs can be based on S’s reasons is widely presupposed in current epistemology but—we argue in this chapter—it is not exhaustive. We give examples of quite ordinary situations in which S’s well-grounded beliefs appear to be based on S’s reasons in neither of the ways previously described. To accommodate these recalcitrant cases, we introduce the notion of enthymematic inference and defend the thesis that S can base a belief that Q on doxastic reasons P1, P2, . . . Pn via inferring enthymematically Q from P1, P2, . . ., Pn. The chapter is structured as follows: in Section 2 we detail the key principle of the orthodox view about the relation between propositional justification and well-grounded belief, and describe cases that are not accounted by it. In Section 3 we introduce the concept of enthymematic inference and correlated notions. By relying on this notional background, in Section 4 we expand the key principle of the orthodox view about propositional justification and well-grounded belief into a more comprehensive principle and show that it accounts for the recalcitrant cases. In Section 5 we draw our conclusions.

2 Justifying and Grounding Epistemic justification is—very roughly—justification for believing that a proposition is true.1 Our approach to epistemic justification throughout this chapter will be internalist accessibilist. The access internalist endorses the intuitive thesis that all the factors that provide S with epistemic justification are accessible to S by mere reflection or introspection.2

The Many Ways of the Basing Relation

75

The externalist denies this claim. It is customary to distinguish between propositional justification and doxastic justification.3 Roughly, a subject S has propositional justification for a proposition Q when S has a reason for believing Q (whether or not S believes Q and does so for that reason). S has doxastic justification for Q when S justifiedly believes Q. A widely accepted or implicitly presupposed principle that links together these two types of justification is the following: (P-WG) If (1) S has propositional justification for Q in virtue of some mental state or set of mental states X, and (2) S does believe Q on the basis of X, then (3) S’s belief Q is doxastically justified.4 For example, Jonathan Kvanvig states that: Doxastic justification is what you get when you believe something for which you have propositional justification, and you base your belief on that which propositionally justifies it. (Kvanvig 2003: 8), Furthermore, according to John Pollock and Joseph Cruz: To be justified in believing something it is not sufficient merely to have a good reason for believing it. One could have a good reason at one’s disposal but never make the connection. [In such a case] what is lacking is that you do not believe the conclusion on the basis of those reasons. (Pollock and Cruz 1999: 35–36) Since the term ‘doxastic’ can be used to refer to different things in different contexts, to avoid confusion, hereafter we prefer to follow the epistemologists who replace the expression ‘doxastically justified belief’ with the equivalent expression ‘well-grounded belief’ (see for instance Conee and Feldman 2004). Most internalist epistemologists take epistemic reasons to divide into two broad categories: non-doxastic and doxastic (cf. Pollock and Cruz 1999: 87). A doxastic reason of S for believing Q is (a belief of S in) a proposition P or a set of propositions P1, P2, . . ., Pn that counts for S as evidence for Q. A non-doxastic reason of S for believing Q is any cognitive mental state of S that supports her believing Q but that is not in turn a belief or set of beliefs (cf. Weisberg 2009). Such a mental state could for instance be a perceptual experience that Q or an apparent memory that Q of S. For the sake of simplicity, we will take perceptual experience to be a paradigmatic instance of the broader category of non-doxastic reasons.

76

Luca Moretti and Tommaso Piazza

In accordance with this dichotomy between types of reasons, S’s propositional justification for Q can flow from either S’s experiences or from S’s beliefs. In particular, in virtue of having an experience with content Q, S would have propositional justification for believing Q, and in virtue of justifiedly believing P1, P2, . . ., Pn, S would have propositional justification for believing any proposition Q that S is aware is an inductive or deductive consequence of P1, P2, . . ., Pn jointly. With this distinction in mind, let’s go back to the general principle introduced before concerning the relation between propositional justification and well-grounded belief. We can now clarify that the way in which (2) is satisfied depends on whether the mental state X referred to in (1) is an experience or a belief (or set of beliefs). When X is an experience that Q, saying that S’s belief that Q is based on X is saying that (2.1) S forms the belief that Q as a spontaneous and immediate response to her having that experience. When X consists of S’s justified beliefs that P1, P2, . . ., Pn for any P1, P2, . . ., Pn such that S is aware5 that Q is an inductive or deductive consequence of them jointly, saying that the belief that Q is based on X is saying that (2.2) S carries out an inference from P1, P2, . . ., Pn to Q. Here are some examples of how condition (2) can be fulfilled in the two alternative ways described by (2.1) and (2.2). Let P be the proposition that Jan is riding a bike. Suppose John has propositional justification for believing that P in virtue of his experience that P, and that John does form the belief that P as a spontaneous and immediate response to his having that experience. In this way, John acquires a well-grounded belief that P. Alternatively, let Q be the proposition that Jan has learned to ride a bike. Suppose that John has propositional justification for believing that Q in virtue of his perceptually justified belief that P and his awareness that Q is an inductive consequence of P. Also, suppose that John actually infers that Q from his belief that P. In this way, John comes to entertain a well-grounded belief that Q. By availing ourselves of the distinction between these two ways to satisfy condition (2), we can re-state the general principle that relates propositional justification to well-grounded belief as follows: (P-WG*) IF (1.1) Q is propositionally justified for S in virtue of her having a perceptual experience with content Q, and (2.1) S forms the belief

The Many Ways of the Basing Relation

77

that Q as a spontaneous and immediate response to her having that experience, OR (1.2) Q is propositionally justified for S in virtue of her justifiedly believing P1, P2, . . ., Pn and her being aware that Q is an inductive or deductive consequence of P1, P2, . . ., Pn jointly, and (2.2) S carries out an inference from P1, P2, . . ., Pn to Q. THEN, (3) S’s belief that Q is well-grounded. In what follows we shall argue that (P-WG*) cannot accommodate all possible cases in which it is intuitive that S has a well-grounded belief that Q. We will describe quite ordinary situations that—somewhat surprisingly— have not attracted much attention in the epistemological literature.6 These are situations in which it is intuitive that S has a well-grounded belief that Q though neither the conjunction (1.1) & (2.1) nor the conjunction (1.2) & (2.2) is fulfilled. We will present cases of two general types. In the first, S has an experience that P and is aware that Q is a consequence of P. However, S doesn’t form a belief that P, and so doesn’t infer Q from P. Rather, S directly forms the belief that Q in response to her entertaining the experience that P. In the second case-type, S is aware that Q is a consequence of P1 in conjunction with other premises P2, . . ., Pn but not from P1 alone. Furthermore, S justifiedly believes P1 and has reasons for believing P2, . . ., Pn, and yet she doesn’t come to believe P2, . . ., Pn. Thus, she doesn’t infer Q from P1, P2, . . . Pn jointly. Rather, S directly switches from P1 alone to Q. Before moving forward, we pause to pre-empt a possible criticism of (P-WG*) appealing to the epistemological view known as conservatism (cf. Wright 2007; Silins 2007; Silva 2013). The conservative contends that when S has propositional justification for believing Q in virtue of a perceptual experience that Q, this can happen only if S also possesses antecedent (possibly non-evidential) justification for believing that (R) S’s experience is reliable.7 On the conservative view, S’s perceptual belief that Q is well-grounded when it is based on both S’s experience that Q and S’s belief that R (cf. Silins 2007). This might motivate a concern about (P-WG*). For (P-WG*) doesn’t seem able to account for the way in which S’s belief that Q is based, specifically, on S’s belief that R. This is so because the belief that Q, when formed as a spontaneous response to S’s having the experience that Q isn’t simultaneously inferred from R. To attenuate this concern, it helps noting that what we take to be the most plausible form of conservatism construes the contribution of S’s experience that Q to S’s justification for Q inferentially (cf. White 2006; Silins 2007). On this view, S’s experience that Q contributes to S’s justification

78

Luca Moretti and Tommaso Piazza

for believing Q via producing S’s introspective belief that (P) she has the experience that Q. So, when S is said to form the belief that Q on the basis of the experience that Q, S is not spontaneously responding to her having that experience. Rather, S is inferring Q from her introspective belief that P and the additional premise R. The way in which S’s belief that Q can be well-grounded, in accordance with what we take to be the most plausible form conservatism, is thus fully captured by (2.2). Let’s now inspect the cases—anticipated before—that appear unaccountable by (P-WG*). In the epistemological literature it is customarily assumed that a proposition X can be relevant for a subject S’s propositional justification for believing another proposition Y, when S is aware that Y is a consequence of X, even if S doesn’t actually believe X but has only evidence for believing X (cf. Pryor 2005: 183). For instance, those who claim that propositional justification can transmit across known entailment are committed to this assumption. Suppose X entails Y. S’s justification for believing X transmits to Y just in case S has justification for believing Y in virtue of both her justification for believing X (whether or not S actually believes X) and her awareness that X entails Y (cf. Moretti and Piazza 2013). Our cases also exploit this assumption. Here is an example of the first, mono-premise type. Suppose Andrea has a visual experience with the content that (P) the little hand of her watch points to 3 and the big hand points to 12. Also, suppose that Andrea doesn’t form the belief that P, so that she doesn’t infer from P the conclusion that (Q) it’s 3 o’clock. Rather, Andrea responds to her having the experience that P by directly forming the belief that Q. In these circumstances it is intuitive that Andrea has propositional justification for Q. For whether or not Andrea forms the belief that P, her experience gives her propositional justification for believing P, and Andrea is aware that Q follows from P. It is intuitive, moreover, that Andrea’s belief that Q, based on her experience that P, is well-grounded. For instance, Andrea doesn’t seem epistemically blameworthy for holding this belief on the basis of her experience that P. Indeed, Andrea’s belief looks perfectly rational or reasonable in these circumstances. This fact, however, is not accounted for by (P-WG*). Recall that Andrea doesn’t form a belief that P, so she doesn’t infer her belief that Q from the belief that P. This shows that neither (1.2) nor (2.2) is satisfied. Note, furthermore, that Q is not (part of) the content of Andrea’s experience that P. So, Andrea doesn’t form the belief that Q as a spontaneous, immediate response to her having that experience. Thus, neither (1.1) nor (2.1) is satisfied. Nevertheless— as said—it appears true that (3) Andrea’s belief that Q is well-grounded. (P-WG*) leaves this fact unexplained. Here is another example of the same type. Kurt is driving his car from Innsbruck heading to Salzburg. He knows very well the geography of this part of Europe. At a certain time, he spots a sign indicating an exit to Rosenheim. As a spontaneous response to his having the experience that (P) there is a sign on the highway indicating an exit to Rosenheim, Kurt

The Many Ways of the Basing Relation

79

forms the belief that (Q) he is now in Germany and no longer in Austria. It is intuitive that Kurt has propositional justification for believing Q. For whether or not Kurt forms the belief that P, his experience gives him propositional justification for believing P, and Kurt is aware that Q is a consequence of P. It is also intuitive that Kurt’s belief that Q, based on his experience that P, is perfectly rational and thus well-grounded. This fact, however, is unaccounted by (P-WG*). Recall that Kurt doesn’t form the belief that P, so he doesn’t infer the belief that Q from the belief that P. Thus neither (1.2) nor (2.2) is satisfied. Note, moreover, that Q is not even a fragment of the content of Kurt’s experience that P. Thus, Andrea doesn’t form the belief that Q as a spontaneous, immediate response to his having an experience that P. Thus, neither (1.1) nor (2.1) is satisfied. Again, (P-WG*) leaves the intuitive fact that S’s belief that Q is wellgrounded unexplained. Here is an example of the second, multi-premise type.8 Suppose Terence knows that John and Jack are indistinguishable twins. Imagine that while walking to his office, Terence receives a call from his secretary who tells him that (P1) one of the twins—the secretary doesn’t know which— is waiting to see him. Terence justifiedly believes P1. Terence also apparently remembers that yesterday Jack flew to Sydney (on the opposite side of the globe), where he intended to spend some months. Terence’s apparent memory produces in him the disposition to believe (P2) that the gentleman who is waiting for him is not Jack. So Terence could also form the belief that P2, and explicitly reason from P1 and P2 to the conclusion that the person waiting to see him is John. But suppose, more realistically, that Terence doesn’t carry out the piece of reasoning described here and that, instead, he directly transitions from his apparent memory and his belief that P1 to confidently believing (Q) that John is waiting for him. Intuitively, Terence has propositional justification for believing Q. In fact, Terence justifiedly believes P1 and—though he doesn’t believe P2—Terence’s apparent memory gives him propositional justification for believing P2. Furthermore, Terence is aware that Q can be inferred from P1 and P2 jointly. It is also intuitive that Terence’s belief that Q, based on his justified belief that P1 and his apparent memory, is well-grounded. This intuition, however, is not accounted for by (P-WG*). In fact, Terence doesn’t form the belief that P1 and the belief that P2,9 from which conjunction he could infer the belief that Q. Thus neither (1.2) nor (2.2) is satisfied. Also, Terence doesn’t have an experience that Q. So, he doesn’t form the belief that Q as a spontaneous, immediate response to her having that experience. Hence neither (1.1) nor (2.1) is satisfied.

3 Plain Inferences and Enthymematic Inferences The last three examples detailed earlier show that, in certain cases, wellgrounded beliefs don’t owe their epistemic status to the satisfaction of either (1.1) & (2.1) or (1.2) & (2.2). To account for these and similar

80

Luca Moretti and Tommaso Piazza

cases, we will need to complement the two disjuncts in (P-WG*) with a third one that describes a third possible way in which S’s belief that Q can rationally be based on mental states of S. So, we will lay down three alternative sets of conditions whose satisfaction suffices to make a belief that Q of S well-grounded. The third set of conditions explains why the subject’s belief that Q is well-grounded in the former examples and suggests that also in cases of these types there is a sense in which the mental state (or set of mental states) of S that makes S’s belief that Q well-grounded is the same state (or set of states) that gives S propositional justification for believing Q. To get started we need to do some preliminary work. So far, we have relied on a standard notion of inference widely adopted in epistemology according to which a belief in the conclusion can only be inferred from a belief in the premise or a set of beliefs in the premises. We think that we can accommodate the recalcitrant cases considered before if we put in use a more liberal notion of inference. Let’s first introduce the notion of a subject’s perspective. In the way we use this notion, S’s perspective is constituted by all of S’s mental states provided with the distinctive phenomenological character—often described as phenomenal force—of presenting their contents to S as true or genuine facts.10 S’s beliefs, apparent perceptions, and apparent memories are for instance constituents of S’s perspective. A proposition P is true from S’s perspective just in case either the belief that P is a constituent of S’s perspective or there is some mental state M constitutive of S’s perspective—e.g. a perceptual experience that P—such that S’s having M causes S to acquire the disposition to believe P in appropriate circumstances. Suppose for example Mary believes that there are three bottles of beer in the fridge. Since this belief is a constituent of Mary’s perspective, the proposition that there are three bottles of beer in the fridge is true from this perspective. Moreover, Mary’s belief or perceptual experience that there are three bottles of beer in the fridge causes her to acquire the disposition to believe many different propositions she doesn’t actually believe which, for this reason, also count as true from Mary’s perspective. These propositions include, for instance, that the fridge is not empty, that there is some alcoholic drink in her house, that Mary is not in urgent need to buy beer, etc. With this notional background in place, we can now characterize a very general notion of inference. (Inference) S carries out an inference from P1, P2, . . ., Pn to Q if and only if S forms a belief that Q because (a) P1, P2, . . ., Pn are true from S’s perspective, and (b) S takes Q to be a consequence of P1, P2, . . ., Pn jointly.

The Many Ways of the Basing Relation

81

To fully understand the content of (INFERENCE), some clarifications are in order. First, P1, P2, . . ., Pn can be true from S’s perspective even if they are false. For instance, suppose Mary’s belief that there are three bottles of beer in the fridge is false. In this case the propositions that there are three bottles of beer in the fridge, or that she is not in urgent need to buy beer, are both false but still true from Mary’s perspective. Furthermore, S can take Q to be a consequence of P1, P2, . . ., Pn, even if Q is not a consequence of P1, P2, . . ., Pn. So, in our liberal sense of inference, S can infer Q from P1, P2, . . ., Pn even if P1, P2, . . . Pn are false, and Q is not a consequence of P1, P2, . . ., Pn. Second, P1, P2, . . . Pn can be true from S’s perspective even if S has no justification for P1, P2, . . . Pn. For instance, suppose Mary believes that there are three bottles of beer in the fridge out of her wishful thinking. In this case the propositions that there are three bottles of beer in the fridge and that she is not in urgent need to buy beer are both unjustified for Mary, but still true from her perspective. Of course, a proposition can be justifiedly true from S’s perspective. This happens either when the belief that P is a constituent of S’s perspective and this belief is justified or when the mental state M of S causing S to acquire the disposition to believe P gives S propositional justification for so believing. For instance, suppose Mary believes that there are three bottles of beer in the fridge because she clearly remembers seeing them in the fridge five minutes ago. In this case the propositions that there are three bottles of beer in the fridge and that she is not in urgent need to buy beer are both justifiedly true from Mary’s perspective. Third, S can take Q to be a consequence of P1, P2, . . ., Pn even if S is unable to grasp notions such as deductive consequence, inductive consequence, and their ilk. In order for S to take Q to be a consequence of P1, P2, . . ., Pn, it is sufficient (and necessary) that Q be available to her as a consequence of P1, P2, . . . Pn in McCain (2014)’s internalist sense. That is to say, it is sufficient (and necessary) that S either entertain the appearance or seeming that if P1, P2, . . ., Pn are jointly true, then Q is true or probable,11 or be at least disposed to entertain that appearance or seeming whenever S is explicitly invited to consider the question. Inference as just characterized is a genus that can be differentiated in different species depending on how condition (a) is fulfilled. In one first range of cases, condition (a) can be satisfied because (a1) S actually believes that P1, P2, . . ., Pn. A second possibility is when (a2) S doesn’t actually believe at least some of the premises P1, P2, . . ., Pn, though some constituents M1, M2, . . ., Mm of S’s perspective cause in S the disposition to believe the non-believed premises.

82

Luca Moretti and Tommaso Piazza

A third possibility, which is included in the second possibility is when: (a3) S doesn’t actually believe any of P1, P2, . . ., Pn, though some constituents M1, M2, . . ., Mm of S’s perspective cause in S the disposition to believe P1, P2, . . ., Pn. If S forms the belief that Q because (a1) is satisfied, and (b) S takes Q to be a consequence of P1, P2, . . ., Pn, then S performs a plain inference. This is the standard interpretation of ‘inference’—the one presupposed in condition (2.2) of (P-WG*). If S forms the belief that Q because (a2) or (a3) is satisfied, and (b) S takes Q to be a consequence of P1, P2, . . ., Pn, then S performs an enthymematic inference. This inference is not an inference in the standard sense presupposed in (2.2). It also deserves emphasis that the transition described by condition (1.2) is not an inference in this enthymematic sense. One might believe the contrary because when S has a perceptual experience that Q and forms the belief that Q as a spontaneous response to her having that experience, S is forming a belief in a proposition Q that she (trivially) takes to be a consequence of the experience’s content (that is to say, Q itself). Crucially, however, in this case S doesn’t form her belief that Q because S takes Q to be a consequence of itself. Rather S forms the belief that Q in response to the phenomenal force of her perceptual experience that Q. To further illuminate the notions of plain and enthymematic inference we need to clarify what it takes for S to form the belief that Q because condition (a) and condition (b) are fulfilled. Let’s first address this question in relation to (a1) and (b). In this case the question is what it takes for S to form the belief that Q because S believes that P1, P2, . . ., Pn are true and considers Q to be a consequence of P1, P2, . . ., Pn. This is the same as asking what it takes for S to plainly infer Q from P1, P2, . . ., Pn. The fact that S believes Q because (a1) is satisfied involves—trivially— that S believes Q because S believes P1, P2, . . ., Pn. The fact that S forms the belief that Q because (b) S takes Q to be a consequence of P1, P2, . . ., Pn involves—less trivially—that S forms the belief that Q because S’s beliefs that P1, P2, . . ., Pn cause in her the belief that Q in a way that is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn. We think that the difference between a causal relation from S’s beliefs that P1, P2, . . ., Pn to S’s belief that Q that is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn and a causal relation from S’s beliefs that P1, P2, . . ., Pn to S’s belief that Q that is not shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn should be intuitively clear, at least to some extent. However, an analogy with Ryle (1949)’s distinction between an action that is intelligently executed and an action that isn’t intelligently executed might further illuminate this difference. S’s breathing or swallowing a sandwich are not actions intelligently executed by S, while S’s carefully folding a sheet to make a paper ship or S’s driving a car

The Many Ways of the Basing Relation

83

are actions that S could only intelligently carry out. Actions of these types are sustained by causal processes that occur at some level. The processes that underwrite the actions that are not intelligently executed by S are not shaped by S’s knowledge of how to perform those actions, while the processes that underwrite the actions that are intelligently executed by S are shaped by S’s knowledge of how to perform those actions. In a similar way, the transitions whereby some beliefs of S cause another belief of S can be executed by S more or less intelligently depending on whether or not the underlying causal processes are shaped by S’s taking the latter belief to be a consequence (deductive or inductive) of the former beliefs. Some philosophers—e.g. Boghossian (2014)—hold that the causal process whereby S forms a belief that Q from the beliefs that P1, P2, . . ., Pn can count as an inference even if the process is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn at a sub-personal level. When the causal relation from S’s beliefs that P1, P2, . . ., Pn to S’s belief that Q is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn at a personal level, S is aware or potentially aware that Q is a consequence of P1, P2 . . . ., Pn. If the same causal relation is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn at a sub-personal level, then S isn’t even potentially aware that Q is a consequence of P1, P2, . . ., Pn (cf. Boghossian 2014). The internalist view of epistemic justification that we have adopted in this chapter imposes that S can acquire inferential justification for Q from P1, P2, . . ., Pn only if S is aware or potentially aware that Q is a consequence of P1, P2, . . ., Pn. In fact, as clarified earlier, on this view, S can take Q to be a consequence of P1, P2, . . ., Pn only if S entertains or is disposed to entertain the appearance that if P1, P2, . . ., Pn are jointly true, then Q is true or probable. Accordingly, saying that S’s beliefs that P1, P2, . . ., Pn cause S’s belief that Q in a way that is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn is just saying that S’s beliefs that P1, P2, . . ., Pn cause that belief in a way that is shaped by S’s having or being disposed to have the appearance that Q is a consequence of P1, P2, . . ., Pn. On this view, the causal process whereby S forms the belief that Q from the beliefs that P1, P2, . . ., Pn can count as a plain inference only if the process is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn at a personal level. To summarize, it appears true that (Plain Inference) S carries out a plain inference from P1, P2, . . ., Pn to Q if and only if (a1) S believes P1, P2 . . . Pn, and (b1) S’s beliefs that P1, P2, .  .  ., Pn jointly cause S’s belief that Q through a process shaped by S’s taking Q to be an inductive or deductive consequence of P1, P2, . . ., Pn at a personal level.

84

Luca Moretti and Tommaso Piazza

One might wonder whether there is a criterion for ascertaining whether psychological causal processes of the type described in (PLAIN INFERENCE) are actually shaped by S’s taking the conclusion to follow from the premises at a personal level. We suggest this criterion: (Personal Level) S’s beliefs that P1, P2, . . ., Pn jointly cause S’s belief that Q through a process shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn at a personal level if and only if S’s beliefs that P1, P2, . . ., Pn jointly cause S’s belief that Q, and S would adduce the reasons that P1, P2, . . ., Pn and that Q is a consequence of P1, P2, . . ., Pn in response to a request to explain why she believes Q. Boghossian (2014) criticizes a principle in the neighbourhood of (PERSONAL LEVEL). This principle says that, roughly, a subject S counts as taking a set of premises to support a conclusion if and only if S would offer the premises as her reasons when asked why she believes the conclusion.12 According to Boghossian, this principle is implausible because it commits their advocates to claiming that what S is disposed to say in response to the question about why she has formed a given belief makes it the case that she has formed that belief for a particular reason. But this claim is very odd and implausible, and so is the principle. Adapted to (PERSONAL LEVEL), the objection would say that this principle commits us to the implausible view that the fact that S would adduce P1, P2, . . ., Pn in response to the question why she has formed the belief that Q makes it the case that the causal link from S’s beliefs that P1, P2, . . ., Pn to S’s belief that Q was shaped by S’s taking Q to follow from P1, P2, . . ., Pn. To pre-empt this possible criticism, we emphasize that (PERSONAL LEVEL) is not proposed as an analysis of what it takes for the causal link from S’s beliefs that P1, P2, . . ., Pn to S’s belief that Q to be shaped at a personal level by S’s considering Q to follow from P1, P2, . . ., Pn. (PERSONAL LEVEL) is only meant to be a criterion that interprets the truth of its right-hand side as a signal (or symptom) of the truth of its left-hand side. Hence, (PERSONAL LEVEL) doesn’t commit us to accepting the odd claim that the satisfaction of the counterfactual embedded in its right-hand side makes it the case that its left-hand side is true. So far, we have focused on plain inference. We have (hopefully) clarified what it takes for S to form the belief that Q because condition (a1) and condition (b) are fulfilled. Let’s now turn to enthymematic inference, the execution of which by S depends on the satisfaction of (a2) or (a3) and (b). An inference of this type differs from a plain inference only because of (a2) or (a3)—namely, only because the subject S who performs it doesn’t move from believing all its premises to believing its conclusion. More precisely, the difference consists in the fact that when

The Many Ways of the Basing Relation

85

drawing an enthymematic inference, S doesn’t believe at least some of the premises, though certain constituents of S’s perspective cause S to have the disposition to believe the non-believed premises. An inference from P1 . . . Pn to Q can be partly enthymematic or fully enthymematic. It is partially enthymematic if S believes some but not all of its premises and fully enthymematic if S believes no premise at all. More accurately: (Partly Enthymematic Inference) S carries out a partly enthymematic inference from P1, P2, . . ., Pn to Q if and only if (a2) S doesn’t actually believe at least some of the premises P1, P2, . . ., Pn, though some constituents M1, M2, . . ., Mm of S’s perspective cause in S the disposition to believe the non-believed premises, and (b2) M1, M2, .  .  ., Mm together with the premises believed by S jointly cause S’s belief that Q through a process that is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn at a personal level. (Fully Enthymematic Inference) S carries out a fully enthymematic inference from P1, P2, . . ., Pn to Q if and only if (a3) S doesn’t actually believe any of P1, P2, . . ., Pn, though some constituents M1, M2, . . ., Mm of S’s perspective cause in S the disposition to believe P1, P2, . . ., Pn, and (b3) M1, M2, . . ., Mm jointly cause S’s belief that Q through a process that is shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn at a personal level. These two characterizations raise a question: suppose M1, M2, . . ., Mm together with any premise believed by S jointly cause S’s belief that Q through a process shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn. Is there any criterion to ascertain whether or not, in this process, S takes Q to be a consequence of P1, P2, . . ., Pn at a personal level? We suggest the following principle, which parallels (PERSONAL LEVEL): (Personal Level*) S’s mental states M1, M2, . . ., Mm and any premises believed by S, among P1, P2, . . ., Pn, jointly cause S’s belief that Q through a process shaped by S’s taking Q to be a consequence of P1, P2, . . ., Pn at a personal level if and only if M1, M2, . . ., Mm and any premise believed by S, among P1, P2, .  .  ., Pn, jointly cause S to believe Q and S would adduce the reasons that P1, P2, . . ., Pn and that Q is

86

Luca Moretti and Tommaso Piazza a consequence of P1, P2, . . ., Pn in response to a request to explain why she believes Q.

4 Expanding (P-WG*) and Accounting for Recalcitrant Cases Let’s return to (P-WG*). Before expanding this principle with the addition of a third disjunct capable of accounting for recalcitrant cases like those described in Section 2, we would like to re-phrase its second disjunct, that is to say, (1.2) Q is propositionally justified for S in virtue of her justifiedly believing P1, P2, . . ., Pn and her being aware that Q is an inductive or deductive consequence of P1, P2, . . ., Pn jointly, and (2.2) S carries out an inference from P1, P2, . . ., Pn to Q, into an equivalent statement, which exploits the notions of a subject’s perspective and plain inference, introduced previously. This is the re-formulation: (1.2*) Q is propositionally justified for S in virtue of P1, P2, .  .  ., Pn being justifiedly true from her perspective because S justifiedly believes P1, P2, . . ., Pn, and in virtue of her being aware that Q is an inductive or deductive consequence of P1, P2, . . ., Pn jointly, and (2.2*) S carries out a plain inference from P1, P2, . . ., Pn to Q. (1.2*) is equivalent to (1.2), for saying that Q is propositionally justified for S in virtue of P1, P2, . . ., Pn being justifiedly true from S’s perspective because S justifiedly believes P1, P2, . . ., Pn is the same as saying that Q is propositionally justified for S in virtue of S’s justifiedly believing P1, P2, . . ., Pn. Furthermore, as already clarified, ‘inference’ in (2.2) stands for ‘plain inference’, used in (2.2*). Suppose now that S doesn’t believe at least some of the premises P1, P2, . . ., Pn but that all these premises are justifiedly true from S’s perspective. This means that the premises that S actually believes, if any, are justifiedly believed by S, and that the mental states M1, M2, . . ., Mm that produce in S the disposition to believe the premises that she actually doesn’t believe give S propositional justification for them. Furthermore, suppose that S is aware that Q is a consequence of P1, P2, . . ., Pn. It is intuitive that when these two conditions are satisfied, S has propositional justification for believing Q in virtue of the premises that she justifiedly believes, among P1, P2, . . ., Pn, and in virtue of the mental states M1, M2, . . ., Mm that produce in S the disposition to justifiedly believe the premises that she doesn’t believe, among P1, P2, . . ., Pn. Imagine that, in these very circumstances, S carries out an enthymematic inference from P1, P2, . . ., Pn to Q. (Where the inference is fully or partly enthymematic depending on, respectively, whether

The Many Ways of the Basing Relation

87

S doesn’t believe any P1, P2, . . ., Pn or believes only some of them.) It is intuitively plausible that S’s belief that Q will be well-grounded in this case. We have just determined a new third set of conditions, unaccounted for by (P-WG*), whose satisfaction makes S’s belief that Q well-grounded. We are now in position to supplement (P-WG*) with a third disjunct. (P-WG*) IF (1.1) Q is justified for S in virtue of her having an experience with content Q, and (2.1) S forms the belief that Q as a spontaneous and immediate response to her having that experience, OR (1.2*) Q is propositionally justified for S in virtue of P1, P2, .  .  ., Pn being justifiedly true from her perspective because S justifiedly believes P1, P2, . . ., Pn, and in virtue of her being aware that Q is an inductive or deductive consequence of P1, P2, . . ., Pn jointly, and (2.2*) S carries out a plain inference from P1, P2, . . ., Pn to Q. OR (1.3) Q is propositionally justified for S in virtue of P1, P2, . . ., Pn being justifiedly true from her perspective, though S doesn’t believe at least some P1, P2, . . ., Pn, and in virtue of S being aware that Q is an inductive or deductive consequence of P1, P2, . . ., Pn jointly, and (2.3) S carries out a (fully or partly) enthymematic inference from P1, P2, . . ., Pn to Q. THEN (3) S’s belief that Q is well-grounded. Let’s have a look at the third disjunct of (P-WG*). When condition (2.3) is satisfied, the constituents M1, M2 . . ., Mm of S’s perspective that produce in S the disposition to believe any premise, among P1, P2 . . ., Pn, and S’s actual beliefs (if any) in P1, P2, . . ., Pn jointly cause S’s belief that Q in a way that is shaped by S’s taking Q to be a consequence of P1 .  .  . Pn at a personal level. That much follows from (ENTHYMEMATIC INFERENCE). Moreover, when condition (1.3) is satisfied, the mental states M1, M2, .  .  ., Mm that produce in S the disposition to believe any P1, P2 .  .  ., Pn also give S propositional justification for so believing. Thus, for any premise P1, P2, . . ., Pn, S either justifiedly believes it or has justification for believing it. Since S is also aware that Q is a consequence of P1, P2, . . ., Pn, S has propositional justification for believing Q in virtue of any premise she justifiedly believes, among P1, P2, . . ., Pn, and in virtue of the mental states M1, M2, . . ., Mm that

88

Luca Moretti and Tommaso Piazza

produce in S the disposition to justifiedly believe the premises that she doesn’t believe. In conclusion, the joint satisfaction of (1.3) and (2.3) guarantees that S’s mental states M1, M2, . . ., Mm and S’s actual beliefs (if any) in any P1, P2, . . ., Pn, which jointly cause S to believe Q in a way that is shaped at a personal level by S’s taking Q to follow from P1, P2, . . ., Pn, are the same mental states in virtue of which S has propositional justification for believing Q to begin with. In this sense, episodes of belief formation involving enthymematic inferences that meet conditions (1.3) and (2.3) can still be regarded as cases in which what provides S with propositional justification for Q is also what rationally grounds S’s beliefs that Q. With this general moral on the table, let’s move to examine how conditions (1.3)–(2.3) accommodate the examples introduced in Section 2 to challenge the completeness of (P-WG*). Consider one more time Andrea’s case. She has a visual experience with the content that (P) the little hand of her watch points to 3 and the big hand points to 12. Andrea doesn’t form the belief that P. Rather, Andrea responds to her having the experience that P by directly forming the belief that (Q) it’s 3 o’clock. In this case, Andrea has propositional justification for Q in virtue of P being justifiedly true from her perspective—Andreas’s experience that P gives her propositional justification for believing P—and in virtue of Andrea’s awareness that Q is a consequence of P. Hence condition (1.3) is satisfied. Furthermore, (2.3) is also satisfied. For Andrea comes to believe Q from her entertaining the experience that P by carrying out a fully enthymematic inference. In fact, her experience that P causes her belief that Q through a process that is shaped by her taking Q to be a consequence of P at a personal level. (This is revealed by the fact that Andrea would certainly cite the reason that P and that Q is a consequence of P in response to a request to explain why she believes Q.) In conclusion, (P-WG*)—in particular its third disjunct—accounts for the intuition that Andrea’s belief that Q is well-grounded in the case envisaged. Kurt’s case can be accounted for by (P-WG*) in quite a similar way. Kurt has a visual experience with the content that (P) there is a sign on the highway indicating an exit to Rosenheim. As a spontaneous and immediate response to this, Kurt forms the belief that (Q) he is now in Germany. Kurt has propositional justification for Q in virtue of P being justifiedly true from his perspective—Kurt’s experience that P gives him propositional justification for believing P—and in virtue of Kurt’s awareness that Q is a consequence of P. Hence, condition (1.3) is satisfied. Furthermore, (2.3) is also satisfied. For Kurt arrives at believing Q from his having the experience that P via performing a fully enthymematic inference. Indeed, his experience that P causes his belief that Q through a process that is shaped by his taking Q to be a consequence of P at a personal level. (This is revealed, again, by the fact that Kurt would surely

The Many Ways of the Basing Relation

89

cite the facts that P and that Q is a consequence of P in response to a request to explain why he believes Q.) Therefore, (P-WG*)—specifically its third disjunct—accounts for the intuition that Kurt’s belief that Q is well-grounded in the imagined scenario. Let’s now turn to Terence’s case. Terence knows that John and Jack are indistinguishable twins. On his way to the office, Terence takes a call from his secretary who tells him that (P1) a gentleman who looks like one of the twins has just asked to meet him. Terence forms the justified belief that P1. Terence also seems to remember that yesterday Jack flew to Sydney. Since Terence’s apparent memory is a reason for believing (P2) that the gentleman waiting for him is not Jack, Terence forms the belief (Q) that John has just asked to see him. In this case, Terence has propositional justification for believing Q in virtue of his being aware that P1 and P2 jointly imply Q, and in virtue of P1 and P2 being justifiedly true from Terence’s perspective. This is so because Terence justifiedly believes P1 and his memory gives him propositional justification for believing P2. Hence condition (1.3) is fulfilled. (2.3) is also fulfilled. For Terence comes to believe Q from his belief that P1 and his entertaining the memory that produces in him the disposition to believe P2. Terence thus carries out a partially enthymematic inference. Indeed, his belief that P1 and his memory jointly cause his belief that Q through a process that is shaped by his taking Q to be a consequence of P1 and P2 at a personal level. (Terence would adduce P1 and P2, and that Q is a consequence of P1 and P2 to explain why he believes Q.) In conclusion, (P-WG*)—specifically its third disjunct—accounts for the intuition that Terence’s belief that Q is well-grounded in this example.

5 Concluding Considerations In this chapter we have defended the view that there are more ways in which a subject S can form a well-grounded belief that Q than epistemologists have been able to identify. In particular, we have defined a notion of inference more liberal than the one typically presupposed in the epistemological literature, and we have shown that by putting in use this more liberal notion of inference we can accommodate many cases in which a subject intuitively forms a well-grounded belief. Interestingly, the account we have proposed is a natural extension of the standard account that we aimed to complement. For all the cases involving a well-grounded belief we have reviewed still count as cases in which what supplies S with propositional justification for believing Q is the ground of S’s belief that Q.

Notes 1. Although the property of being epistemically justified can be predicated of propositional attitudes other than belief, for the sake of simplicity, in this

90

2. 3. 4.

5.

6. 7. 8. 9. 10. 11. 12.

Luca Moretti and Tommaso Piazza paper we focus only on belief. Throughout this work we also assume that the claim that S has justification (or reason) for believing P and the claim that P is justified for S are substantially equivalent in meaning. A less popular and more controversial form of internalism—called mentalism— states that all the factors that provide S with epistemic justification are mental states of S. Whenever we speak of justification or reason, unless otherwise specified, we mean prima facie justification or prima facie reason. Turri (2010) has found apparent counterexamples to this general principle. This is not the place to assess Turri’s arguments. Since Turri’s cases have no direct bearing on the questions we deal with in this paper, we prefer to set the whole issue aside. Furthermore, we tend to agree with Silva (2015), that a very modest revision of the above general principle would suffice to make it survive Turri’s objections. From an internalist point of view, this type of awareness should not be thought of as factive. For the internalist, S can have inferential justification for believing Q from P even if it is false that Q is a consequence of P, though it appears to S that Q is a consequence of P. (See for Huemer 2016 for discussion.) The type of awareness required for S can be cashed out in terms of S’s entertaining an inferential seeming that P supports Q, or S’s being disposed to entertain such a seeming. More on this later. Pryor (2005: 183) consider cases of this type but in a different context. Or that relevant sceptical conjectures are false (depending on the variant of conservatism). Suggested by a similar example described in Wright (2003). One might contend that an apparent memory that P is nothing but a belief that P. But this is implausible because one can have an apparent memory that P while disbelieving P. See mainly instance Pryor (2000) and Huemer (2001). For theories of inferential appearances of this type see for instance Chudnoff (2014), Brogaard (2016) and Huemer (2016). This is similar to the view of basing advanced by Leite (2004).

References Boghossian, P. 2014. ‘What Is Inference?’. Philosophical Studies 169(1): 1–18. Brogaard, B. 2016. ‘Staying Indoors: How Phenomenal Dogmatism Solves the Skeptical Problem without Going Externalist’. In B. Coppenger and M. Bergmann (eds.), Intellectual Assurance: Essays on Traditional Epistemic Internalism. Oxford: Oxford University Press: 85–104. Chudnoff, E. 2014. ‘The Rational Roles of Intuitions’. In A. R. Booth and D. P. Rowbottom (eds.), Intuitions. Oxford: Oxford University Press: 9–35. Conee, E. and R. Feldman. 2004. Evidentialism: Essays in Epistemology. Oxford: Oxford University Press. Huemer, M. 2001. Skepticism and the Veil of Perception. Lanham, MD: Rowman and Littlefield. Huemer, M. 2016. ‘Inferential Appearances’. In B. Coppenger and M. Bergmann (eds.), Intellectual Assurance: Essays on Traditional Epistemic Internalism. Oxford: Oxford University Press: 144–160. Kvanvig, J. 2003. ‘Propositionalism and the Perspectival Character of Justification’. American Philosophical Quarterly 40: 3–18.

The Many Ways of the Basing Relation

91

Leite, A. 2004. ‘On Justifying and Being Justified’. Philosophical Issues 14: 219–253. McCain, K. 2014. Evidentialism and Epistemic Justification. New York: Routledge. Moretti, L. and T. Piazza. 2013. ‘Transmission of Justification and Warrant’. Stanford Encyclopedia of Philosophy. https://plato.stanford.edu/entries/transmissionjustification-warrant/ Pollock, J. and J. Cruz. 1999. Contemporary Theories of Knowledge (2nd ed.). Lanham, MD: Rowman & Littlefield. Pryor, J. 2000. ‘The Skeptic and the Dogmatist’. Nous 34: 517–549. Pryor, J. 2005. ‘There Is Immediate Justification’. In M. Steup and E. Sosa (eds.), Contemporary Debates in Epistemology. Oxford: Blackwell: 257–269. Ryle, G. 1949. The Concept of Mind. Chicago: Chicago University Press. Silins, N. 2007. ‘Basic Justification and the Moorean Response to the Skeptic’. In T. S. Gendler and J. Hawthorne (eds.), Oxford Studies in Epistemology (Vol. 2). Oxford: Oxford University Press: 108–140. Silva, P. 2013. ‘How to Be Conservative: A Partial Defence of Epistemic Conservatism’. Australasian Journal of Philosophy 91: 501–514. Silva, P. 2015. ‘On Doxastic Justification and Properly Basing One’s Beliefs’. Erkenntnis 80: 945–966. Turri, J. 2010. ‘On the Relationship between Propositional and Doxastic Justification’. Philosophy and Phenomenological Research 80: 312–326. Weisberg, J. 2009. ‘Commutativity or Holism? A Dilemma for Jeffrey Conditionalizers’. British Journal for the Philosophy of Science 60: 793–812. White, R. 2006. ‘Problems for Dogmatism’. Philosophical Studies 131: 525–557. Wright, C. 2003. ‘Some Reflections on the Acquisition of Warrant by Inference’. In S. Nuccetelli (ed.), New Essays on Semantic Externalism and Self-Knowledge. Cambridge, MA: MIT Press: 57–77. Wright, C. 2007. ‘The Perils of Dogmatism’. In S. Nuccetelli (ed.), Themes from G. E. Moore: New Essays in Epistemology and Ethics. Oxford: Oxford University Press: 25–48.

5

Reasons and Basing in Commonsense Epistemology Evidence From Two Experiments John Turri

Introduction Images and Methods The most important development in philosophy this century has been the rapidly growing dissatisfaction with a primarily armchair methodology, and a corresponding growth in the use of empirical evidence and science to address philosophical questions. To appreciate some of the motivation for this change, consider Wilfrid Sellars’ evocative and rightly famous distinction between the manifest and scientific images. The manifest image consists of our ordinary ways of understanding the world and our place in it, our “pre-reflective orientation and common heritage” that “took shape in the mists of prehistory” (Sellars 1963: pp. 3, 5). The scientific image consists of the deliverances of organized scientific inquiry, which can diverge radically from commonsense. Sellars proposed that contemporary philosophy’s distinctive objective is to unify these different images into a coherent whole, to transcend this “duality” of images. While it is immediately obvious that getting straight on the scientific image requires empirical evidence and, well, science, it turns out that the same is true for the manifest image. For it is an “essentially social” phenomenon, marked by “common standards of correctness and relevance,” that “transcends” individual thinkers and has “an objective existence . . . in human thought generally” (Sellars 1963: p. 14). Objective social phenomena are objects of scientific investigation in their own right. Setting aside the wonderful metaphor, which I find deeply fascinating and insightful, the basic idea can be stated as follows. There is what commonsense says the world is like, and then there is what our best evidence says the world is like. Empirical science provides our best evidence about what the world is like, so this side of the ledger must involve empirical scientific evidence. It is tempting to think that the other side of the ledger needn’t involve empirical scientific evidence. After all, if it’s commonsense, then we all share it and one could successfully characterize it through careful and systematic reflection, perhaps supplemented by

Commonsense Epistemology

93

comparing notes occasionally with others engaged in reflection of similarly high quality. Sellars gives voice to this temptation when he claims that many philosophical systems have “more or less adequate[ly]” characterized commonsense, and that twentieth-century anglophone philosophy has captured it “in something like its pure form” (Sellars 1963: p. 15). It is true that social observation and reflection provide insight into our shared standards and practices. Life as we know it would be impossible otherwise. But the insight afforded, albeit impressive, is far from complete, and systematic scientific inquiry is needed to begin filling in the gaps, as Sellars went on to acknowledge (Sellars 1963: p. 15). Generations of social and cognitive scientists have made progress on this front, including work ranging from intuitive physics (e.g. McCloskey, Washburn, and Felch 1983), to folk biology (e.g. Berlin, Breedlove, and Raven 1973), to commonsense psychology (e.g. Kelley 1992). They have also begun the daunting task of reconciling commonsense and the deliverances of science in ways that are meaningful and useful to us. They do this by identifying the limitations of our pretheoretical understanding and developing techniques to integrate hard-won theoretical knowledge to improve judgment and decision-making (e.g. Schtulman 2017; Gigerenzer 2014; Kosko 1993). For example, people can learn to respect base rates and remedy their ignorance about how diseases are actually transmitted. Decades into these challenging pursuits, there is much left to learn. But if Sellars is correct that reconciling science and commonsense is philosophy’s distinctive task, then it turns out that these impressive philosophical accomplishments are due to scientists. Another motivation for the increasing role of scientific methods in philosophy is less inspiring, what we might call the decades long false start. Philosophers have made numerous false claims about ordinary thought and speech. Of course, researchers in every field make mistakes, so in itself, this is unremarkable. What makes these mistakes remarkable, regardless of the field they occur in, is that they follow a pattern that teaches a valuable lesson to those willing to learn (for reviews, see Turri (2016), Buckwalter and Turri (2018), Turri (2019)). The pattern begins with one philosopher making an empirical claim, based on introspection or anecdotal social observation, about how we ordinarily speak or classify objects. The philosopher relies on this claim to motivate or defend a theory of some ordinary concept of philosophical interest. Other philosophers accept the claim but dispute its theoretical significance and propose alternative explanations. An entire literature grows up around the dispute, which enters the annals via textbooks, handbooks, and encyclopedia entries. A central tendency in this literature is to treat the original empirical claim as a datum to be explained, or, at worst, explained away. Many years later, someone decides to go back and check whether the original empirical claim is actually true. A few simple experiments provide clear evidence that it isn’t. Thus the debate was predicated on a

94

John Turri

false assumption and all the time and energy devoted to it would have been better spent theorizing about actual facts. In addition to opportunity cost for the field as a whole, let’s not lose sight of the human element. This sort of outcome can be especially disappointing and frustrating for people invested in the debate. For many of us, it’s not easy to let go of something that has been a source of personal satisfaction and professional success. This can create additional obstacles to advancing collective understanding of the underlying issues. We should do our best to avoid unforced errors like this, and normalizing experimentation within the discipline has helped. Again, philosophy is not special in this regard. Researchers in all fields should be quick to rely, when possible, on empirical evidence and controlled experiments to separate wheat from chaff, especially early in a research program when formative clues can put researchers on more promising paths. Reasons and Basing Epistemic reasons and the basing relation are a closely related pair of philosophical topics that have not, as far as I’m aware, been directly subjected to empirical scrutiny. My goal in the present chapter is to begin filling that gap. With respect to the basing relation, I will focus on the lead premise in the principal argument for a causal theory of the basing relation. According to this theory, to believe something for a reason is for the reason to non-deviantly cause the belief (Turri 2011). The principal argument for this view begins from the premise that, intuitively, reasons are “differencemakers.” Simplifying, the argument then proceeds to claim that the best explanation for why reasons are difference-makers is that reasons are causes. The argument relies on intuition or verdicts about cases at more than one juncture, all of which could be worth exploring. My focus here will be the most important juncture, which is whether reasons are intuitively viewed as difference-makers. If they aren’t, then that poses a serious problem for the argument. I am not aware of existing findings on the concepts of “difference-making,” reasons, or inference that would lead us to expect one outcome or another. With respect to epistemic reasons, I will focus on three competing theories about the kind of thing reasons are (Turri 2009). One view is that all reasons are psychological items of the agent, such as beliefs, knowledge, emotions, and perceptual and sensory states. A second view is that all reasons are non-psychological items, such as physical objects, external facts, or states of affairs. A third view is that reasons include both psychological and non-psychological items. Call these views psychologism, anti-psychologism, and dualism, respectively. The principal case for psychologism and against anti-psychologism and dualism relies on the premise that being given a description of the

Commonsense Epistemology

95

agent’s mental condition “is sufficient for us to understand” his reasons, in which case “it is superfluous” to identify additional, non-psychological items as reasons (Turri 2009: p. 504). This appears closely related to the following assumption about radically deceived agents: Ideally a theory should respect the following intuition. Your victimized twin, Vic, recently ensnared by an evil genius, undergoes an indefinite succession of experiences indistinguishable from yours. Responding to the very same kinds of sensory experiences as you, Vic believes that he or she is reading an epistemology paper. Intuitively you and your twin believe what you do for the same reasons. (Turri 2009: p. 493) With respect to this assumption, I believe existing findings should lead us to expect that it is false. In particular, research on the mistaken “extramission” folk theory of perception counts against it (Cottrell and Winer 1994; Winer et al. 1996, 2002). According to this theory, when people perceive, it is because a force emanates from their sense organs and interacts with the perceived object. For example, one sees an object because rays emanate from one’s eyes and strike the object. Vision science shows that this is exactly backwards: one sees because rays emanating from the object strike one’s eye. The naive extramission theory is present from early in childhood, persists into adulthood, shows up on a variety of measures, and is surprisingly resistant to explicit instruction. One psychologist characterizes the ordinary concept of perception in a way consistent with the extramission theory: “Perceiving is experienced as a direct contact with the environment; it is a means whereby objective facts enter the life space” (Heider 1958: p. 15). In light of existing evidence, then, we should expect that psychologism is false as a theory of the ordinary reason concept. The remaining question is whether a direct test of the matter will favor anti-psychologism or dualism. Experiment 1 focuses on reasons and difference-making. Experiment 2 focuses on the ontology of reasons. To preview the main findings, the results of experiment 1 provide initial support for the view that reasons are difference-makers. The results of experiment 2 provide strong initial support for a dualist theory of reasons.

General Methods The following statements are true of all studies reported here. I report all manipulations and measures used. All participants were adult residents of the United States. No participants were excluded from analysis. I recruited and tested people using an online platform of Amazon Mechanical Turk (www.mturk.com), TurkPrime (Litman, Robinson, and Abberbock 2017), and Qualtrics (www.qualtrics.com). Participants completed

96

John Turri

a brief demographic questionnaire after testing. I used R 3.5.2 for all analyses (R Core Team 2018).

Experiment 1: Difference Making Method No research on the topic existed to inform an a priori power analysis regarding sample size, so I defaulted to the conventional “rule of thumb” of 30 participants per condition, plus a few extra as a precaution against participants dropping out or refusing to sign the consent form after initially clicking to participate. Participants Participants’ mean age was 35.29 years (range = 19–76, sd = 10.69), and 56% (172 of 305) were female. Materials and Procedure Participants were randomly assigned to one of ten conditions in a 2 (Status: none, some) × 5 (Scenario) mixed experimental design. Participants first read a brief scenario about an agent considering a question, then responded to two test statements (within-subjects). The Status factor manipulated whether a potential basis (reason) was described as making no difference to the agent’s thinking (none), or as making a difference (some). The Scenario factor manipulated what question the agent was considering, and other features of her situation. I included this factor to support generalization of the results beyond the specific stimuli studied here (Judd, Westfall, and Kenny 2012; Clark 1973; Baayen, Davidson, and Bates 2008). All stimuli used in this study are included in an appendix. To give readers a sense of the materials, I include one scenario and the test statements here, with the Status (none/some) manipulation shown in brackets. (Juror) Lauren is on a jury hearing a criminal case. After listening carefully throughout the trial, Lauren concludes that the defendant was innocent. A grainy photograph seemed to place the defendant a mile away from the scene of the crime. This photograph made [absolutely no/a] difference to Lauren’s thinking on the matter. 1. Lauren’s conclusion was partly based on the photograph. (basing) 2. The photograph made a difference to Lauren’s thinking. (difference) The basing statement appeared below the scenario on the first screen of the study. Participants then advanced to a new screen (they could

Commonsense Epistemology

97

not go back) and responded to the difference statement from memory. Responses were collected on a standard 7-point Likert scale, 1 (“strongly disagree”) — 7 (“strongly agree”), arranged left-to-right on the participant’s screen. Data Analysis One principal research question is whether Status would affect response to the basing statement. If it doesn’t, then the results would undermine a primary argument for a causal theory of the basing relation. If it does, then the results would support the causal theory. Relatedly, it will be informative to discern whether such a Status effect is robust across the scenarios and the demographic variables of sex and age. A second research question pertains to the relationship between the basing and difference statements. To the extent that participants respond differently to them, it puts pressure on the causal theory of the basing relation. To the extent that participants respond similarly to them, it supports the causal theory. To address these questions, I conducted a linear mixed effects analysis on participant response and followed up with appropriate correlation analysis and t-tests. Results I conducted a linear mixed effects analysis of participant response. I included as fixed effects Status (none, some), the type of judgment (basing, difference), an interaction between Status and type of judgment, and participant age and sex. I included random intercepts for Scenario (five variants) and participant nested within Scenario. Step-wise elimination of non-significant effects (Kuznetsova, Brockhoff, and Christensen 2017) resulted in a reduced model including only Status and participant, which did not differ significantly from the full model, χ2(5) = 3.27, p = .659. The reduced model explained 92 per cent of variance in participant response (Nakagawa and Schielzeth 2012). The difference between Status conditions explained 54%, and the difference among participants explained 38%. A bivariate correlation analysis on the basing and difference statements found that they were extremely strongly correlated, r(303) = 0.92 [0.9, 0.93], p < .001. (See Figure 5.1.) A paired samples t-test on the two statements detected no difference, t(304) = −0.7, MD = −0.04 [−0.14, 0.07], p = .483. Independent samples t-tests revealed that mean response to the basing statement was significantly lower in the none condition (M = 2.54, SD = 1.75) than in the some condition (M = 5.78, SD = 1.18), t(264.79)  =  −18.98, MD = −3.24 [−3.58, −2.91], p < .001, d = −2.18. Similarly, mean response to the difference statement was significantly lower in the none condition (M = 2.58, SD = 1.8) than in the some condition (M = 5.82, SD = 1.17),

98

John Turri

Figure 5.1 Experiment 1. (A) Mean response to the basing and difference-making statements (within-subjects) across the two Status conditions (none, some) (between-subjects). Scales ran 1 (“strongly disagree”)—7 (“strongly agree”). Error bars show 95% bootstrapped confidence intervals. (B) Scatterplot showing the correlation between differencemaking and basing judgments. Points are jittered to avoid overplotting.

t(259.04) = −18.58, MD = −3.24 [−3.58, −2.89], p < .001, d = −2.18. The median response to both statements was “agree” (=6) in the some condition and “disagree” (=2) in the none condition; the modal response to both statements was “agree” in the some condition and “strongly disagree” (=1) in the none condition. Discussion This experiment began looking at the relationship between judgments about what makes a difference to whether an agent draws a conclusion, and the reasons an agent’s conclusion is based on. The two judgments were strongly correlated and statistically indistinguishable by their mean, median, and mode. This same pattern was robust across a range of stimuli and the demographic variables of biological age and sex. The results support the principal argument for a causal theory of the basing relation, which begins from the premise that reasons are difference-makers.

Experiment 2: Reason Comprehension and Attribution Method No research on the topic existed to inform an a priori power analysis regarding sample size. Because this experiment used less sensitive measures than experiment 1, I recruited more participants per condition, plus a few extra for the same reason described in experiment 1.

Commonsense Epistemology

99

Participants Participants’ mean age was 36.58 years (range = 19–69, sd = 11.85), and 55% (101 of 184) were female. Materials and Procedure Participants were randomly assigned to one of three conditions (Normal, False, Brain) in a mixed experimental design.1 Participants first read a brief scenario about an agent who draws a conclusion, then they rated whether they had enough information to understand the agent’s reasons, and finally they rated whether ten items (within-subjects) were among the agent’s reasons for drawing the conclusion. The scenarios were closely adapted from previous research on belief evaluation (Turri 2015). In all conditions, the agent, Victor, concludes that there is a fox nearby. The conditions differed in whether Victor is a normal human whose conclusion is true (Normal), a normal human whose conclusion is false (False), or a human brain-in-a-vat whose conclusion is false (Brain). (Normal/False) Victor is a healthy human adult sitting on his patio in a fine neighborhood. Victor is currently enjoying a variety of perfectly vivid sensory experiences, thanks to a team of scientists who helped save his life with a supercomputer that detected a heart condition. Victor was unaware that scientists could do that, just as he was unaware that he had a bad heart condition in the first place. Everything seems perfectly normal to him now. The scientists monitor him regularly.¶2 As Victor sits there on his patio, it seems like a reddish four-legged animal with pointy ears and a bushy tail is walking through a nearby flowerbed. Victor’s experiences seem entirely natural and normal. He concludes that it is a fox. [And/But] things [are/are not] exactly as they seem to Victor: as he sits there on his patio, he [is/is not] looking at a fox. (Brain) Victor is a healthy human brain sitting in a vat of fluid in a fine laboratory. Victor is currently enjoying a variety of perfectly vivid sensory experiences, thanks to a team of scientists creating them through a supercomputer that is hooked up to Victor. Victor was unaware that scientists could do that, just as he was unaware that his body died and he was put in a vat in the first place. Everything seems perfectly normal to him now. The scientists monitor him regularly.¶ As Victor sits there in his vat, it seems like a reddish four-legged animal with pointy ears and a bushy tail is walking through a nearby flowerbed. Victor’s experiences seem entirely natural and normal. He concludes that it is a fox. But things are not as they seem to Victor: as he sits there in his vat, he is not looking at a fox.

100

John Turri

Participants were then asked to rate the agreement or disagreement with this statement: Based on the scenario described above, I have enough information to understand Victor’s reason(s) for concluding that it is a fox. (comprehension) Responses were collected on a standard 7-point Likert scale, 1 (“strongly disagree”) — 7 (“strongly agree”), arranged left-to-right on the participant’s screen. Participants then went to a new screen and, while the story remained visible atop the screen, read these instructions, Victor concluded that it is a fox. What are Victor’s reasons for drawing that conclusion? For each item, please rate whether it is (“Yes”) or isn’t (“No”) one of Victor’s reasons for concluding that it is a fox. Below the instructions was a matrix table with ten items arranged vertically (order randomly rotated). To the right were three possible response options for each item, in the following order, left-to-right on the participant’s screen: “Yes,” “Unclear,” “No.” Participants selected their response by clicking a radio button. These were the ten items (all presented within-subjects): 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Sensory experiences (sense) The situation seems normal (normal) Background knowledge of what foxes looks like (background) The scientists’ activities (scientists) The supercomputer’s operations (computer) The physical animal itself (animal) Consciousness (conscious) Brain chemistry (brain) It seems like there is a fox (seems) The fact that there is a fox (fact)

Data Analysis One principal research question is whether participants would agree that they had enough information to comprehend Victor’s reasons, and whether they would do so at similar rates across conditions. A second question is whether assignment to condition would affect which reasons participants attributed to Victor. To address these questions, I conducted an analysis of variance on response to the comprehension statement, followed up by appropriate t-tests. I then conducted a linear mixed effects analysis on reason attributions from the list of items, followed up by appropriate analyses of variance.

Commonsense Epistemology

101

Results I conducted an analysis of variance on response to the comprehension statement, with assignment to condition as the independent variable. This revealed a significant difference across conditions, F(2, 181) = 9.65, p  < .001, 2 = 0.096. (See Figure 5.2.) Follow-up independent samples t-tests showed that mean response was higher in the Normal condition (M = 5.82, SD = 1.02) than in the False condition (M = 4.64, SD = 1.71), MD = 1.18 [0.68, 1.69], t(97.3) = 4.65, p < .001, d = 0.84. Mean response was higher in the Normal condition than in the Brain condition (M = 5.15, SD = 1.67), MD = 0.68 [0.18, 1.17], t(98.75) = 2.7, p = .008, d = 0.49. Mean response was nonsignificantly higher in the Brain condition than in the False condition, MD = 0.51 [−0.1, 1.11], t(119.93) = 1.66, p = 0.1, d = −0.3. Despite the differences, one sample t-tests showed that mean response was above the neutral midpoint (= 4) in all three conditions: Normal, t(61) = 14.12, p .999 < .001

Table 5.3 Experiment 1. F tests for the fixed effects and their order of elimination from the fully specified linear mixed effects model.  

Eliminated Sum Sq

NumDF DenDF F value

Status:Judgment Judgment sex age Status

1 2 3 4 0

1 1 1 1 1

0.002 0.198 0.380 0.735 156.865

305 305 305 305 305

p value

0.004 .947 0.495 .482 0.945 .332 1.830 .177 390.561 < .001

Table 5.4 Experiment 2. Likelihood ratio tests for the random effects and their order of elimination from the fully specified linear mixed effects model.  

Eliminated

logLik

AIC

LRT

Df

p value

  ID

  0

−1817.230 −1831.993

3702.460 3729.985

  29.526

  1

  < .001

Table 5.5 Experiment 2. F tests for the fixed effects and their order of elimination from the fully specified linear mixed effects model.  

Eliminated

Sum Sq

NumDF

DenDF

F value

p value

Sex Age Condition:Item

1 2 0

0.006 0.158 106.466

1 1 18

184 184 1656

0.014 0.397 14.814

.906 .530 < .001

6

Inference and the Basing Relation Keith Allen Korcz

One under-discussed issue in epistemology is the relationship between basing relations and inferences. I shall argue that errors about this relationship have led to some confusion about the nature of the basing relation. The aim of this chapter is to clarify some of that confusion. One consequence of this clarification will be a clarification of the basing of analytic truths. Another will be a proposed modification of the standard view that the action of making an inference is sufficient to establish a basing relation from a set of reasons to the inferred belief. I shall also explore some closely related issues regarding the relationship between so-called propositional justification and doxastic justification.

1 Some Issues Regarding Inference What is an inference? We’ll look at two issues: first whether inferences can occur only among beliefs, and second whether an inference must be stated as a premise within an argument. Regarding whether inferences can only occur among beliefs, the literature is not consistent. Sometimes inferences are understood to occur only among beliefs—one can only infer from premises one believes to a conclusion one thereby comes to believe (or, perhaps, already believed).1 Such a view seems implausible given the standard use of the term ‘inference’, as has been argued by Raymond Bradley and Norman Swartz.2 For example, when we make an inference of the reductio ad absurdum form,3 we may not take the conclusion to be true (e.g., where we are unsure as to which premise should be rejected). In this case, we infer to a conclusion we are unsure whether to believe. Similarly, when we end our reasoning at a self-contradiction, it is unlikely that we believe the explicit self-contradiction we have derived. Here, we disbelieve the conclusion inferred. Finally, if we include the self-contradiction among our premises, we may still infer our conclusion, though we do not believe the self-contradiction. Here, we infer from premises not all of which we believe.4 More generally, if we make an inference from premises we take to be false (e.g.,

114

Keith Allen Korcz

for purposes of exploring the implications of an hypothesis), there is no reason to expect that we would believe the conclusion thereby inferred. Bradley and Swartz claim that, “when [an] inference is made from a proposition which is believed to be true, then this inferred proposition is also believed to be true.”5 I have two objections to this view. First, suppose I casually read Plato’s paradoxical joke from Euthydemus6 and reason along as follows: (P1) I own a male dog with puppies. (P2) The dog is a father. (P3) The dog is mine. (C) The dog is my father. I may believe all the premises and make the inference (in a sense to be discussed shortly), yet reject the conclusion. Similarly for cases where the inference is impeccable, yet I am surprised by the conclusion and hesitate to believe it though I clearly recognize the validity of the inference and I continue to believe the premises.7 People’s beliefs are not always consistent. The surprise need not be a defeater or a sign of one (though, regarding this particular example, I have defeaters for believing that a dog is my father, the same may not be the case for other surprising inferences). In many cases, the unexpectedness of a conclusion may simply cause me to hesitate to form a belief, and, from there, distractions may prevent belief formation indefinitely. Second, I might also have a defeater for the conclusion, for example, other things I believe that indicate to me that the conclusion is not true. Yet I may continue to believe the premises and believe that the inference is correct, being unsure how to resolve the inconsistency posed by the defeater. Many arguments for skepticism about knowledge may have a similar paradoxical nature.8 Inference seems to be not as much about the relata as the relation. I can engage in the same sort of action (e.g., draw an inference that is an instance of modus ponens) whether I believe the premises or not, or even when I am manipulating variables and do not know the propositions that might be substituted for them. When I fail to believe the premises, I will typically not believe the conclusion as the result of the inference. Believing a conclusion seems to be a frequent but not necessary result of the action of making an inference. It isn’t too far a step, then, to hold that inference need not result in a believed conclusion. A second issue is whether, to function, inferences must be explicitly stated within the premises of the argument. For example, if I believe that (1) either Nate will bring the bourbon or Cathie will bring the bourbon, and I believe that (2) it is not the case that Cathie will bring the bourbon, must I also believe that (3) if (1) and (2) are true, then (C) is true in order to infer that (C) Nate will bring the bourbon? One could believe

Inference and the Basing Relation

115

that an inference is valid, but must such a belief be added to the premises of an argument in order to infer its conclusion? I don’t think so. As Lewis Carroll has pointed out, reasons for such a requirement are likely to result in a vicious infinite regress.9 One way to end such a regress is to take an inference to be an action, and this seems to be the standard view. One might often consciously believe that one has made a particular inference, but one need not do so to have made the inference. We can then distinguish inferences from argument forms, such as affirming the consequent, affirming the antecedent, etc., whether valid or invalid (restricting myself for the moment to deductive reasoning with the assumption that similar points will apply to inductive reasoning, mutatis mutandis) and that, we could suppose, are not mental entities. A valid inference is then an instance of an argument form,10 and similarly for other sorts of inferences. A deductive inference might then be, roughly, understood as follows: An action, being an instance of an argument form, that when performed results in the impression11 that if a given set of premise(s) is true, then a given conclusion is true. This definition at least avoids the problems with reductios and paradoxes previously discussed. Note that an inference will often, but not always, cause one to believe the conclusion one has inferred. The impression mentioned in the definition could take the form of a sophisticated belief about logical connectives and such, but it need not. There is an issue here related to the well-known Problem of Distant Reasons,12 also referred to as the Problem of Forgotten Evidence.13 Goldman has presented the problem as a problem for what he calls Weak Internalism, the view that only facts concerning what conscious and/or stored mental states an agent is in at time t are justifiers of the agent’s beliefs at t. Goldman describes the problem: Many justified beliefs are ones for which an agent once had adequate evidence that she subsequently forgot. At the time of epistemic appraisal, she no longer possesses adequate evidence that is retrievable from memory. Interestingly, some psychologists have suggested that forgetting may, at least typically, not involve loss of beliefs, but merely difficulty accessing them from long-term memory.14 Providing the proper help may allow someone to make the recollection, or at least provide evidence that the memory trace remains. If this is true, and if one is concerned merely with developing an internalist account of human knowledge, perhaps this would blunt the force of the objection. It depends on how flexible an internalist is with regard to the accessibility of beliefs. One could likely

116

Keith Allen Korcz

still rig up examples of missing evidence, but they might not arise in the usual course of human knowing. Given that inferences are actions that (at least typically) occur and are over fairly quickly, how long does the epistemic effect (e.g., of justifying the belief) last? I have no idea if memories of inferences often make it to long-term memory. But certainly inferences are a key element of epistemic justification. Here we may have a problem similar to that of the Problem of Forgotten Evidence, except regarding inferences. We might call it the Problem of Forgotten Inferences. A similar problem might also occur regarding forgotten perceptual states that function(ed) as reasons for a belief: the Problem of Forgotten Perceptions. The standard sort of reply to the Problem of Forgotten Evidence is to point to reasons a person currently does have and can easily access, such as the justified belief that beliefs recalled in a particular way tend to be true, even if the original reasons for the belief are not easily accessible. This sort of reply could also easily be extended to forgotten inferences and forgotten perceptions, and it seems to me, anyway, that it will ultimately be adequate.

2 Turri’s Challenge: Basis and Basing John Turri15 and others16 have attributed to several philosophers a view that Turri summarizes as follows: (Basis) IF (i) p is propositionally justified for S in virtue of S’s having reason(s) R, and (ii) S believes p on the basis of R, THEN S’s belief that p is doxastically justified. Turri then argues that Basis is false on the basis of some counter-examples he gives. Here is one of the counter-examples:17 Consider two . . . jurors, Miss Proper and Miss Improper, sitting in judgment of Mr. Mansour. Each paid close attention throughout the trial. As a result, each knows the following things: (P1) Mansour had a motive to kill the victim. (P2) Mansour had previously threatened to kill the victim. (P3) Multiple eyewitnesses place Mansour at the crime scene. (P4) Mansour’s fingerprints were all over the murder weapon. is propositionally justified for both jurors because each knows (P1—P4). As it happens, each comes to believe as the result of an episode of explicit, conscious reasoning that features (P1—P4) essentially. Miss Proper reasons like so:

Inference and the Basing Relation

117

(Proper Reasoning) (P1—P4) make it overwhelmingly likely that Mansour is guilty. (P1—P4) are true. Therefore, Mansour is guilty. Miss Improper, by contrast, reasons like this: (Improper Reasoning) The tea leaves say that (P1—P4) make it overwhelmingly likely that Mansour is guilty. (P1—P4) are true. Therefore, Mansour is guilty. Each juror satisfies conditions (i) and (ii), yet only Miss Proper’s belief that Mansour is guilty is doxastically justified. Only Miss Proper is justified in believing as she does that he is guilty. Miss Improper’s belief that Mansour is guilty is based on the reasons that propositionally justify for her, but only in light of the tea-leaf reading, which spoils things. Her belief is improperly based on (P1—P4), whereas Miss Proper’s is properly based. The same example shows that Basis+ fails.18 However, I think that a propositional justification so understood is not always a reason, and for precisely the reason Turri gives: no inference need be specified. If propositional justifications need not include any sort of inference, then Basis appears mistaken in assuming that propositional justifications are reasons. I take it that for something to be a good reason for believing p, it must indicate the truth of p. Regarding inferential justification, possessing some set of beliefs and believing another, nonequivalent, proposition p, without any indication of an inference to p, is not a situation where the set of propositions indicate the truth of p. For example, believing that q is no reason to believe some non-equivalent proposition that p unless an inference is somehow indicated or involved, for example, a belief or realization that q entails p. What is it for an inference to be indicated or involved in a propositional justification? Having a meta-belief to the effect that a reason is a good reason to hold a belief would be one way for a propositional justification to indicate the relevant inference, but it cannot be the only way on pain of infinite regress and issues with epistemically unsophisticated knowers. Another would be a de re realization “if this [reason], then that [conclusion].” Again, this is not necessary for there to be a propositional justification, but it is sometimes there. When it is there, then (all else being equal) the propositional justification is a reason. But when it is absent, the propositional justification is not a reason at all. At best, it is a possible reason.19 Such a conception of propositional justification fits best with discussions over whether epistemic reasons are possible, as in many discussions of skepticism. One possible objection is that a propositional justification’s failing to specify an inference might make it a bad reason, but it is still a reason. In

118

Keith Allen Korcz

response, I am not sure how exactly to draw the distinction between bad reasons and no reasons. But it seems like a good place to start would be to require that propositional reasons (good or bad) involve an inference (good or bad), but non-reasons do not. As Turri notes, requiring that one have a belief regarding the relevant inference cannot plausibly be required for propositional justification: Must the jurors also believe that (P1—P4) constitute good evidence to conclude that Mansour is guilty in order for their knowledge of (P1—P4) to propositionally justify that conclusion for them? No, because that would impose an unacceptably stringent condition on propositional justification. Human children, who may not even possess the concept of evidence, are propositionally justified in believing many things. Lacking the concept of evidence, they cannot so much as entertain the thought that X is evidence for Y, much less believe such a thing.20 For those propositional justifications that, for example, include such a belief indicating the relevant logical inference (e.g., the reason is a fully expressed argument), Turri’s counter-examples do not work. In the Mansour example, if there were such a belief, the difference between Miss Proper and Miss Improper would be evident in their respective propositional justifications. On the other hand, for such propositional justifications that make no use of an inference, if I am correct, those propositional justifications are not reasons. Where one has a propositional justification, but substitutes the inference indicated in that propositional justification for some other inference, that propositional justification simply isn’t being used. Suppose that S has the belief that q, a belief that q entails p, and a belief that modus ponens is the right way to infer that p given these other beliefs. Let’s suppose that those three beliefs constitute a propositional justification for S. But then S infers that p utilizing a different inference rule, say that from the two beliefs infer r and from r always infer q. In that case, S is simply not forming a belief using the given propositional justification. Different inferences may distinguish different propositional justifications. So, if, as Basis says, p is doxastically justified for S in virtue of S’s having reason(s) R, then that propositional justification must specify the relevant inference. Otherwise, p is not had in virtue of reasons R.21 It is very confusing to call propositional justifications justifications if, in some cases, they cannot function as reasons as I have explained them here, but that seems to be a common usage.

3 Turri’s Solution Turri offers his own solution to the counter-example of the jurors and Miss Proper he presents and suggests that it may offer an important insight into the nature of propositional justification. The suggestion is that

Inference and the Basing Relation

119

Judgments about propositional and doxastic justification are moderately idealized, in that they abstract away, to a greater or lesser degree, from the abilities and powers of the specific agent in question, to the abilities and powers manifested in a typical performance by a competent member of the agent’s kind.22 Divorcing the concept of propositional justification from individual knowers to kinds of knowers as Turri does is a fairly radical step, but one that is unnecessary if the previous two alternative suggestions are correct. To resolve the problem of propositional justifications lacking inferences, Turri proposes the following account of the relationship between propositional and doxastic justification: (PJ) Necessarily, for all S, p, and t, if p is propositionally justified for S at t, then p is propositionally justified for S at t BECAUSE S currently possesses at least one means of coming to believe p such that, were S to believe p in one of those ways, S’s belief would thereby be doxastically justified.23 Turri then bravely fights an uphill battle to avoid several potential counterexamples to this account and to clarify what is meant by possessing a means of coming to believe. The resulting account will be quite complex.24 Turri might find a way around the various counter-examples he discusses, but the effort to clarify the idea of possessing a means of coming to believe seems to me to be more problematic, as it is what pushes Turri to suggest that propositional justification applies to kinds of knowers rather than individual knowers. The difficulties with getting precise about the counterfactuals and possibilities regarding when someone has a capability in some relevant sense, and when they do not, are well known, so I will not repeat them here. The solution I have suggested, that of holding that some propositional justifications are not reasons, has the advantage of avoiding at least some of these difficulties and complexities.

4 Are Basing Relations and Inferences Identical? There is also an alternative way of attempting to solve the problem posed by Turri’s counter-examples. The idea is to hold that in some cases basing relations are themselves inferences. Alvin Goldman appears to suggest, but does not commit to, the idea: I am inclined to say that inference is a causal process, that is, when someone bases his belief of one proposition on his belief of a set of other propositions, then his belief of the latter propositions can be considered a cause of his belief of the former proposition. But I do not wish to rest my thesis on this claim.25

120

Keith Allen Korcz

Accepting such an account, one could draw a distinction between properly and improperly based beliefs, the former affording justification by means of a good inference (ceteris paribus) and the latter supplying only a bad inference. Then, only propositional justifications that are properly based can amount to doxastic justifications.26 However, if I am correct in arguing that inferences need not result in beliefs, and it is the case that basing relations always result in beliefs, then the two will not necessarily coincide. I take it we can have a basing relation without an inference (e.g., if we believe on the basis of a perceptual state or some such). Moreover, it seems we can have inferences without basing relations, as discussed earlier. It may, however, be the case that every inferentially justified belief involves an inference coinciding with a basing relation.

5 Doxastic Theories of the Basing Relation Goldman seems to have had in mind a causal theory of the basing relation, according to which, very roughly, a belief is based on a reason if the reason causes the belief in an appropriate way. But what about doxastic theories of the basing relation? There are two main types of these: one holds, very roughly, that S’s belief that p is based on S’s reason R if S has a meta-belief B to the effect that reason R is a good reason to hold the belief that p—thus, such meta-beliefs are, if of the appropriate kind, sufficient, but not necessary, to establish basing relations.27 The other type holds that such meta beliefs are both necessary and sufficient for a basing relation. I’ll just discuss the former view, that certain meta-beliefs are sufficient to establish a basing relation. A point of clarification: at first glance, what I have described as inferences, on the one hand, and doxastic basing relations, on the other, may appear identical. However, for a doxastic theory to be plausible it clearly must require that the person in question hold the belief to be so based. This distinguishes even doxastic basing relations from inferences, at least conceptually, and the distinction is exactly similar to that for causal theories of the basing relation. Turri offers a purported counter-example to such doxastic theories of the basing relation, but it is not clear to me that it works. The example and edited portions of his discussion are as follows:28 (EXHAUSTED) Martin believes that Mars contains significant amounts of water buried just below its surface (Q). He judges that this is good evidence to believe that life exists elsewhere in the universe (P). Martin also is certain that the conditions for life are overwhelmingly abundant throughout the universe (S). He judges that this too is good evidence to believe that life exists elsewhere in the universe. But Martin is utterly exhausted and in despair from

Inference and the Basing Relation

121

several grueling and fruitless months on the academic job market, which understandably and predictably impairs his cognitive functioning, especially at the present moment. He consequently neglects his evidential judgment about the relevance of subterranean Martian water, and bases his belief that life exists elsewhere solely on his belief that the conditions for life are abundant throughout the universe. Turri continues: If this is a possible case, then the doxastic theory is false. It would appear to be a case in which one has an appropriate meta-belief to the effect that Q is good evidence for P, but does not base his belief that P on Q. And it certainly seems possible. The job market may be bad enough to make Martin slightly irrational. But it’s not bad enough to make his situation impossible. What of Martin’s “neglected” evidential judgment? A doxastic theorist might respond as follows. If by ‘neglect’ I mean ‘forgot’, then the case poses no threat to the sufficiency of the doxastic theorist’s condition. If by ‘neglect’ I mean ‘reject’, then again the case poses no threat. In response, by ‘neglect’ I mean neither ‘forgot’ nor ‘reject’. I simply mean that Martin is unaffected by this evidential belief, in the same way that Michael Stocker’s jaded politician is unaffected by some of his moral beliefs. . . . There once was a young politician who cared about the plight of suffering people worldwide. . . . But he became jaded as he aged. He no longer cared about anyone outside his circle of friends and family. He still believed that it would be a very good thing to help the downtrodden, and knew there was much he could do to promote that goal. But he was no longer the least bit motivated to do so. . . . In broad outline, Martin’s case is to epistemic psychology as Stocker’s is to moral psychology. Both cases involve the failure of an evaluative belief to play its typical role. And just as the lack of motivation in the politician’s case needn’t indicate loss of the relevant moral belief, the lack of basing in Martin’s case needn’t indicate loss of the relevant evidential belief. It’s hard to see what Martin lacks the motivation to do, since there is nothing left for him to do, according to many doxastic theories. Once Martin has formed the relevant meta-belief regarding his reason and the belief it is a reason for, and he retains the belief it is a reason for, there is nothing else left for Martin to do: the belief is based on his reason, according to the basic doxastic theory I outlined earlier. Turri suggests that Martin is “unaffected by this evidential belief.” But all that is required by such doxastic theories is that Martin have the evidential belief (and that the belief be thereby based on a reason); it need not additionally affect him

122

Keith Allen Korcz

in any other way. The evidential belief (or, as I referred to it earlier, the meta-belief) is the effect. Turri’s example of the exhausted Martin makes the most sense if Turri is thinking that the meta-belief that establishes a basing relation must also cause a person to hold the belief for which the meta-belief establishes the person’s reasons. But, at least typically, this is not what doxastic theories require and, in fact, sometimes, the whole point is to avoid such a condition (e.g., in Korcz’s theory, a mixture of causal and doxastic theories,29 where the relevant meta-belief MB establishes a basing relation between reasons R and the belief that p even though there is no causal relation from either reasons R to the belief that p nor MB to the belief that p).

6 Are Analytic Truths Based on Anything? An interesting question related to the issue of inference and basing is the question of how to understand the basing of analytic truths. I will offer an account of the basing of analytic truths, and then briefly discuss an argument that it makes the justification of analytic truths inferential. The standard, if famously controversial,30 slogan about analytic truths is that they are true by definition. Setting aside the controversies, this suggests a straightforward account of what beliefs that are instances of analytic truths (when known by virtue of their analyticity and not, for example, the testimony of someone else) are based on: an understanding of the meaning of the analytic truth, presumably consisting of an understanding of the relevant concepts along with an understanding of how those concepts are organized to express the analytic truth in question. Thus, my belief that no bachelor is married is based on my understanding of the meaning of the word “no,” the word “bachelor,” the word “is,” and the word “married,” along with my understanding of the relevant grammar. It is very difficult to see what else could cause me to believe that an analytic truth is true merely by understanding it if it is not the components of my understanding it. Such an account of analytic truths has some virtues. It explains how analytic truths are self-verifying while, at the same time, not being known by people who hear one but do not understand what it means. Second, content, but not propositional content, is connected to knowing the analytic truth. This is analogous to beliefs based on perceptual states and such: it is presumably in virtue of their content that they justify the beliefs immediately based on them, but this is not (I assume) propositional content. One virtue of the usual sort of causal analysis of the basing relation is that a common account of basing is provided for both inferential and non-inferential justification, and this account of the basis of beliefs that are instances of analytic truths preserves that virtue. I have assumed that applications of grammatical knowledge are skillbased, and that skill-knowledge (sometimes called “knowledge-how”) is

Inference and the Basing Relation

123

non-propositional.31 However, some have argued that skill knowledge is in fact propositional. This would make knowledge of analytic truths in part inferential as well, on the account I have given. Since this would undermine the usual role of analytic truths, this seems to me to be a good reason to deny that skill knowledge is propositional.32

Notes 1. David H. Sanford, If P, Then Q: Conditionals and the Foundations of Reasoning (London, UK: Routledge, 1992), p. 129. 2. Raymond Bradley and Norman Swartz, Possible Worlds: An Introduction to Logic and Its Philosophy (Indianapolis, IN: Hackett, 1979), p. 193. 3. Of course, similar examples can be rigged up without appeal to the use of this particular argument form. 4. As a reviewer pointed out, we could believe the conditional from assumption to contradiction. However, we do not have to, and this is all I need for my point here. 5. Bradley and Swartz, p. 202. 6. Plato, “Euthydemus,” The Collected Dialogues of Plato, ed. by Edith Hamilton and Huntington Cairns (Princeton, NJ; Princeton University Press, 1963), Stephanus 298e, p. 412. 7. This is the reason why closure principles for knowledge are obviously false. Of course, the real issue for closure is with propositional justification, not knowledge. 8. Thank you to a reviewer who suggested I point this out. 9. Lewis Carroll, “What the Tortoise said to Achilles,” Mind, New Series, Vol. 4, No. 14, April 1895, pp. 278–280. 10. It seems a lot more natural to talk of following an inference rule here, but this seems to require knowledge of the rule, something which may be beyond many knowers. It seems sufficient to make an inference and thereby acquire knowledge that the knower realize that the conclusion follows from the premises in this instance, even without knowledge of the general rule. 11. This impression could be a belief, but need not be, accommodating those without sophisticated logical concepts. A simple de re “this because of that” should suffice. The realization need not be conscious, but would ordinarily, at least, be accessible upon reflection. The impression need not be accurate— that is, one could make a bad inference. 12. Marshall Swain, Reasons and Knowledge (Ithaca, NY: Cornell University Press, 1981), pp. 87–88. 13. Alvin Goldman, “Internalism Exposed,” Epistemology: Internalism and Externalism, ed. by Hilary Kornblith (Malden, MA: Blackwell, 2001), pp. 214–216. 14. David A. Lieberman, Human Learning and Memory (New York, NY: Cambridge University Press, 2012), pp. 409–410. 15. John Turri, “On the Relationship between Propositional and Doxastic Justification,” Philosophy and Phenomenological Research, Vol. 80, No. 2, March 2010, pp. 312–326. 16. For example, Paul Silva, Jr., “On Doxastic Justification and Properly Basing One’s Beliefs,” Erkenntnis, Vol. 80, 2015, pp. 945–947. 17. Turri, 2010, pp. 315–316. I have removed the footnotes. 18. Basis+ is a slightly modified version of Basis. The difference is not relevant to my point here.

124

Keith Allen Korcz

19. Similarly for defeaters. Possessing a defeater for a justification undermines the potential of that justification to justify. Where the defeater is propositional, for example, it is only if one infers from the defeater to the conclusion that, say, a premise of your justification is false does it become your reason to reject your justification, in the way the terminology is being used here. 20. Turri, 2010, p. 316, footnote 10. 21. It may be relevant in some cases that most of the quotes Turri cites (on pp. 313–314) as endorsements of Basis are phrased in terms of reasons. 22. Turri, 2010, p. 324. 23. Turri, 2010, p. 320. Turri adds a couple of explanatory footnotes here, but I don’t think they effect my point so I shall not repeat them. 24. Silva, p. 950, sketches out his own interpretation of the various conditions that appear to be required. 25. Alvin Goldman, “A Causal Theory of Knowing,” Essays on Knowledge and Justification, ed. by George S. Pappas and Marshall Swain (Ithaca, NY: Cornell University Press, 1978), p.  73. A similar idea seems to be what Gilbert Harman proposed in his, “How Belief Is Based on Inference,” Journal of Philosophy, Vol. 61, No. 12, June 11, 1964, pp.  354–355. More recently, Ram Neta discusses the notion that an inference is sufficient to establish a basing relation (and, perhaps according to some, both necessary and sufficient) in his, “The Basing Relation,” The Routledge Companion to Epistemology, ed. by Sven Bernecker and Duncan Pritchard (London, UK: Routledge, 2011), p. 111. Also, Paul Boghossian takes it for granted that inference and basing are identical in, “What Is Inference?,” Philosophical Studies, Vol. 169, 2014, p. 8. 26. This is very similar—perhaps identical, but I am not certain of that—to a proposal made by Paul Silva, Jr. in his “On Doxastic Justification and Properly Basing One’s Beliefs,” Erkenntnis, Vol. 80, 2015, pp. 953–954. 27. For example, Keith Korcz, “The Causal-Doxastic Theory of the Basing Relation,” Canadian Journal of Philosophy, Vol. 30, No. 4, December 2000, pp. 547–548. This chapter lays out a number of additional conditions that must be met for it to be plausible that such a belief establish a basing relation, but I have not included them here because they would just complicate the discussion. 28. John Turri, “Believing for a Reason,” Erkenntnis, Vol. 74, No. 3, May 2011, pp.  386–387. Here, I only discuss what amounts to an aside in the paper. Turri goes on in the paper to develop an original and interesting account of the basing relation intended to avoid deviant causal chains. 29. Keith Allen Korcz, “The Causal-Doxastic Theory of the Basing Relation,” Canadian Journal of Philosophy, Vol. 30, No. 4, December 2000, pp. 525–550. 30. The controversy, of course, is due to W. V. O. Quine, “Two Dogmas of Empiricism,” From a Logical Point of View (New York, NY: Harper & Row, 1961), ch. 2. I make no effort to address these issues in this paper. 31. The classic discussion of this issue is Gilbert Ryle, The Concept of Mind (New York, NY: Barnes & Noble, 1949), ch. 2. 32. I would like to thank Micah Dugas, Jesse Saloom, and the editors and reviewers for their helpful comments on previous drafts of this paper.

7

The Superstitious Lawyer’s Inference Patrick Bondy and J. Adam Carter

I Introduction The epistemic basing relation, which is the relation that obtains between a belief and the reason(s) for which it is held, is of central importance in epistemology. This is because epistemologists care about epistemic justification, and the epistemic basing relation is often taken to be necessary for distinguishing between the two principal kinds of epistemic justification: propositional and doxastic. The core idea is roughly as follows: a proposition p is propositionally justified for a subject S just in case S possesses good epistemic reasons for believing p; whereas S’s belief that p is doxastically justified (i.e. S’s belief is held in an epistemically permissible fashion) if and only if S believes p on the basis of S’s good epistemic reasons for believing p, viz., those reasons that propositionally justify p for S. Many epistemologists are happy to accept that doxastic and propositional justification are ‘connected’ in this way by the epistemic basing relation.1 However, it is highly contentious how to spell out the conditions that must be satisfied for a belief to count as being based on a reason. Although it is not a perfect match, externalists about epistemic justification have generally aligned themselves with some form of a causal account of the basing relation. To a first approximation, this account maintains that for S, p, reason R, S’s belief that p is based on R iff S’s belief that p is (non-deviantly) caused by R,2 where the matter of how to spell out ‘non-deviance’ has divided proponents of the causal account.3 By contrast, doxastic theories of the basing relation—accounts that are often found to be attractive by epistemic internalists—take basing to be fixed (in short) principally by beliefs rather than causes; that is, typical doxastic accounts hold that for S, p, reason R, S’s belief that p is based on R only if S has a meta-belief to the effect that R is a good reason to believe p (e.g., Audi 1982). Interestingly, the most famous and controversial counterexample to any version of the causal account was the very first counterexample ever raised to it: Keith Lehrer’s (1971) case of the superstitious lawyer, a case initially posed to Gilbert Harman’s (1970) early causal account.

126

Patrick Bondy and J. Adam Carter

The matter of whether Lehrer’s counterexample worked against causal accounts was a lively topic in the 1970s and 1980s,4 though little consensus was reached—perhaps at least in part because of the sheer complexity of the case. And then, for whatever reason, discussion of the case waned, while the causal account has since gained popularity as the default view. We take exception to this orthodoxy, which we think is largely unearned. As we see it, Lehrer’s counterexample is far worse news for causal accounts than the contemporary thinking about the causal account would suggest. Yet this is for reasons that have thus far not been fully appreciated, including by proponents (early and present) of doxastic accounts. The time, we believe, is ripe to revisit Lehrer’s famous example with a critical eye, both to the example and to the causal theory it is meant to challenge. In doing so, we will show why it is—as Lehrer initially thought—a case where one bases a belief on the basis of a reason that is not a cause of the belief. More specifically, as we shall argue, the case is best understood as a case of normal inference from premises to a conclusion, one that is sufficient both for the lawyer to base his belief in the conclusion at least in part on the good evidence he has, and, plausibly, for the lawyer’s belief to be doxastically justified by that good evidence. If this diagnosis is correct, then the problem that the case poses for causal accounts can be expressed much more simply, and we think forcefully, than previously thought. Here is the plan. In §2, we discuss basing, illustrating how causal and doxastic theories handle a wide range of ordinary cases of basing and non-basing correctly. In §3, we discuss inference, in particular arguing in support of a type of doxastic ‘taking’ condition on inference. In §4, we explain the central features of the original superstitious lawyer case, and we provide a new, much simpler case to illustrate the point of the original one. In both the original case as well as in our new one, the lawyer bases his belief on the evidence, and the evidence doxastically justifies his belief, even though the evidence is causally inert. Interestingly, however, because the lawyer’s belief is unsafe, it might very well be that critics were right to say that the superstitious lawyer lacks knowledge.

II Basing Although giving accounts of inference and the basing relation has proven to be a difficult task, identifying clear cases of basing and non-basing is comparatively easy. To orient ourselves in reflecting on the basing relation, then, it’s useful to begin with some clear cases of beliefs held on the basis or not held on the basis of reasons. Identifying features present in the cases of basing and absent in the cases of non-basing will help us in thinking about more difficult cases, such as the case of the superstitious lawyer.

The Superstitious Lawyer’s Inference

127

Case 1: No Reason, No Basing Sam believes that it will rain today. Sam has not seen a weather forecast, no one has told her anything about the weather, and she has not even looked outside. Sam simply woke up feeling sure that it would rain. In No Reason, No Basing, there just is no reason on the basis of which Sam holds her belief.5 Plausibly, Sam’s belief is doxastically unjustified,6 but the point here isn’t about justification: it’s about whether her belief is held on the basis of a reason. There is nothing that Sam recognizes or would recognize on reflection as her reason for believing that it will rain. Intuitively, it appears that Sam doesn’t hold her belief on the basis of a reason.7 Case 2: Good Reason, No Basing Sam believes that it will rain today. Sam saw a weather forecast from a reliable source indicating a 95% chance of rain today. But Sam places no trust in weather reports, and the weather report that she saw plays neither an initiating nor a sustaining causal role with respect to her belief. She simply woke up feeling sure that it would rain. This case is just as in No Reason, No Basing, except that this time there is a fact or feature of the case that could act as Sam’s motivating reason for belief, and if she were to hold her belief on the basis of the weather report, her belief could be doxastically justified. But she does not see the weather report as providing good reasons for belief, and it is causally inert with respect to Sam’s being in the mental state of believing that it will rain. Her belief, just as in No Reason, No Basing, is based on nothing at all. Case 3: Bad Basis Sam believes that it will rain today. She believes that because her weekly horoscope from six days ago indicated that she would see rain this week, and she has not seen rain yet this week. Unlike in the previous cases, in Bad Basis there is something that Sam regards as a good reason for thinking that it will rain today: her weekly horoscope. Horoscopes do not generally provide good reasons for believing their predictions, so Sam’s reason is not a good reason her belief, and all else equal her belief is doxastically unjustified. Still, in this case there is clearly a (motivating) reason on which her belief is based.

128

Patrick Bondy and J. Adam Carter Case 4: Well-Founded Sam believes that it will rain today. She believes that because she saw a weather forecast from a reliable source, which indicated a 95% probability of rain. Sam trusts weather reports in general, and believes that this report provides a good reason for thinking that there is a high probability of rain today.

As in Bad Basis, here Sam has a reason which she sees as making it the case that she should believe that it will rain today. In this case, the reason is a good one, and all else equal, it doxastically justifies her belief. Nonetheless, there is a clear sense in which the relation between Sam’s belief that it will rain today and the good reason that she possesses in WellFounded is the same kind of relation which obtains between Sam’s belief that it will rain today and the bad reason she possesses in Bad Basis.8 That relation is the basing relation. By contrast, in the first two cases, there is no basing relation instantiated. The intuitive responses to these four cases are compatible with a range of views about the nature of the basing relation. For instance, they are compatible with a purely causal account according to which a belief B is based on a reason R iff R non-deviantly causes B.9 In No Reason, No Basing, and in Good Reason, No Basing, there is no reason which causes Sam’s belief, so the causal account of basing yields the result that Sam does not hold her belief on the basis of a reason in these cases. That is the intuitively correct result. In Bad Basis and Well-Founded, by contrast, there is a reason which appears to non-deviantly cause Sam’s belief, and so the causal account yields the result that Sam’s belief is held on the basis of these reasons. Again, that is the intuitively correct result. But these results do not by themselves settle the issue in favour of causal accounts of the basing relation. For example, we can correctly categorize these cases if we opt for a doxastic account of basing according to which a belief B is based on a reason R iff the subject S has an appropriate meta-belief, to the effect that an available reason is a good reason for belief. This sort of doxastic account also correctly categorizes the cases we’ve seen so far: in the first two cases, Sam has no such meta-belief, so the doxastic account entails that her belief is not based on a reason, while in the latter two cases, plausibly, she does have such a belief. So, in a wide range of clear cases where subjects hold beliefs on the basis of reasons, the reasons are causally efficacious in sustaining the subject’s belief, and the subject believes that the reason is a good one for holding the target belief. And in a range of clear cases where beliefs are not held on the basis of reasons, there is no reason which the subject takes to be a good reason for the belief, and there is no reason that is nondeviantly effective in causing or sustaining the belief.

The Superstitious Lawyer’s Inference

129

That there is a wide range of cases of basing in which (i) subjects take there to be good reasons for their beliefs and (ii) those reasons causally support the target beliefs should be unsurprising. After all, we are creatures capable of rationally regulating our beliefs in accord with the reasons that appear to us to be good ones, and when things are going well, we do causally regulate our beliefs in that way. We should therefore expect that in normal cases, where our cognitive faculties are functioning normally, our recognition of a reason as a good reason for belief will prompt us to hold the target belief. And in such cases, when a reason causes us to hold a target belief, it does so only because we take the reason to recommend holding the target belief. So it is unsurprising that causal and doxastic accounts yield the same verdicts in ordinary cases. Each type of account identifies a relation that clearly appears to obtain in ordinary cases where beliefs are held on the basis of reasons, and that fails to obtain in clear cases of non-basing. To help us decide whether the basing relation is really a causal relation or a doxastic relation (or both,10 or neither),11 it will help to consider cases where causal and non-causal accounts of the basing relation diverge, such as superstitious-lawyer types of cases. In these cases, the intuitive response regarding whether the belief is held on the basis of a particular reason is not so clear—or at least, intuitions have historically been divided. In the superstitious-lawyer case, a lawyer (in short) infers his client’s innocence from some good evidence; the evidence is causally inert with respect to the lawyer’s belief in its conclusion; but he still draws the inference.12 Lehrer used the example to argue that the lawyer can have knowledge of the conclusion of the inference, crucially, on the basis of the causally inert inference. If Lehrer is right about that, then—contra the causal account—beliefs can be held on the basis of reasons which have no causal efficacy. In Section IV we will argue that because the lawyer draws the relevant inference, we should see him as holding the belief in his client’s innocence on the basis of his appraisal of the evidence. In other words, we argue, inference from believed premises to a believed conclusion is sufficient for basing belief in the conclusion on belief in the premises. That is just what we see in the case of the superstitious lawyer. Before turning to the full case of the superstitious lawyer, it will therefore be profitable first to clarify what we mean by ‘inference’, and why we think inference is sufficient for basing.

III Inference The term ‘inference’ has several related uses. Sometimes we talk about inferences as conscious mental events; sometimes we talk about implicit inferences that we can be said to draw, when we unconsciously or

130

Patrick Bondy and J. Adam Carter

subconsciously base beliefs on perceptual input. Sometimes we talk about inferences as actual events or processes that have taken place in a person’s mind, while other times we are interested in potential inferences that a person might draw; the former are concrete events, while the latter are abstract objects, consisting of propositional contents and inferencewarrants which permit the acceptance of some propositional contents on the basis of some other accepted propositions. The relevant sense of ‘inference’ here is that of a concrete mental act, involving the consideration of some premises and a conclusion, where the subject takes the premises to support the conclusion.13 The key question in articulating an account of inference is how that ‘taking’ condition should be understood: what does it mean to say that a subject takes a set of premises to support a conclusion? For example, does it mean that the subject believes that the premises support the conclusion? Or perhaps it means that the subject is disposed to follow a rule according to which one ought to form a belief in the conclusion, given that one accepts the premises? One way to approach the taking condition is to consider cases where a subject S believes that the premises do not support the conclusion. In such cases, clearly, S does not take the premises to support the conclusion. Similarly, we can consider cases where a subject has no opinion on whether some premises support a conclusion. Again, cases like this are likely to strike us as clearly cases where the subject does not take the premises to support the conclusion. And so the most natural interpretation of the taking condition is in doxastic terms: Taking (Doxastic): S takes a premise P to support a conclusion C iff S believes that P supports C. Because this is the most natural interpretation of the taking condition, it is the one philosophers tend to consider first, before moving on to other interpretations when problems come up for the doxastic construal.14 In our view, the objections that have been raised against the doxastic construal are far from decisive. They do show that Taking (Doxastic) is too crude as it stands, but what the objections point to is the need for a more refined understanding of the doxastic construal. And, with the refined doxastic construal in hand, we will be in a position to see that superstitious-lawyerstyle cases are genuine cases of inferential basing of beliefs on reasons. The two most important objections to Taking (Doxastic), in our view, are the over-intellectualization objection and the causal deviance objection. We will address these in turn. III (i) Over-Intellectualization Paul Boghossian (2014) writes:

The Superstitious Lawyer’s Inference

131

A child, we are inclined to think, can reason. Luke and Drew are playing hide-and-seek. Seeing Drew’s bicycle leaning against the tree, Luke thinks: “If he were hiding behind that tree, he would not have left his bicycle there. So, he must be behind the hedge.” That looks like reasoning. But do children have meta-beliefs about the relations between their premise judgments and their conclusions? Do children have the concepts of premises and conclusions? Do they have the normative concept of one belief justifying another? (2014, pp. 6–7) The worry is that (surely) children engage in reasoning. But they are also (surely) too unsophisticated to have the concept of a premise, or a reason. So they cannot have the meta-belief required to satisfy Taking (Doxastic). Examples like this one, we think, give children too little credit. From a very young age, children are able to ask and answer ‘why’ questions; and to ask why one ought to hold a belief, or why (for example) one must go to school, is to ask for normative reasons. Young children—and probably most adults—no doubt have never heard the phrase ‘normative reason’, or even ‘premise’ and ‘conclusion’, but they do know what it is to hold beliefs or perform actions for good reasons, or for bad reasons. That is all the conceptual apparatus required to satisfy Taking (Doxastic). And, in Boghossian’s example, it looks as though Luke does have the required conceptual sophistication to form the necessary meta-belief. It is a dispositional belief, to be sure; Luke likely does not consciously dwell on it. But it is a belief of his nonetheless. So that is not a counterexample to Taking (Doxastic). On the other hand, if we turn our attention to children who are so young and unsophisticated that they do not even know what it is to hold beliefs for good or bad reasons, then it is not clear that such children are really capable of drawing inferences, in the reflective, consciously aware sense of considering premises and drawing conclusions on the basis thereof.15 III (ii) Causal Deviance Ram Neta (2013)16 argues that if the Taking Condition is construed as the condition that the subject represents (doxastically or nondoxastically) that the premises support the conclusion, then there will be counterexamples where a subject S satisfies the relevant taking condition, and S’s belief in the premises causes S to believe the ‘conclusion’, but where the belief in the conclusion is caused in a deviant manner by the belief in the premises and the meta-belief. In such a case, S does not draw the inference.17 Causation of belief in a conclusion by belief in premises, in conjunction with possessing a meta-belief that would

132

Patrick Bondy and J. Adam Carter

satisfy the Taking Condition, is thus insufficient for drawing an inference. For example: Roderick has done nothing of value throughout his life except for offering a solution to the Gettier problem. Now, on his deathbed, Roderick thinks back on his otherwise worthless life and says “well, at least I solved the Gettier problem”. But as he says this, Timothy bursts into his room and proves to Roderick that his solution to the Gettier problem is unsuccessful. Roderick now believes that his solution to the Gettier problem is unsuccessful, and also believes that this very fact supports the conclusion that his life was worthless. Believing these two things, Roderick falls into a state of despair, and, out of despair, believes that his life was worthless. Of course, that his life was worthless is a conclusion that Roderick could have inferred from two beliefs that he has, viz., that his solution to the Gettier problem fails, and that the failure of his solution to the Gettier problem supports the conclusion that his life was worthless. But Roderick does not make this inference, and it is not on the strength of any such inference that Roderick believes that his life was worthless. Rather, it is out of despair that he believes that his life was worthless. (Neta 2013, p. 390) The case of Roderick appears to be a counterexample to Taking (Doxastic) because Roderick has two first-order beliefs (P: he has failed to solve the Gettier problem; C: his life was worthless), and he has a meta-belief about those first-order beliefs (that belief in P supports belief in C), but he does not draw the inference from P to C. This counterexample shows that in formulating a doxastic construal of the taking condition, we need to be explicit that the subject does not simply have a mental representation of separate propositions P and C, and a meta-belief that belief in P would support belief in C. S must represent her own belief in P as a good reason for her own belief in C. Taking (Reflexive): S takes P to support C iff S believes that her own belief in P supports (/is a good reason for) her own belief in C.18 Taking (Reflexive) explains what is going wrong in the case of Roderick: Roderick has two beliefs, P and C, and he has a belief that P supports (the content of) C, but he has not drawn the connection between them. He does not believe that his very belief in P supports his very belief in C. That is why he does not inferentially base his belief in C on his belief in P. But whenever S believes P, and believes C, and believes that her very belief in P is a good reason for holding her very belief in C, S inferentially bases her belief in P on her belief in C. One potential worry about Taking (Reflexive) is that it brings the overintellectualization problem back into focus. Earlier, we considered the

The Superstitious Lawyer’s Inference

133

problem that the Taking (Doxastic) rules out young children and higher animals from drawing inferences because they appear to lack the concept of a normative reason. We responded that they do after all have a rudimentary concept of a normative reason. And now the overintellectualization worry comes back as a problem for Taking (Reflexive) because young children and higher animals might have beliefs that are based on reasons (e.g., based on visual or olfactory perception), but they might also lack the concept of belief, in which case they cannot have beliefs about their own beliefs. If that is right, then young children and higher animals cannot satisfy Taking (Reflexive), and so according to our account, they cannot draw inferences after all. We can respond to the overintellectualization worry for Taking (Reflexive) as we did in the context of Taking (Doxastic). For one thing, just as young children and higher animals plausibly have a rudimentary concept of good reasons, so too they plausibly have at least a rudimentary concept of belief. For example, when a gazelle hides from a lioness, the gazelle is trying to ensure that the lioness doesn’t notice that the gazelle is there. Plausibly, the gazelle represents its own perspective on the world as distinct from the lioness’s perspective. It tries to ensure that the lioness’s perspective does not include a correct representation of the gazelle’s location, and so the gazelle stays downwind so that the lioness doesn’t discover the gazelle by smelling it. If that is right, then the gazelle does have at least a rudimentary conception of the lioness’s beliefs. Further, as we note in section III (i), if we turn our attention to children so young or animals so intellectually undeveloped that they don’t have even a rudimentary conception of belief, then it just is not clear that such children and animals are capable of drawing inferences, and we should not want our account of inference to include them.19 Might one, at this stage, attempt to challenge Taking (Reflexive) by appealing to a ‘Lewis Carroll’-style regress argument? In his famous regress argument, Carroll (1895) envisaged a student who was told P, and told that P entails Q, and told that if P & (P entails Q), then Q. But yet, the student (we may imagine) still does not know how to draw the inference; the student, despite this propositional knowledge, failed to see that her belief that Q follows from the premises the student believes. And this would remain the case for any further iteration of a belief about the premise and conclusion (or their relationship to one another) that the teacher could possibly tell the student. Carroll’s argument—borrowed to similar effect by Ryle (1945)—was meant to cast doubt upon the idea that the kind of ‘know-how’ involved in skillfully drawing an inference is something that will ever be secured by simply possessing propositional knowledge (and ipso facto, by simply possessing beliefs) about how the premise and conclusion stand in relation to each other. By parity of reasoning, we might imagine Carroll challenging Taking (Reflexive) along broadly analogous lines: S might (as the argument would go) believe that her own belief in P supports (/is a good reason for)

134

Patrick Bondy and J. Adam Carter

her own belief in C, but (like Carroll’s student—failing to see the connection between P and C) nonetheless fail to take P to support C. The problem with this objection is that the taking relation that is relevant to characterizing inference is weaker than the kind of relation Carroll purported to show could not be secured by the possession of propositional attitudes. That relation was a success-oriented relation: one that involved a kind of factive seeing. We are happy to grant Carroll (and Ryle) that a thinker’s believing that her own belief in P supports (/is a good reason for) her own belief in C does not imply that that thinker sees how P supports C. But this is perfectly compatible with countenancing that the fact that thinker’s believing that her own belief in P supports (/is a good reason for) her own belief C suffices for that thinker to take her P belief to support her C belief. This reply can be helpfully appreciated against the background of a more general point: not all inferences must be either good inferences or for that matter skillful inferences. Novice (and intermediate, and even expert) logicians sometimes mistakenly draw inferences in accord with faulty inference-rules. And what goes for inference goes for basing: sometimes, a subject S will hold a belief in C on the basis of a belief in a good reason P, but S misunderstands the support-relation at hand. In such cases, S’s holding a belief on the basis of the available good reason is insufficient for doxastic justification.20 What is required for doxastic justification is not just basing of one’s belief on good reasons, but correct, or perhaps skillful, basing of one’s belief on good reasons.

IV The Superstitious Lawyer, Reconsidered With the foregoing considerations in hand, let’s now look again at the Superstitious Lawyer case, which is summarized pithily by Kvanvig (1985) as follows: The counterexample concerns a [sic. superstitious] lawyer who, like the rest of his contemporaries, takes his client to be guilty. However, because of his [sic. superstitious] nature, the lawyer is inclined to trust what the tarot cards say, and upon learning that the tarot cards say that his client is innocent, comes to believe that his client is innocent. What the tarot cards say also prompts the lawyer to re-examine the evidence, which the lawyer comes to recognize conclusively establishes that his client is innocent. However, given his rather impressionable character, the lawyer also realizes that were the sustaining power of the tarot cards removed, the sway of public opinion would cause him to be unable to see that the evidence establishes his client’s innocence. Nonetheless, the lawyer now justifiably believes that his client is innocent on the basis of his examination of the evidence. But this examination of the evidence neither prompts

The Superstitious Lawyer’s Inference

135

his belief that his client is innocent nor does it sustain his belief that his client is innocent—his belief in what the tarot cards say holds the dubious distinction of being responsible for both. (1985, pp. 153–154) Despite the complexity of the case, a simple point can be put on the table. Once the lawyer’s superstition causes him to re-examine the evidence, it’s clear that the lawyer believes that his own belief in the evidence (call this ‘P’) supports (/is a good reason for) his own belief in the conclusion that the client is innocent (call this ‘C’). According to Taking (Reflexive), it follows that the lawyer thereby takes P to support C. This point, in conjunction with our argument in §III, makes for a simple diagnosis of the superstitious lawyer case. (i) An inference from believed premises to a believed conclusion is sufficient for basing belief in the conclusion on belief in the premises; and (ii) the superstitious lawyer makes an inference from believed premise(s) to a believed conclusion. And so the case is straightforwardly a case of inferential basing. Finally, because the evidence is a good reason to believe the conclusion—and, moreover, we have no good reason to think the inference itself is not a skillful or competent one21—the superstitious lawyer’s belief is plausibly doxastically justified. And this is the case despite the evidence not causing the belief. The sheer complexity of the case, we think, is at least partly responsible for confusions shared on both sides of the debate.22 These complexities, we believe, are ultimately unnecessary. The thrust of the example can be made much more simply. Consider the following ‘pared down’ version of the case, one that does what the original case aims to do, but without unnecessary distractions: Brain Scientists A lawyer is working for a cabal of brain scientists. She has good evidence for believing her client is innocent, and through careful and skilled reasoning recognizes it as good evidence, but the scientists have rendered evidential considerations with respect to her client’s guilt or innocence causally inert for her. The lawyer would make the evidence causally effective, if she could, but she cannot. Nevertheless, she fully believes that the evidence supports her belief in her client’s innocence. Intuitively, the lawyer in Brain Scientist has drawn an inference to the conclusion that her client is innocent. Further, because the lawyer has drawn this inference and done so skillfully, her belief in her client’s innocence appears to be doxastically justified. At least, this is clearly not a textbook case of propositional justification without doxastic justification.

136

Patrick Bondy and J. Adam Carter

In standard cases of that sort, a subject possesses good evidence but disregards it, or fails to properly appreciate it. By contrast, the lawyer in Brain Scientist properly appreciates the evidence she possesses; the evidence is just incapable of causally supporting her belief. This case controls for the kind of unnecessarily loaded aspects of the explanation for why the evidential considerations are causally inert, which we find in Lehrer’s original case. By simply appreciating (i) that the evidential considerations are causally inert; along with the fact that (ii) normal inference takes place, the force of the strategy the superstitious lawyer case initially used can be better appreciated as seriously problematic for proponents of the causal theory of the basing relation. By focusing our attention on the simple structure of the strategy, the burden on the causal theorist sharpens: the causal theorist can dodge this kind of case only if they can give a good explanation for why in the presence of normal inference the causal inefficacy of a reason should matter for basing. Thinking about the case in this way will accordingly make certain kinds of responses less relevant. This includes Goldman’s response (on behalf of the proponent of the causal theory) that the lawyer (contrary to what Lehrer originally suggests) lacks knowledge. Our reading of the landscape commits us to viewing the situation of the superstitious lawyer as analogous to the situation in the Well Founded case we discuss in Section II, in that it is a case of inferential basing that leads to doxastic justification. Importantly, though, inferential basing that leads to doxastic justification can be unsafe in a way that leads the target belief to fall short of knowledge.23 (Compare: subjects in barn façade cases have paradigmatically well-founded beliefs, and yet their beliefs are unsafe because very easily they could have been false given how they are formed.) And, of course, the superstitious lawyer could easily have drawn a different tarot card than the one he did. What this means is that the structure of the superstitious lawyer case— which is highlighted in our revised version—is one that suffices to make the case against the causal theory regardless of whether the belief one ends up with counts as knowledge—and thus, regardless of whether one’s evidence lacking causal efficacy is enough to undermine knowledge of the target proposition.

V Conclusion Superstitious lawyer-style cases are surely unusual—indeed, for all we have said here, they might even be necessarily exceptional. For example, perhaps the capacity to draw inferences depends on a subject’s having the capacity to causally regulate her beliefs in accord with her appraisal of the available evidence in normal cases. We do not argue that that is so, but neither have we said anything to rule out that sort of possibility. And

The Superstitious Lawyer’s Inference

137

it also bears emphasizing that we have only been concerned here with beliefs involved in conscious inferential processes. For all we have said here, non-deviant causation of beliefs by reasons might suffice for basing in cases of non-inferential belief formation. Although we have not argued that the superstitious lawyer has knowledge of his client’s innocence, we find it intuitively very plausible that he bases his belief on the available evidence, and that—given that he does this normally and plausibly skillfully—his belief is consequently doxastically justified. But this verdict about the case does not simply rest on intuition. For if a refined doxastic construal of the taking condition on inference of the kind we have proposed in Section III is correct, then in superstitious lawyer-style cases, the subject genuinely draws an inference from the available evidence to the target belief, and consequently holds the target belief on the basis of the evidence. Because the lawyer’s evidence is good evidence for the target belief, and he competently appreciates its force, the lawyer’s belief is doxastically justified, even though the evidence for it is causally inert.24

Notes 1. See, however, Turri (2010) and Lord and Sylvan (this volume) for notable lines of resistance to this orthodoxy. 2. Champions of this kind of approach include, e.g., Harman (1970), Swain (1978), Moser (1989), Turri (2011), and McCain (2012). 3. See Korcz (2015) for an overview. See also Bondy (2016). 4. See, for example, Kvanvig (1985, 1987), Lemke (1986), Tolliver (1982), Swain (1981), and Goldman (1979). 5. When we say “reason” without qualification in what follows, we mean what are often called motivating reasons: reasons which prompt belief or action, or which the subject takes or would take upon reflection to recommend belief or action. Motivating reasons for belief can be good or bad. When they are good, they are normative reasons, and they can doxastically justify belief. When they are bad, the subject might still hold beliefs on the basis of them, but her beliefs will not thereby be doxastically justified. 6. It is worth noting that in ‘no reason’ cases, such as this, the subject is sometimes claimed to possess what is called a ‘no reason defeater’. For discussions of no-reason defeaters, see Plantinga (1994) and Bergmann (1997). 7. Perhaps it will appear to some as though Sam’s belief is based on something, viz. her hunch. If so, that is no problem; Sam’s feeling upon waking up is a natural but dispensable piece of the example. Instead, we can imagine that Sam has been hypnotized into believing that it will rain, say, as part of a course of treatment for a debilitating case of heliophobia. Then there is clearly nothing on which her belief is based. 8. At least, this is the prevailing view among epistemologists, but see Sylvan and Lord (this volume) for an argument to the contrary. 9. Non-deviance needs to be spelled out in order for the causal account to be satisfactory, of course, but we can set that aside for now. Our aim in this paper is not to raise causal deviance problems. See Ye (this volume) and Turri (2011) for attempts to handle causal deviance in the account of the basing relation.

138

Patrick Bondy and J. Adam Carter

10. See Korcz (2000) for an example of a hybrid account. 11. For instance, according to Lord and Sylvan (this volume), the basing relation is prime. 12. See Section IV for a full description and discussion of the case. 13. Some philosophers also distinguish theoretical inference from practical inference, in virtue of the type of attitude featuring in the conclusion: theoretical inferences conclude with beliefs, and practical inferences conclude with intentions. For the purpose of this chapter, we are only addressing theoretical inferences, and we do not take a stand on this distinction. Even if practical inferences are distinguished from theoretical inferences in that way, the doxastic construal of the taking condition that we defend here will apply to practical inferences in the very same way as it applies to theoretical inferences. 14. For example, Boghossian (2014) and Neta (2013). 15. Of course, even these very young children must make some kinds of mental moves, transitions in thought that fall short of basing. A natural way to think about such transitions in thought is in terms of cognitive heuristics, mental shortcuts which do not correspond with any kind of rational regulation, but which nonetheless can be practically useful. For a recent discussion of cognitive heuristics—described in the language of Type-1 thinking—see Kahneman (2011). 16. Neta is following Boghossian (2014) here, but Neta makes this point much more explicitly than Boghossian does. 17. Another kind of objection in this neighbourhood is highlighted by Wallbridge (2018), who writes that doxastic accounts fail: to capture the sense in which the basis of a belief is what explains why one believes as one does. To the extent that the kinds of meta-beliefs proposed by doxastic accounts are able to provide an explanation, they do so in virtue of their covariance with causal or counterfactual factors. (2018, 1619) This strategy of objection, we think, problematically presupposes that the only kind of explanation that would be captured by the pretheoretical intuition that bases explain why one believes what they do is causal. If it were, then a causal theory would be the only kind suited to satisfying the pretheoretical intuition. However, the doxastic account proponent can reasonably resist Wallbridge’s preferred way of reading the pretheoretical intuition while maintaining that the doxastic account satisfies a more ecumenical version of it. 18. We treat Taking (Reflexive) as a condition requiring full belief here, but note that it can be weakened by giving it a “de re awareness” reading, according to which S need not form the full-fledged belief that her belief in P supports her belief in C; S only needs to be aware of her own belief in P as supporting her own belief in C. (See Korcz (2015, section 3), and Moser (1989) for more on the de re awareness condition.) What we are most concerned to emphasize in Taking (Reflexive) is not what kind of belief-like representation or awareness the subject has, but simply that the representation is reflexive. 19. Of course such children and animals would make some mental transitions, but those transitions need not count as inferences. See note 15. 20. Turri offers a number of examples to illustrate the problem of beliefs held badly on the basis of good reasons. For example: in a case where P1-P4 make it very likely that Mansour is Guilty, two members of the jury form the belief that he is guilty. Miss Proper reasons like so:

The Superstitious Lawyer’s Inference

139

(Proper Reasoning) (P1—P4) make it overwhelmingly likely that Mansour is guilty. (P1—P4) are true. Therefore, Mansour is guilty. Miss Improper, by contrast, reasons like this: (Improper Reasoning) The tea leaves say that (P1—P4) make it overwhelmingly likely that Mansour is guilty. (P1—P4) are true. Therefore, Mansour is guilty. (2010, p. 316) 21. One might object here that the inference is not skillful because, were the sustaining power of the tarot cards removed, the sway of public opinion would have prevented the inference from being drawn successfully. This reasoning, though, is problematic; it conflicts with ordinary thinking about skill in epistemology. To borrow an example from Sosa (2015), suppose a skilled basketball player makes a free throw, and that the free throw manifests the shooter’s skill. Now, add to the story that the electricity to the gymnasium could have suddenly failed, and had it failed, the shot would have missed due to darkness. But it did not suddenly fail. Does the fact that the electricity could have failed—even if it could easily have failed—count against the shooter’s possessing and manifesting free-throw shooting skill in making the shot while the lights are on? Very plausibly, the answer here is ‘no’; and a similar point we think goes for the superstitious lawyer. True, he would be unreliable if the initial conditions were altered. But, situated as he is, he is, we may presume, in an analogous position to that of a skilled basketball shooter whose reliability requires that the lights stay on and would be thwarted if they weren’t. Put another way, the lawyer is in a very different position from a hopeless counterpart lawyer who (like the original lawyer) is also a follower of tarot who is easily swayed by public opinion, but who is, in addition, inferentially challenged. The lawyer, but not the counterpart, is situated so as to infer reliably. 22. Consider, for example that the following counterfactual is true of the lawyer in the original case: if they did not base the belief on the good reason they have, then they would have (rather than withheld judgement) based the belief on a bad reason they have. We can of course grant this conditional while stressing that the lawyer nonetheless does in fact base their belief on a good reason. Korcz (2000) presents other useful variations on the superstitious lawyer case, but they are at least as complex as the original. Nowhere in the literature do we find straightforward examples that illustrate the relatively simple idea that Lehrer was driving at with his complicated case. 23. On one canonical way of thinking about safety and knowledge due to Pritchard (2005), a belief is known only if it is safe in the sense that it couldn’t easily have been false given the conditions of its formation. 24. We are grateful to an anonymous referee for helpful comments that have improved the paper.

References Audi, Robert. 1982. “Believing and Affirming.” Mind 91 (361): 115–120. Bergmann, M. A. 1997. Internalism, Externalism and Epistemic Defeat (Doctoral dissertation, University of Notre Dame). Boghossian, Paul. 2014. “What Is Inference?” Philosophical Studies 169: 1–18. Bondy, Patrick. 2016. “Counterfactuals and Epistemic Basing Relations.” Pacific Philosophical Quarterly 97 (4): 542–569.

140

Patrick Bondy and J. Adam Carter

Carroll, Lewis. 1895. “What the Tortoise Said to Achilles.” Mind, 104 (416): 691–693. Goldman, Alvin I. 1979. “What Is Justified Belief?” In Justification and Knowledge, 1–23. Springer. Harman, Gilbert H. 1970. “Knowledge, Reasons, and Causes.” Journal of Philosophy 67 (21): 841–855. Kahneman, Daniel. 2011. Thinking, Fast and Slow. New York: Farrar, Straus and Giroux. Korcz, Keith Allen. 2000. “The Causal-Doxastic Theory of the Basing Relation.” Canadian Journal of Philosophy, 30 (4): 525–550. ———. 2015. “The Epistemic Basing Relation,” in Stanford Encyclopedia of Philosophy (Fall ed.), ed. Edward N. Zalta. https://plato.stanford.edu/archives/ fall2015/entries/basing-epistemic/. Kvanvig, Jonathan L. 1985. “Swain on the Basing Relation.” Analysis 45 (3): 153. ———. 1987. “On Lemke’s Defence of a Causal Basing Requirement.” Analysis 47 (3): 162–167. Lehrer, Keith. 1971. “How Reasons Give Us Knowledge, or the Case of the Gypsy Lawyer.” Journal of Philosophy 68 (10): 311–313. Lemke, Lory. 1986. “Kvanvig and Swain on the Basing Relation.” Analysis 46 (3): 138–144. McCain, Kevin. 2012. “The Interventionist Account of Causation and the Basing Relation.” Philosophical Studies 159: 357–382. Moser, Paul. 1989. Knowledge and Evidence (Cambridge, UK: Cambridge University Press). Neta, Ram. 2013. “What Is an Inference?” Philosophical Issues 23: 388–407. Plantinga, Alvin. 1994. “Naturalism Defeated.” Unpublished manuscript. Pritchard, Duncan. 2005. Epistemic Luck. Oxford: Oxford University Press. Ryle, Gilbert. 1945. “Knowing How and Knowing That: The Presidential Address.” Proceedings of the Aristotelian Society (46): 1–16. Sosa, Ernest. 2015. Judgment and Agency. Oxford: Oxford University Press. Swain, Marshall. 1978. “Reasons, Causes, and Knowledge.” Journal of Philosophy 75 (5): 229–249. ———. 1981. Reasons and Knowledge. Vol. 92. Ithaca, NY: Cornell University Press. Tolliver, Joseph. 1982. “Basing Beliefs on Reasons.” Grazer Philosophische Studien 15: 149–161. Turri, John. 2010. “On the Relationship between Propositional and Doxastic Justification.” Philosophy and Phenomenological Research 80 (2): 312–326. Turri, John. 2011. “Believing for a Reason.” Erkenntnis 74: 383–397. Wallbridge, Kevin. 2018. “The Peculiar Case of Lehrer’s Lawyer.” Synthese 195 (4):1615–1630.

8

Prime Time (for the Basing Relation)* Errol Lord and Kurt Sylvan

1 Introduction Sometimes we believe things for reasons. For example, Sam’s reason for believing that Billy is kind is that he gives to Oxfam. Lois’s reasons for believing that the paper should be accepted are the paper’s excellent ideas and examples. Sam and Lois differ from Jean, whose belief that Abuja is the capital of Nigeria was caused by a blow to the head. While there is a reason why Jean believes what she does (i.e., an explanatory reason), there are no reasons for which she believes what she does (i.e., no motivating reasons).1 Sometimes we also believe things for sufficient normative reasons, which are factors that count strongly enough in favor of our beliefs to provide justification for them.2 Billy’s charitable giving may be a sufficient reason to believe that he is kind. And the paper’s examples and ideas may be good enough to justify believing that it should be accepted. If this is so and Sam and Lois believe these things because they have these sufficient normative reasons, then they are not only to be distinguished from Jean, but also from Freddie. Freddie’s reasons for believing the temperature on April 12 will be 48 degrees are that 12 × 4 = 48 and that the temperature on any day is the product of the number of the month and the number of the day. Freddie too believes for reasons—just crazy ones. In this chapter, we are interested in what it takes to believe for sufficient normative reasons.3 Many may find this question less interesting than the broader question of what it takes to have certain considerations as one’s motivating reasons for belief. This is partly because it is natural to think that Freddie has more in common with Sam and Lois than he does with Jean. Sam, Lois, and Freddie are all believing for reasons— Sam’s and Lois’s just happen to line up with sufficient normative reasons. Thus, it is tempting to think that the main question to ask is the broader question of what it is to believe for reasons, period. Believing for sufficient normative reasons is merely a special case, on this tempting view, of the more general phenomenon of having certain considerations as one’s

142

Errol Lord and Kurt Sylvan

motivating reasons for belief. Yielding to this temptation leads people to tacitly assume: The Composite View: Believing that p for a sufficient normative reason r involves nothing more than (i) believing that p on the basis of a motivating reason r, and (ii) r’s corresponding to a sufficient normative reason to believe that p, where (i) and (ii) are conditions that could obtain independently of each other.4 If this view were true, the question that interests us would reduce to two questions: (i) the question of what it takes for a consideration to be one’s motivating reason for belief, and (ii) the question of what it is for something to be a sufficient normative reason for belief. Once we answer (i) for subjects like Sam and Lois, we can just tack on that the considerations for which they believe line up with sufficient normative reasons and thereby answer our question. End of story. The Composite View is widely assumed in epistemology, rarely flagged explicitly, and never defended. (Audi, 1993, 267), for one, gives eloquent voice to it: Since believing for a good reason is believing for a reason (one that is good), the account [of believing for a reason] clarifies believing for a good reason. . . . Indeed, if an indirectly (prima facie) justified belief is simply a belief held for at least one good reason, then if our conditions [for believing for a reason] are supplemented with an account of what constitutes a good reason, we shall have all the materials we need to understand one of the main kinds of justified belief and, in good part, one of the main kinds of knowledge. Revealingly, this passage occurs in a short section in a long paper on believing for reasons. Here we see confidence that a satisfactory account of believing for normative reasons will fall out as a special case of an account of motivating reasons for belief. What we see in Audi surfaces elsewhere. Often one finds lengthy discussions of what it takes for a consideration to be one’s motivating reason for belief that are followed either by brief discussions of believing for normative reasons, or by no discussion at all. Consider (Armstrong, 1973, 96): In all this discussion of what it is for a proposition actually to function in a man’s mind as a reason for belief, nothing has been said about the goodness of such reasons. This is because the goodness or badness of a reason has nothing essential to do with its actually operating for somebody as a reason.

Prime Time (for the Basing Relation)

143

As in Audi, this passage appears in a short section on believing for normative reasons that follows a lengthy discussion of believing for reasons. Some who tacitly assume the Composite View don’t even go so far as to separately discuss believing for normative reasons. (Swain, 1981, 82) starts a section on basing beliefs on reasons by saying: “Let me note at the outset that I am concerned in this section to define the basing relation without regard to justificatory considerations.” While he goes on to say nothing about basing beliefs on normative reasons, he does appeal to his account of basing beliefs on reasons in giving an account of doxastic justification. Such moves make sense if, and perhaps only if, the Composite View is granted.5 As we will see later, the Composite View is also crucially assumed in recent work on the relationship between propositional and doxastic justification, and on whether reasons can play a fundamental explanatory role in epistemology.6 It is required to support some of the surprising claims that have recently been made about these topics. So the Composite View is an important view, in spite of the fact that it is never foregrounded in discussion. Of course, part of why it has never been foregrounded is that it strikes many as too obvious to be worth discussing. It admittedly has some attraction. But it is worth thinking about why. For the Composite View is, we believe, mistaken. As we will see in §2, the reasoning needed to support the Composite View is non-trivial. Indeed, the similarities that this reasoning bears to manifestly problematic reasoning in other areas (e.g., the philosophy of perception) will lead nicely to an argument we will give in §3 against the Composite View. Having given this argument, we will turn in §4 to our alternative to the Composite View, and in §5 to payoffs and broader morals.

2 The Non-Triviality of the Composite View 2.1 The Reasoning Behind the Composite View In this section, we consider two arguments that may seem to make the Composite View irresistible. As we will see, there are very general reasons for finding these arguments suspect; structurally similar arguments in other areas are obviously flawed, and it is easy to see how the flaws could extend to the case at hand. So, if anything, the fact that these are the only obvious arguments for the Composite View should reduce our confidence in the view, not increase our confidence. The first argument starts with the plausible thought that Sam, Lois, and Freddie all have something important in common. This is not to deny that Sam and Lois achieve something that Freddie does not achieve.

144

Errol Lord and Kurt Sylvan

But there is a single thing that Freddie, Sam, and Lois are doing. Since it cannot be true that they are all believing for normative reasons—Freddie is insane, after all—it must be that what they all have in common is that they believe for reasons. Now we can ask: what more do Sam and Lois achieve? Well, the reasons for which Sam and Lois believe line up with honest-to-goodness normative reasons. Given this, it is tempting to think that the main difference between Sam and Lois on the one hand and Freddie on the other hand is that Sam’s and Lois’s motivating reasons line up with sufficient normative reasons, where this further fact is an independent fact—i.e., one that obtains independently of the fact that they believe for those motivating reasons. This independent fact accounts for their achievement. So, it may seem that the achievement of believing for sufficient normative reasons amounts to nothing more than (i) having some motivating reasons for belief, where (ii) these motivating reasons happen to line up with sufficient normative reasons, and where (i) and (ii) are independent factors. Here the success factor at which the achievement aims is the correspondence of the motivating and normative reasons; this is, of course, a “success” in a somewhat wider sense than that normally intended when epistemologists talk about successes and achievements. While this reasoning may seem impeccable, notice that it is an instance of the following broader pattern of thinking:7 Mere Composition From the Bad Case (1) One identifies some achievement—say, seeing that p, intentionally producing an outcome, or believing for a normative reason. (2) One notes that there is a feature in common between the achievement case and a botched case—say, having an experience as of p when p corresponds to no fact, intending an outcome without the outcome’s occurring, and believing for a motivating reason that corresponds to no normative reason. (3) One infers that the state in the botched case must be independent of whatever success factor is present in the achievement case (e.g., the intended outcome’s occurring, the world’s being as represented by the experience, and the motivating reason’s corresponding to a normative reason). (4) One concludes that the achievement is a mere composite of the state in the botched case and the success factor. This pattern of reasoning is bad. Perception provides a vivid illustration of why it is bad. Seeing that p cannot just be having an experience as of p, plus p’s corresponding to a fact. Veridical hallucination is a clear possibility. Even if we follow Grice (1961) and company in thinking that what we need to add is a causal relation between the fact that p and the

Prime Time (for the Basing Relation)

145

experience as of p, we will also face the problem of deviant causal chains. The fact that this pattern of reasoning is dubious should lead us to doubt the first line of support for the Composite View.8 Let’s consider a second line of support for the Composite View. It rests on the thought that the following sort of argument is obviously valid: Substitution Argument (1) Sam’s motivating reason for believing that Billy is kind is that Billy gives to Oxfam. (2) The fact that Billy gives to Oxfam is a sufficient normative reason to believe that Billy is kind. (3) So, Sam has the belief that Billy is kind for a motivating reason that corresponds to a sufficient normative reason. (4) Therefore, Sam believes that Billy is kind for a sufficient normative reason. We can stipulate that (1) and (2) are true. The move from (1) and (2) to (3) is obviously valid. And the move from (3) to (4) looks unexceptionable. But looks are deceiving. The final move is in fact questionable. Once again the case of perception is revealing. Consider the following analogous argument. Bad Argument (1*) Sam’s visual experience has the content that the house is on fire. (2*) That the house is on fire is a fact. (3*) So, Sam has a visual experience with a content that corresponds to a fact. (4*) Therefore, Sam visually experiences the fact that the house is on fire. The move from (3*) to (4*) is not valid. Once again, cases of veridical hallucination and cases involving deviant causal chains will be cases in which (3*) is true but (4*) is false. Perhaps we can get ourselves to hear “experiences a fact” as meaning has an experience with a content that corresponds to a fact. But even if that phrase could be heard in that way, there is another relation we can naturally take to be picked out by this phrase, and this relation is not a composite one. This obvious precedent in the philosophy of perception should at least make us pause when we hear the case for the Composite View. One might naturally worry that just as a deviant connection between (i) the fact that p and (ii) an experience as of p can prevent one from seeing that p, so a deviant connection between (i*) the fact that a consideration happens to correspond to a normative reason and (ii*) the fact that this consideration is one’s motivating reason can prevent one from believing for the normative reason to which the motivating reason corresponds.

146

Errol Lord and Kurt Sylvan

We will develop this worry for the Composite View in detail in §3. For now, we are simply marking that there are general reasons for being wary of the kind of reasoning that supports the Composite View. The symmetry between the Substitution Argument and the Bad Argument should make us expect that there is an important relation that is not a mere composite of having a motivating reason for belief and that reason happening to correspond to a normative reason. This relation will stand to the relation of having a motivating reason for belief as the relation of seeing that p stands to having an experience as of p. The basic point of this chapter is to present a case for the existence and importance of this relation. Naysayers might insist on stipulating that “believes for a sufficient normative reason” should be understood to mean has a motivating reason for belief that corresponds to a sufficient normative reason, just as some might conceivably insist on stipulating that “experiences the fact that p” be understood to mean has an experience with a content that corresponds to the fact that p. But in both cases, what really matters is that there is another recognizable relation worthy of the name that cannot be understood compositely. This is the relation, we claim, that constitutes a genuine achievement. Moreover, we think that in the case of believing for normative reasons, the relation to which epistemologists should appeal in doing the important theoretical work is the one that constitutes an achievement.9 Thus, if we can establish that there is a recognizable and important relation that we can bear to normative reasons worth calling “belief for a sufficient normative reason” that is not composite, we will have achieved everything we wanted to achieve. 2.2 What Alternatives? The points from the last subsection should at least make us take alternatives to the Composite View seriously. But what are the alternatives? Perhaps because the Composite View has always been a tacit assumption, alternatives to it have never been clearly articulated. In this subsection, we will outline the basic structure that any satisfactory alternative must have. Later in the chapter (§4) we will develop and defend a specific alternative. The first constraint on any successful alternative is that it must entail that believing for a sufficient normative reason is prime, where a condition (e.g., some property or relation) is prime iff it cannot be analyzed into proper parts that can obtain independently of each other.10 The Composite View entails, of course, that believing for sufficient normative reasons is not prime. This is because the conjuncts that constitute the analysans can obtain independently of each other. As we will see, this feature of the Composite View exposes it to counterexamples. Importantly, failure to secure primeness is also what dooms the bad views in the philosophy of perception mentioned in the last subsection.

Prime Time (for the Basing Relation)

147

The problem that arises in cases of veridical hallucination is that there is not the right connection between having an experience as of p and the fact that p. By building into the analysis that these two conditions do not obtain independently of each other, prime views go a long way toward blocking cases where the right connection is missing. Of course, one cannot avoid problems of deviance simply by adopting a prime view. One can see this by thinking about a simple-minded causal theory of perception. On this view, seeing that p is analyzed in terms of (i) having an experience as of p, (ii) it being a fact that p, and (iii) condition (i) holding because condition (ii) holds. This view is prime. The conditions on the right hand side cannot be met independently of each other. However, this simple-minded view is also open to counterexamples involving deviant causal chains. Our ultimate view of believing for sufficient normative reasons will avoid analogous counterexamples. It will not be a simple causal view but rather a sophisticated dispositional view. Still, the first step toward the truth is to adopt a view on which believing for sufficient normative reasons is prime, in the sense that the relation that one bears to a normative reason in believing for that normative reason is not independent of the fact that the reason is a normative reason. In the same way that classic arguments tell against composite accounts in the philosophy of perception, our case against the Composite View will show why the question “What does it take to believe for a normative reason?” is not a special case of the question “What does it take to have a motivating reason for belief?” To believe for a normative reason, there must be a connection between the fact that what one has is a normative reason and the fact that one believes as one does. If there isn’t, there will be room for deviance of a kind that precludes one from believing for a normative reason. A common view in the philosophy of perception is that seeing that p and having an experience as of p are two different relations, though they have things in common.11 This is what the counterexamples show, many think. Seeing that p is more than having an experience as of p plus something else independent.12 Our view resembles this view in the philosophy of perception. Believing for a sufficient normative reason and believing for a motivating reason are two different relations, though they have things in common. This is what the counterexamples to the Composite View show. Just as seeing that p is not just having an experience as of p plus something else independent, so believing for sufficient normative reason is not just believing for a motivating reason plus something else independent.13 Thus, we think that the failure of the Composite View calls for two things. First, it demands that we hold that believing for sufficient normative reasons is prime. Second, it demands that we hold that the believingfor-sufficient-normative-reasons relation is a different relation than the believing-for-reasons relation, even plus independent factors. In §4, we

148

Errol Lord and Kurt Sylvan

will defend an account that has these features. Before doing so, however, we will further explain why the Composite View fails.

3 The Deviance of the Composite View 3.1 The Core Argument in Outline Our core argument against the Composite View is simple: 1. If believing on the basis of a sufficient normative reason were a mere composite of two independent components—viz., (i) having a motivating reason for belief r, and (ii) r’s happening to correspond to a sufficient normative reason—then cases where the connection between (i) and (ii) is deviant would be cases of believing for a sufficient normative reason. 2. But subjects do not believe for sufficient normative reasons in these cases. 3. So the relation of believing for a sufficient normative reason is not a mere composite of (i) and (ii) (and thus the Composite View is false). In the next two subsections, we explain and defend (1) and (2), which will suffice to refute the Composite View. It will not suffice to establish our specific alternative. We save that task for §4. 3.2 Independence and Deviance The deviance that plagues Composite Views arises in cases like the following:14 Fortunate Consequent-Affirmer Sam wonders whether Terry took the bus to work. He knows that Terry’s car is in the driveway. This is, in fact, a sufficient abductive reason to think that Terry took the bus. Sam also believes that if Terry took the bus, then Terry’s car is in the driveway. But he comes to believe that Terry took the bus by inferring that he took the bus from his belief that Terry’s car is in the driveway and his belief that if Terry took the bus, then Terry’s car is in the driveway by following the invalid deductive rule: from , and , infer . Sam hereby manifests a general consequent-affirming incompetence. By stipulation, the fact that Terry’s car is in the driveway is a sufficient reason to believe that Terry took the bus.15 And Sam believes Terry took the bus on the basis of the consideration . But intuitively he doesn’t believe for a sufficient normative reason. (If you

Prime Time (for the Basing Relation)

149

don’t find that intuitive, we will argue for it at length in the next subsection.) For one thing, he affirms the consequent! More importantly, he is insensitive to whether the fact that Terry’s car is in the driveway is actually a good reason to believe Terry took the bus. To see some evidence for this, consider an extension of the story. Suppose the fact that Terry’s car is in the driveway isn’t always a sufficient reason to believe Terry took the bus. For Terry carpools on Wednesdays. But today is Tuesday, and so the fact that the car is in the driveway is a sufficient reason to believe that Terry took the bus. Notice that in this revised scenario, it is true that Sam would have been motivated by the same considerations even if it were Wednesday and the fact were an insufficient reason. This is evidence that Sam is insensitive to what makes the driveway fact a sufficient reason to believe that Terry took the bus. Indeed, as long as Sam believes the car is in the driveway, he will believe Terry took the bus. Now, to see why premise (1) is true, note that as long as the fact that a believer’s motivating reason corresponds to a sufficient normative reason is independent of her having that motivating reason, there will be room for deviant connections like the one in Fortunate Consequent-Affirmer. If the Composite View is true, one believes for a sufficient normative reason even in these cases. So, (1) is true. An objection should be addressed. Notice that in Fortunate ConsequentAffirmer, it is stipulated that Sam’s reason for believing that Terry took the bus is that Terry’s car is in the driveway and that if Terry took the bus, Terry’s car will be in the driveway. A friend of the Composite View might ask: “If this is Sam’s complete reason for believing that Terry took the bus, why can’t we just say that it is a bad reason, and that Sam’s belief is unjustified because his reason is bad?” This objection fails for a simple reason. We can imagine subjects who reason from the same beliefs as Sam but who manifest abductive competence rather than deductive incompetence. These subjects will be doxastically justified in holding their conclusion beliefs. But their premise beliefs will be the same (or so we can stipulate). So, one cannot escape the problem by calling the reason bad: this will lead us to wrongly regard as unjustified those who move from the same premises to the same conclusion while manifesting abductive competence. Of course, real people who reason abductively may well have further premise beliefs (e.g., that the observed effect is best explained by the posited cause). But not all competent abductive reasoners must have these further beliefs. One can imagine competent abductive reasoners who are just virtuously disposed to reason (e.g.) from effects to causes in a way that manifests sensitivity to the deeper explanatory relations, without having explicit beliefs about these relations. This last point also disposes of a more ambitious version of the objection. One might have instead objected that that Sam’s complete reason

150

Errol Lord and Kurt Sylvan

for believing that Terry took the bus must include either the false belief that affirming the consequent is a valid deductive rule, or a belief in the conditional that, for all p and q, if p  q and q, then p. One might have then insisted that that makes his complete reason bad, and this is why Sam does not believe for a sufficient normative reason. But we can stipulate that Sam lacks any such belief in the case, and our intuition doesn’t change if we do. More importantly, the objector’s move will require overintellectualization to have traction across the board. People usually do not have beliefs about the validity of certain patterns of thought or about the conditionals whose truth would ensure that what they are treating as normative epistemic reasons are normative epistemic reasons. More often, people simply use their inferential competence (or incompetence) to move directly from premises to conclusions, whereby they systematically conform to good (or bad) rules. And so not all cases of deviance must involve false beliefs about normative relations: they need only involve manifestations of inferential incompetence. In any case, even when people understand the rules to which they systematically conform, these rules will not figure in their reasoning as believed premises; Carroll (1895) showed that we cannot replace all rules with extra believed premises. The rules serve instead as enabling conditions. For the objector’s move to rule out all cases of deviance, the distinction between normative reasons and enabling conditions would need to be wholly collapsed. This is a mistake. There are many cases where we do not need to know why something is an epistemic reason to stand in the relation to it that lets us get justified by it. This is vivid for noninferential justification. Children can gain justified beliefs about the world via experience without having the concepts needed to understand why this is so. Children do not need to have the concept of an experience or of a reliable connection in order to gain justified beliefs from experience. This suggests that on views on which reliability is part of what explains why experience gives normative reasons, the reliable connection will serve as an enabling condition. The core point is not local to reliabilist stories. Internalists also tell stories about why experience gives normative reasons that need not be grasped by justified believers. Suppose the story has to do with facts about the phenomenology of perceptual experience. Such facts are not ones to which a subject needs reflective access to gain justified beliefs. Yet they may explain why experience gives reasons anyway. They would do so as enabling conditions, not as further reasons. The point also holds in the inferential case. A person may not have the fact that p as a reason to believe q, where p entails q via some inferential route, unless this person has certain inferential competences. But if a person has these competences to a minimal degree, she does not also need to believe that p entails q in order to be in a position to get a justified belief in q from p. She just needs to exercise her competence. So while

Prime Time (for the Basing Relation)

151

propositions about the links underwriting the inferences could serve as further premises, they need not serve as further premises. So, the objection will fail in deviance cases involving subjects with unsophisticated psychological profiles—that is, subjects who manifest inferential incompetence while lacking the psychological sophistication either to represent the bad rules to which they reliably conform or to have de dicto beliefs about the normative relations that would need to be in place for these rules to be good rules. 3.3 Deviance and Belief for Normative Reasons While (2)—the claim that Sam and his ilk don’t believe for sufficient normative reasons—is a more controversial premise, we think it is intuitively plausible on its face, as we have indicated. Sam and subjects like him simply do not believe for sufficient normative reasons. If true, this would capsize the Composite View. But it is possible that not everyone will share this intuition. So in this subsection, we will offer some independent arguments for (2) from some platitudes. Since we already explained in §2.1 why there is no valid intuitive argument for the opposite conclusion, we will rest content with these independent arguments. Let’s consider the first argument. Intuitively, when you base your belief in p on x rather than y, your believing p is explained by certain features that x has that y does not have. This is an instance of what we take to be the intuitively obvious principle that if a condition C rather than C* explains some state of affairs, this must owe to a difference between C and C*; after all, if C and C* had all the same features (including relational features), then how could C rather than C* explain the state of affairs? Call this the Explanation Platitude. Now, if you base your belief on a normative reason r, what features of r will have to explain your belief? A natural answer is that the very fact that r is a normative reason in favor of your belief must explain why you have that belief. Otherwise your believing p isn’t explained by the presence of a normative reason. It is just explained by your having a certain consideration in mind, which you may take to be a normative reason without being sensitive to the real normative relation that makes it a normative reason. The fact that this consideration corresponds to a real normative reason will be explanatorily irrelevant. This is what we see in deviant cases like Fortunate ConsequentAffirmer. Plausibly, then, subjects in deviant cases do not believe for normative reasons, but just for motivating reasons that by luck correspond to normative reasons. Let’s put this argument more officially:   i. If one’s belief is based on x rather than y, one’s belief must be explained by a distinctive feature that x has that y does not have.

152

Errol Lord and Kurt Sylvan

 ii. So, for one’s belief to be based on a normative reason as opposed to a motivating reason that merely happens to correspond to this normative reason, the very fact that there is a normative reason in favor of one’s belief must explain why one has that belief. (We will call this claim Explanation by Normative Reasons, and refer back to it by this name in the next section.) iii. In deviant cases like Fortunate Consequent-Affirmer, the very fact that there is a normative reason in favor of the deviant subject’s belief does not explain why the deviant subject has that belief. iv. So, in deviant cases, the subject will not believe for a sufficient normative reason as opposed to a motivating reason that merely happens to correspond to a sufficient normative reason. (iv) restates premise (2) in our core argument. So (2) is true. A similar argument can be given from a different platitude. In order to base your belief in p on x rather than y, your belief must be sensitive to certain features that x has that y does not have as such.16 Call this the Sensitivity Platitude. Now, if you base your belief that p on a normative reason r that you have to believe that p, to what features of r must you be sensitive? A natural answer is that you have to be sensitive to the normative relation that r bears to your belief in virtue of which r counts as a normative reason for believing that p. Otherwise you aren’t sensitive to the normative reason that you have to believe that p. In deviant cases like Fortunate Consequent-Affirmer, subjects aren’t sensitive to the normative relations that make their motivating reasons correspond to normative reasons. Plausibly, then, their beliefs aren’t based on the relevant normative reasons. Putting this more officially:   I. If your belief in p is based on x rather than y, your belief is sensitive to a distinctive feature that x has that y does not have.   II. So, for one’s belief to be based on a normative reason as opposed to a motivating reason that merely happens to correspond to this normative reason, one’s belief must be sensitive to the normative relation that holds between such reasons and such beliefs. (We will call this claim Sensitivity to Normative Reasons, and refer back to it by this name in the next section.) III. In deviant cases like Fortunate Consequent-Affirmer, the subject’s belief will not be sensitive to the normative relation that holds between her reason and her doxastic attitude. IV. So, in deviant cases, the subject won’t believe for a sufficient normative reason as opposed to a motivating reason that merely happens to correspond to a sufficient normative reason. (IV) simply restates premise (2) in our core argument. So (2) is true.

Prime Time (for the Basing Relation)

153

3.4 Deviance Deferred? Our core argument appears to work. There is a last move that the friend of the Composite View could make that is worth addressing. The strategy involves trying to accommodate the intuition that motivates (2) in other terms. The main version of this strategy begins with the claim that the only thing deviant subjects clearly lack is doxastic justification. The strategists then claim either (i) that something further beyond propositional justification and basing is needed to analyze doxastic justification, or (ii) that doxastic justification cannot be analyzed in terms of propositional justification at all, and that we should instead analyze propositional justification in terms of doxastic justification. Turri (2010) suggests a view of type (ii) by appealing to cases similar to Fortunate Consequent-Affirmer. But he does not even consider the possibility of a prime view. As it stands, we think his argument should be reversed. It is plausible that justified believing just is believing for sufficient normative reasons. It is also, as we have seen, implausible that subjects like Sam really believe for sufficient normative reasons. Since Turri doesn’t even consider the possibility of a prime view, and the prime view captures more than his view, cases like Fortunate ConsequentAffirmer are better seen as inputs to arguments for the prime view rather than for inputs to arguments against the traditional analysis of doxastic justification. What about views of type (i)? Views of this type hold that cases like Fortunate Consequent-Affirmer show that we must add a condition to the standard analysis of doxastic justification. Alan Millar suggests a view of this sort in his Millar (1991). For him, doxastic justification requires the satisfaction of a basing condition, a “rational connection” condition (i.e., that the basis does provide a reason for the based belief), a competence condition, and also that, if the basis is a believed proposition, this belief is “legitimately taken for granted”.17 Millar’s rationale for the competence condition was to solve what he called the “problem of the accidental rational connection”, a problem generated by cases like Fortunate Consequent-Affirmer. The reason to prefer our view to this view is similar to the reason to prefer our view to Turri’s view. Millar’s view fails to capture the compelling idea, motivated by the more general Sensitivity Platitude, that when one successfully bases one’s belief on a normative reason, one is sensitive to the normative relation between that reason and one’s belief. It also fails to capture the compelling idea, motivated by the more general Explanation Platitude, that when one successfully bases one’s belief on a normative reason, the fact that one has a normative reason is what explains why one believes as one does. Moreover, the view adds unnecessary complexity to the standard analysis of doxastic justification.

154

Errol Lord and Kurt Sylvan

Hence, prime views are to be preferred in virtue of their simplicity, explanatory power, and ability to aid analyses like the standard analysis of doxastic justification in terms of propositional justification and basing.

4 Developing a Prime Alternative The Composite View is, as we have seen, implausible. While defenders of this view might gerrymander it further, we suspect that any extensionally adequate version of it will be less plausible than an adequately developed prime view. It is a further step from here to any specific prime view, however. In this section, we will explain and motivate a particular prime view. We start with some constraints on a satisfactory prime view:    I. Transparency-Bad Case: The view must explain why having motivating reasons for belief feels like believing for normative reasons from the first-person perspective, given the bigger gap between the two on prime views.  II. Transparency-Good Case: The view must explain the entailment from believing for normative reasons to having considerations that correspond to these normative reasons as one’s motivating reasons for belief, given the bigger gap between the two phenomena on prime views. III. No Deviance: The view must rule out all relevant deviant cases and provide an illuminating explanation of what is going on in them. IV. Explain the Platitudes: The view should explain why Sensitivity to Normative Reasons and Explanation by Normative Reasons are true. We will argue that our account provides an attractive way to meet these constraints. 4.1 The View (and How It Accounts for III and IV) We will introduce the core elements of our view by way of showing how it can explain our two platitudes—that is, Sensitivity to Normative Reasons and Explanation by Normative Reasons—and solve problems of deviance. To warm up to this, focus again on the fortunate consequent affirmer. The consideration that is his motivating reason happens to line up with a sufficient normative reason. But there is a deviant connection between the fact that his motivating reason is that Terry’s car is in the driveway and the fact that the corresponding state of affairs provides a good reason to believe that Terry took the bus. The trick is to find out what the right connection involves. In order for there to be the right kind of connection, we think one must believe q in virtue of the fact that p is a sufficient reason to believe q,

Prime Time (for the Basing Relation)

155

where this “in virtue of” is understood in a broad sense consistent with a certain causal reading to be unpacked later. More officially: Belief for Sufficient Normative Reasons (BSNR): What it is for S to believe q for a sufficient normative reason p that S possesses is for S to believe q in virtue of the fact that p is a sufficient reason to believe q that S possesses.18 BSNR contains two new pieces of ideology that must be explained. First, the in virtue of relation in BSNR is a causal in virtue of relation. Here are some other examples of this relation: the building collapsed in virtue of the heat (not the force of the collision), and the laborer died in virtue of starvation (not the bad taste of the poor food he was able to afford). One might worry that BSNR’s appeal to this relation renders it unhelpful: it merely gives a name to what we need to solve deviance problems. But we have a particular view of what the in virtue of relation here involves.19 It is a dispositional view.20 Before we state the view, we should emphasize that like many metaphysicians,21 we do not take dispositions to be subject to a conditional analysis. This includes dispositions that involve conditional expressions, like dispositions of the form being disposed to  if p, which will figure in our analysis. That said, we think that very often there is a match between the possession of a particular disposition and the truth of corresponding conditional claims. Because of this, we will often use counterfactuals as a way to test whether someone has a disposition. As we’ve learned from the failure of conditional analyses, there isn’t always a match between the possession of a disposition and the truth of some conditional. Thus, this test isn’t infallible. It is, we claim, still helpful.22 With that disclaimer in mind, here is our way of unpacking the relevant in virtue of claim: In Virtue Of: What it is for S to believe q in virtue of the fact that S has a sufficient normative reason p is for S to be disposed to revise her belief that q if p ceases to be a sufficient normative reason S has to believe q. This yields an unpacked version of BSNR: BSNR-Unpacked: What it is for S to believe q for a sufficient normative reason p that S possesses is for S to be disposed to revise her belief if p ceases to be a sufficient normative reason S has to believe q.23 The second piece of new ideology is possession. The idea here is that some reasons are poised to play certain normative roles only when they

156

Errol Lord and Kurt Sylvan

are in one’s epistemic ken.24 For example, the fact that Anne is smiling is a reason to believe she is amused. It is not a reason I possess, however, until it is in within my epistemic ken. To be in my epistemic ken, I have to have some type of epistemic access to that fact. What this access amounts to is extremely controversial. Popular candidates include knowing the fact, merely believing the fact, and being in a position to know the fact.25 A reason lacks the normative power to justify my beliefs when it is not in my epistemic ken (we will return to possession in the next subsection). It is essential to avoiding problems of deviance that the normative property is what is doing the causal work. It cannot just be the reasongiving fact that does the causal work. This was the problem with the fortunate consequent affirmer. The fact that is the normative reason—the fact that Terry’s car is in the driveway—plays a causal role in the formation of Sam’s belief that Terry took the bus. But it is not the right role. This is because in Sam’s case, the normative property—being a sufficient reason to believe she took the bus—doesn’t do the causal work. Now, some might balk at talk of causally efficacious normativity. But we think that when the relevant causal claims are explained dispositionally, such talk needn’t be mysterious—not more mysterious than other normative talk, at any rate. If there are such properties as the property of being a sufficient reason to , we see no reason why a person couldn’t be sensitive in their doxastic or other responses to the instantiation of these properties via their having a disposition. But plausibly, one way for some property instance to cause some response just is for the instantiation of that property to be the triggering condition for a disposition to have that response. (If one rejects this as a way for causation to proceed, then we are happy to drop talk of causal efficacy in favor of talk of dispositional sensitivity.) Of course, if one understands the triggering of a disposition by a property instantiation as a kind of physical “pushing around”, then it would indeed be mysterious. But we shouldn’t understand all triggering of dispositions as a kind of physical “pushing around”. No one should balk, for example, at the idea that some people can be sensitive to logical relations like entailment. But what is it to be sensitive to these relations? Plausibly, it is to be disposed to form beliefs in certain ways only on the condition that these logical relations are in fact instantiated. We think it is obvious that people can have such dispositions. But if they can, then certain abstract properties can serve as triggering conditions of dispositions. While the condition to which we are appealing is more mysterious because it is normative rather than logical, it doesn’t follow that the relation that it bears to the response is more mysterious. We don’t see why it should be more mysterious to view the instantiation of a normative property as a triggering condition for the manifestation of some disposition

Prime Time (for the Basing Relation)

157

than it should be to view the instantiation of a logical property as a triggering condition for the manifestation of some disposition. Of course, one might ask: why bother appealing to the normative properties rather than the logical properties? In the epistemological case, won’t certain logical properties just subvene the relevant normative properties? This response is forceful in the case of deductive reasoning. But when one turns from deductive to ampliative reasoning, one cannot plausibly avoid appealing to some epistemically normative properties unless one views the probabilistic relations in play as themselves logical a la Keynes (1921). Consider, for instance, someone who is sensitive to the probabilistic relation that underwrites the move from the perceptual appearance as of p in a case without defeaters to p. The relevant probabilistic relation is not a logical relation, barring some Keynesian account of epistemic probability.26 Nor is it a relation of reliable indication if one agrees that it is present even in skeptical scenarios like the demon world. It is rather a relation of evidential probability. But that relation is epistemically normative in precisely the way that the relation of being a sufficient epistemic reason to believe is epistemically normative. So, if one wants a general account of the kind of sensitivity that underwrites good reasoning, one will have to appeal to epistemically normative properties. With those clarifications in mind, we think that it is eminently plausible that all the deviance cases that we want to explain are cases where the reasons are doing causal work but in the wrong way. In all these cases, the reason-giving facts are playing a causal role, but the relevant normative properties of these facts are causally inert, in the minimal sense of “causally” discussed earlier. Our account solves the deviance problem by picking out the feature of the reason-giving facts that must do the causal work in order for it to be of the right type. That feature is the normative property. What else could it be? That this suggestion is on the right track is further confirmed by the elegant explanations our account can give of Sensitivity to Normative Reasons and Explanation by Normative Reasons. Consider the former. Our account entails that in order to believe q for a normative reason p, one’s belief must be held in virtue of the fact that p is a reason to believe q. Since our causal in virtue of relation was spelled out dispositionally, it is plausible that this by itself entails that when one believes for a normative reason, one is sensitive to the normative reason as such. Similar things can be said for Explanation by Normative Reasons. When you believe q for a normative reason p, your belief that p is causally sustained by the fact that p is a normative reason to believe q. It is very plausible that when such causal relations exist, we can provide a correct causal explanation of the fact that you believe that q in terms of the fact that p is a normative reason to believe q. This is sufficient to capture Explanation by Normative Reasons.

158

Errol Lord and Kurt Sylvan

4.2 Accounting for I and II Let’s turn to consider how I and II can be explained:  I. Transparency-Bad Case: The view must explain why having motivating reasons for belief feels like believing for normative reasons from the first-person perspective, given the greater gap between the two on prime views. II. Transparency-Good Case: The view must explain the entailment from believing for normative reasons to having considerations that correspond to these normative reasons as one’s motivating reasons for belief, given the larger gap between the two phenomena on prime views. These constraints flow from two ideas that one might have taken to support the Composite View. First, from the first-person perspective, the bad case seems just like the good case. When we believe q on the basis of p, we “see” p as a normative reason to believe q. Second, when we believe for normative reasons, we seem to endorse the reasons for which we believe in the same kind of way we do when we have merely motivating reasons for belief. These ideas put pressure on theorists like us who claim that believing for normative reasons is a distinctive achievement, not merely a special case of having propositions as motivating reasons for belief. In order to explain these facts, we need to say something about how we understand motivating reasons for belief. As we have stressed throughout, our main goal is to understand believing for normative reasons. Accordingly, we will be brief in discussing our view of having motivating reasons for belief, and proceed to show how it helps account for I and II.27 The motivating reasons for which one believes are one’s rationale for so believing: they render one’s belief intelligible. We think the only way for a consideration to play this role is if one treats that consideration as a normative reason to have the belief, where such treating needn’t be understood as an explicit mental state but could be constituted by an implicit attitude.28 In order for a consideration to be one’s rationale for believing something, one has to take it that the consideration has some positive bearing on the truth of the proposition believed. The following account of motivating reasons for belief can be extracted from these core ideas: Motivating Reasons for Belief: What it is for p to be S’s motivating reason for believing q is for S to believe q in virtue of the fact that S treats p as a normative reason to believe q. Once again, the in virtue of relation is a causal in virtue of relation, and more specifically a dispositional in virtue of. So, we can unpack the story as follows:

Prime Time (for the Basing Relation)

159

Motivating Reasons for Belief—Unpacked: What it is for p to be S’s motivating reason for believing q is for S to be disposed to revise her belief that q if she ceases to treat p as a normative reason to believe q. We will say more later about the notion of treating something as a normative reason. For now, however, it is clear enough how to capture our two desiderata. Take Transparency-Bad Case first. It is essential to our account of the bad case that when one has a consideration as one’s motivating reason for belief, one treats that consideration as a normative reason in favor of one’s belief. Thus, whenever one has a consideration as one’s motivating reason for belief, it will in a natural sense seem from the first-person perspective like one is believing for a normative reason. This is why it seems like one is believing for normative reasons when one has a motivating reason for belief, even if it corresponds to no normative reason. Transparency-Good Case is less trivial. The key to explaining it lies in the nature of possessing a reason. In order to base one’s belief in q on a normative reason given by the fact that p, one must possess the fact that p as a reason to believe q. Moreover, we think that in order to possess the fact that p as a reason to believe q, one must treat the corresponding proposition, p, as a reason to believe q. Finally, one needs to treat this proposition as a reason to believe q in virtue of the fact that the fact that p is a reason to believe q. A full defense of these claims would take us too far afield, but here is a sketch. First, a brief defense of the claim that in order to believe q for a normative reason given by the fact that p, one must possess the fact that p as a normative reason to believe q. This is pretty innocuous in this context, since most epistemologists embrace a version of the thought. The driving thought is that only facts within one’s ken can be normative reasons for which one believes.29 If one is in the dark about the fact that p, one can’t base a doxastic attitude on the fact that p. So one can’t base one’s belief on the fact that p if one fails to possess this fact as a reason.30 Now, possessing a normative reason is not just a matter of having access to the fact that gives the normative reason. There are cases where it is plausible that even though the fact that p is in one’s epistemic ken and this fact is a normative reason to believe q, one does not have this fact as a normative reason to believe q, in virtue of lacking appropriate dispositional sensitivity to the normative relation between the fact and the belief.31 A variant of our case with Sam and Terry brings out why this is the case. Suppose Sam knows that Terry’s car is in the driveway but is reasonably unsure that this fact is a reason to believe that Terry is at home: it is Wednesday and there were rumors that Terry’s carpool was being discontinued. Sam might thus withhold judgment whether the presence

160

Errol Lord and Kurt Sylvan

of the car in the driveway is a reason to believe that Terry is at home. Suppose, though, that the carpool has not been discontinued. So, the fact that Terry’s car is in the driveway actually is a good reason to believe she is at home. Here Sam knows the fact that is the normative reason, but does not possess it as a normative reason to believe Terry is at home. What’s missing in this case and cases like it, we claim, is that Sam doesn’t treat the fact that Terry’s car is in the driveway as a reason to believe that Terry is at home. This is why he fails to possess that fact as a reason to believe Terry is at home. Finally, we will (briefly) defend the claim that in order to possess p as a normative reason to believe q, one must treat p as a normative reason to believe q in virtue of p’s being a normative reason to believe q. This is plausible because one can know a fact and treat it as a reason to believe q, but thereby manifest a disposition to treat it as a reason in a confused way (and only in a confused way). Think of a consequent affirmer who is inveterately disposed to treat the fact that Terry’s car is in the driveway as a sufficient normative reason to believe he took the bus only in the consequent-affirming way. If Terry cannot treat this consideration as a reason in any other way, it is hard to see how he can possess the relevant normative reason. After all, it is plausible that someone so confused lacks even propositional justification. If all this is correct, then we can account for Transparency-Good Case. One will always treat p as a reason to believe q when one believes q for the normative reason p, because in order to believe q for the normative reason p, one must possess q as a normative reason to believe p, and in order to do that one must treat p as a normative reason to believe q. Thus, one will always endorse p in the right way when one believes q for a normative reason p. Moreover, if one is sensitive to p’s normative relationship to q, then one will be sensitive to whether one possesses p as a reason to believe q. Since treating is a necessary condition on reason possession, it follows that when one believes q for a normative reason p, one’s belief that q will be sensitive (in the dispositional sense elucidated earlier) to whether one treats p as a reason to believe q. But that is just what is required to have p as a rationale for q. So, when one believes q for a normative reason p, one has p as a rationale for q. 4.3 More on Treating Our story appealed to the notion of treating something as a reason.32 Although we think that this is a pretheoretically intelligible notion, one might object that our account is unsatisfying until more is said about this notion.33 We will say more about this notion, but we want to stress again that our focus in this chapter is on believing for normative reasons and its relation to believing for reasons simpliciter, and not directly on believing

Prime Time (for the Basing Relation)

161

for reasons simpliciter. Although our project may indirectly reveal important constraints on a theory of believing for reasons simpliciter, it would not undermine our project to take the latter notion for granted and to gloss treating something as a normative reason in terms of that notion. While one might hope to analyze believing for reasons simpliciter in terms of treating, doing so is not critical for the success of our project here.34 What, then, is it to treat p like a normative reason to ϕ? To treat p like a normative reason to ϕ should not, we think, be understood in terms of beliefs that are explicitly about normative reasons (e.g., a belief with the content

) or in terms of any other psychological state that would require possession of the concept of a normative reason. Some and perhaps many people who are capable of reasoning (e.g., bright young children) do not have the sophisticated concept of a normative reason. It is not even clear that they must have the simpler concept of a reason to be capable of reasoning. Nonetheless, these people can—and necessarily do, if they are engaging in reasoning—treat certain considerations as normative reasons of various kinds. A better way to think about treating p as a normative reason to ϕ is dispositionally. A simple dispositional view would say that to treat p as a normative reason of some kind to ϕ is to be disposed to think or act in the ways that would be correct if p were in fact a normative reason of that kind to ϕ. One might, however, naturally worry in this context that such a disposition could be deviant in a way that would prevent it from illuminating the nature of reasoning. Accordingly, one might worry that having the disposition just mentioned cannot be sufficient for treating something as a normative reason. To get a firmer grip on this worry, let’s consider the more specific dispositions induced by the general disposition that the simple view claims is sufficient for treating. And for simplicity, let’s focus on the more specific dispositions that one will have when one treats p as a sufficient reason to believe that q. The obvious more specific disposition is the disposition to form the belief that q upon forming the belief that p. Is it possible to have this disposition without treating p as a reason to believe that q? One might argue that it is in the following way. Imagine that someone has suffered a brain lesion that disposes her to believe that q upon believing that p, where it is the brain lesion that causes her to form the belief that q upon believing that p. Is this person treating p as a sufficient reason to believe that q? Not, some might insist, in any sense that will illuminate the nature of reasoning. This person is not, or need not be, reasoning from p to q. One could insist in reply that treating p like a sufficient reason to believe that q involves dispositions to reason in certain ways. And one could clarify that when the simple view said earlier that to treat p as a reason to ϕ, one must do the things that would be correct if p were a

162

Errol Lord and Kurt Sylvan

normative reason to ϕ, these reasoning dispositions were included among the things that would be correct. But now there is a circularity worry: for these dispositions will presumably include the disposition to do what would count as correct reasoning if p were a normative reason to ϕ. The worry should now be clear. If we want to understand treating as a disposition of a kind that will illuminate the nature of reasoning, the disposition can’t be a deviant one of the sort induced by a brain lesion (at least not in the way mentioned earlier). But one might worry that a non-deviant disposition just will be a disposition to do things that would count as correct reasoning if the proposition that one treats like a reason to ϕ were indeed a reason to ϕ. Hence, one might worry that we face a dilemma: if treating is understood dispositionally, it will either be consistent with deviance of a sort that precludes reasoning, or have itself to be understood in terms of correct reasoning, which is our object of analysis. We think that this dilemma is a false one. In particular, we think that there is a way to isolate the dispositions constitutive of treating that doesn’t require them to be understood in terms of correct reasoning. This way involves opting for a kind of virtue epistemology. To avoid the dilemma, one can hold, following Sosa (2015), that the relevant dispositions are competences, that competences are a special case of dispositions of the sort mentioned earlier, but that the relevant concept of competence is (epistemologically) primitive, as virtue epistemologists suggest. We are happy with this result, since we independently want to defend a view we call reasons first virtue epistemology (cf. Section 5.3) and in fact regard this chapter as the prolegomenon to a multiple-paper defense of this view. This view is already appealing in the case of correct reasoning. Correct reasoning involves basing that is apt: basing which not only aligns a subject’s rationale with a good reason, but does so as a manifestation of a competence to form reason-supported beliefs in the presence of the relevant reasons. Such a competence cannot be understood merely as a disposition to form reason-supported beliefs in the presence of the relevant reasons, as a variant on the previous brain lesion case would suggest. But that does not show that the competence should be analyzed in terms of good reasoning (though of course it will entail a competence to reason well). Rather, it just suggests—in support of virtue epistemology—that the concept of competence is an indispensable one in epistemology, one that can’t be eliminated in favor merely of the concept of a success condition and the concept of a disposition. Since there is independent reason to draw that conclusion that has nothing to do with reasoning,35 it is not ad hoc to make this move in this context. If competences in general are a special case of dispositions to meet some success condition but the notion of competence is (epistemologically) primitive, we can answer the earlier dilemma. To avoid the deviance horn of the dilemma, we invoke competences rather than mere dispositions. To avoid the circularity horn of the dilemma, we note that

Prime Time (for the Basing Relation)

163

there is independent reason to regard the notion of competence as (epistemologically) primitive and hence note that a competence to do what is supported by good reasons when and only when those good reasons are present needn’t be analyzed in terms of good reasoning. Now, one might be puzzled by the move we have made. After all, we were originally proposing to rely on the notion of treating something as a normative reason. And we ended up agreeing that this notion can’t be understood merely dispositionally. But isn’t it clearly possible to treat something like a reason in a confused or radically mistaken way? If so, how could competence help us to understand treating? Of course, the competence involved in treating something like a reason is not a competence to reason correctly, since one might treat p like a reason to believe q via affirming the consequent. But this doesn’t show that there is no relevant competence in terms of which treating should be understood.36 Here is a suggestion that follows some ideas in Broome (2013). Every rule of reasoning sets up a standard of correctness, though some rules are bad in a more substantive sense. One minimal kind of competence is a competence to make mental transitions in conformity with a rule. Perhaps that standard is a bad one. Still, it is a standard and that is all that we need to appeal to in order to get talk of competence, in the minimal sense invoked by reliabilist virtue epistemologists, off the ground. With these ideas in hand, we can give a more satisfactory analysis of treating in terms of competence. To treat p like a reason to believe that q is to have a rule-relative competence to think in the ways that would be correct if p were a reason to believe that q of the kind encoded by the relevant rule. Those ways are ones that conform to some rule that permits one to move from p to q in certain conditions, where that rule encodes p’s being a reason to believe that q of a certain kind (e.g., a deductive reason). One must then distinguish between mere rule-relative competences and robust competences, where the latter are competences relative to the correct epistemic rules. Given this discussion, one might worry that all of our invocations of dispositions are doomed because of cases that are similar to the lesion case. We agree. Because of this, we think that a further change to our characterization of believing for normative reasons is required. Our settled view of believing for normative reasons should now be stated as follows: BSNR-Competence: What it is for S to believe q for a sufficient normative reason p that S possesses to believe q is for S’s believing that q to be sustained by her competence to revise her belief if p ceases to be a sufficient normative reason S possesses to believe q. Here the competence invoked is a robust competence, not merely a rulerelative competence. That is, it is a competence that makes one sensitive

164

Errol Lord and Kurt Sylvan

to the actual normative facts. This revision does not, however, have any negative impact on our stories about how our view secures the four central desiderata on a satisfactory prime view. Competences will do all of the requisite work in the same way we explained before.

5 Payoffs and Other Implications We saw in the last section that our specific prime view neatly captures some central desiderata. We also think that going prime in the way that we have recommended has some substantial payoffs and implications. We turn to mention three. 5.1 Saving the Usual Analysis of Doxastic Justification One payoff that we noted indirectly in §3.4 is that our account saves an elegant analysis of doxastic justification in terms of propositional justification—namely: DJ=BPJ: To be doxastically justified in believing that p just is (i) to have propositional justification for believing p, and (ii) to believe p on the basis of this propositional justification.37 DJ=BPJ has recently been attacked by Turri (2010). His argument is anchored in the thought that the specific way in which one “uses” reasons makes a difference to whether one gets doxastic justification from them and that DJ=BPJ is insensitive to this fact, since one can base a belief on certain reasons while using those reasons in the wrong way. Central to his argument are cases that resemble Fortunate Consequent-Affirmer. Here is one of his cases: Spurs Win Mr. Ponens and Mr. F. A. Lacy know that the Spurs will win if they play the Pistons, and know that the Spurs will play the Pistons. Thus, they both possess sufficient reasons to believe that the Spurs will win. They both infer that the Spurs will win from those two propositions (and only those two propositions). But Ponens follows the modus ponens rule, while Lacy follows the modus profusus rule: for any p, q, and r, infer r from p & q. But Spurs Win doesn’t refute DJ=BPJ. One could reject the Composite View, say that Lacy doesn’t believe for a sufficient normative reason, and note that it is natural to understand basing one’s belief on propositional justification as believing for a sufficient normative reason. Lacy merely has a motivating reason that happens, by luck, to correspond to a

Prime Time (for the Basing Relation)

165

normative reason. Yes, one can treat certain considerations as reasons for belief in the wrong way even if they happen to correspond to normative reasons. But this is simply evidence against the Composite View! Once we realize this, the objection to DJ=BPJ is averted, so long as we assume that basing on propositional justifiers is the same thing as believing for sufficient normative reasons (as it is natural to think). 5.2 Experiential Justification and Speckled Hens There is a more specific payoff related to the foregoing one. Our view provides a simple, principled way to protect the thought that experiences can justify belief from the problem of the speckled hen.38 Specifically, it protects the following claims from assault by speckled hens: (*) The fact that one’s visual experience has the content that p partly explains why there is a normative epistemic reason for one to believe p when one has a visual experience with this content. Believing for this reason suffices for doxastically justified belief (absent defeaters).39 It is common to reject (*) because:  I. Perceptual experiences can have extremely precise contents (e.g., about the exact number of speckles on the facing side of a speckled hen), II. Experience-caused beliefs in these extremely precise contents are not even prima facie doxastically justified for ordinary subjects. But in cases that clearly illustrate (II), it is false that subjects hold their specific beliefs because of the specific fact that their experiences represent the truth-makers of these beliefs. They are obviously insensitive to that feature of their experience. Yet according to (PJ), the fact that one’s experience represents these facts is part of what explains why there is a normative epistemic reason for one to take them to obtain. If so, and believing for a normative reason requires sensitivity to that very fact—as our view suggests—(I) and (II) do not undermine (*). Once we understand what it takes to believe for the relevant experiential reasons, these cases don’t harm (*). For (II) does not show that subjects can believe for the relevant experience-given normative reasons. It just shows that subjects might be caused in some way by experience to believe the precise contents. It will be a deviant way, if these subjects really lack the discriminatory abilities to make out what precise facts are represented by experience. If so, these subjects also cannot take these precise facts to obtain because experience represents them as obtaining. If so, they cannot believe for the relevant normative reasons, on our view.

166

Errol Lord and Kurt Sylvan

So, our Prime View also provides a neutral way to protect an attractive reasons-based epistemology of perception from the problem of the speckled hen. 5.3 The Unavoidability of (a Kind of) Competence The foregoing point is, however, something of a double-edged sword for a reasons-based epistemology of perception. The problem of the speckled hen is often wielded by theorists who think that the epistemology of perception embodied in claims like (I) and (II) wrongly excludes competence from playing a role in perceptual justification. They suggest that what is missing in subjects who cannot get justified beliefs about the precise number of speckles is a reliable ability. Adding competence to the explanation of perceptual justification is necessary. Once it is added, these theorists insist, it becomes unclear how experiences as such are playing a significant role in justifying beliefs in these cases. Now, that last claim is not well motivated. The reason why this claim is not well motivated is suggested by our account. Our diagnosis of these cases is compatible with the thought that experience as such is what is giving subjects epistemic reasons for belief, and hence propositional justification. Experience is not shown to have no significance as such by these cases. Indeed, it is also not shown that basing beliefs on experiencegiven reasons cannot give one doxastic justification. What subjects in these cases lack is the competence to coordinate their motivating reasons with the experience-given normative reasons. But the other edge of the sword becomes visible at this point. One might have thought that showing that the whole story about doxastic justification in perceptual cases can be accounted for in terms of basing beliefs on experience-given epistemic reasons would constitute a blow to a virtue-theoretic epistemology of perception. This is not true. If believing for sufficient normative reasons is an achievement, doxastic justification will require the exercise of a competence.

6 Concluding Remarks Let’s take stock. We have argued that believing for sufficient normative reasons is not a mere composite of having certain motivating reasons for belief and those reasons happening to line up with sufficient normative reasons. The Composite View faces a deviance problem illustrated by cases like Fortunate Consequent-Affirmer. Intuitively and in the light of arguments from §3.3, it is implausible that subjects believe for sufficient normative reasons in these cases. The Composite View is mistaken. This conclusion matters, as we have seen. It impacts how we are to understand the relationship between propositional and doxastic justification, the place of reasons in epistemology,

Prime Time (for the Basing Relation)

167

and the place of competence in theories of doxastic justification that appeal to reasons. The failure of the Composite View should not come as a surprise. Believing for sufficient normative reasons is an achievement. Our thinking about other achievements—e.g., perception and intentional action— cannot be correctly modeled by anything like the Composite View. Achievements are not mere sums of success factors and states that can equally well obtain in the good and bad cases. We should not expect the achievement of believing for sufficient normative reasons to differ. This is not to say that a disjunctive account should be accepted. Nor is that what we have recommended. Consider again the perceptual analogy. Almost everyone since Grice (1961) has assumed that seeing that p is not merely a composite of the fact that p and one’s having a visual experience as of p. But not almost everyone is a disjunctivist. They would only be if they added (i) that there is no common feature in the good and the bad case (save for indistinguishability), and (ii) that there is one phenomenon—that is, visual experience—that, like jade, turns out to be disjunctive. Most people believe instead that having a non-factive perceptual experience and seeing that p are just two relations, though they have things in common. Our picture resembles this familiar picture in the philosophy of perception. On our view, there is an achievement (believing for normative reasons) and a lesser doing (believing for reasons). They have something in common. Believing for normative reasons entails believing for reasons, similarly to how seeing that p entails having a visual experience as of p. The key is that this common feature is not a common factor, in the sense of an independent component in both. It cannot be, given the achievement involved in believing for a sufficient normative reason.

Notes * Thanks to David Black, Bob Beddor, Pat Bondy, J. Adam Carter, Daniel Greco, Miriam Schleifer McCormick, Ram Neta, and anonymous referees for several journals. Thanks especially to Joe Cunningham for engaging with this paper in Cunningham (2019). We fail to discuss the details not because we think it is not required but because we don’t want to rob Joe of his achievement by changing the paper he’s responding to before it appears in print. We plan to engage with Joe’s work in the future. 1. We follow Schroeder (2007) in not using “explanatory reason” and “motivating reason” synonymously. On this usage, all motivating reasons are explanatory reasons, but not all explanatory reasons are motivating reasons. In speaking of motivating reasons for belief, we are not assuming doxastic voluntarism. We are using “motivating reasons for belief” as a semi-technical locution to refer to the reasons for which we believe, for lack of a better locution. 2. Despite the prominence of reasons in post-Gettier epistemology (see Shope (1983) for a fine catalogue of prominent theorists using reasons to do epistemology) and the recent resurgence of reasons-based epistemology, many

168

3.

4.

5. 6.

7. 8.

Errol Lord and Kurt Sylvan epistemologists are allergic to reasons talk. But as long as they are congenial to appeals to evidence, there is something to be learned from our discussion. This is because we think that all normative reasons for belief are pieces of good evidence. Thus, by giving an account of believing for normative reasons, we are giving an account believing on the basis of good evidence. We focus on normative reasons instead of evidence because we think that there are reasons to withhold belief about p that needn’t be evidence that p or evidence that ¬p. It is important to recognize this fact in order to solve some problems that face traditional evidentialists (see Schroeder (2012), Schroeder (2015a), and Lord (MS) for more on this). We focus throughout on belief for normative epistemic reasons. We take no stand here on whether there are non-epistemic reasons for doxastic attitudes. We will focus on sufficient normative reasons merely for ease of exposition, and intend our view to generalize to the case of belief for pro tanto normative reasons. When we speak of a sufficient normative reason to believe p, we just mean a reason that is weightier than the reasons to disbelieve and to suspend judgment on p. See Lord and Maguire (2016), Lord (2018a), Schroeder (2012), Schroeder (2015a), and Schroeder (2015b) for discussions of sufficiency in our sense. Notice that we do not define sufficiency in terms of doxastic justification. We view reasons and their comparative weights as more fundamental. What is it for a motivating reason to correspond to a normative reason? On one view, a motivating reason corresponds to a normative reason iff the former is a proposition made true by the concrete state of affairs that is the normative reason. On another view, a motivating reason corresponds to a normative reason iff the motivating reason is literally identical to the normative reason, which might then be viewed either as a true proposition or as a concrete state of affairs. The second kind of view has become popular due to Dancy (2000), who insists that only an ontology of reasons that allows for motivating and normative reasons to be identical in good cases can do justice to the idea that we can act for good reasons. We agree with Dancy that motivating reasons are not mental states, but we are neutral in this paper about whether they are the sorts of things that can be identical to normative reasons (and disagree amongst ourselves); see Mantel (2017) for further discussion. A further note of clarification is in order. In some of the literature in epistemology, a view close to the Composite View is held but phrased not in terms of reasons but rather directly in terms of justification. On this view, believing with doxastic justification is a matter of (i) having propositional justification for believing p constituted by some justifier J and (ii) believing that p on the basis of J (where it is left open whether one of these conditions depends on the other). This view is more general than the view addressed in our paper, since one might allow with Goldman (1999) and Lyons (2009) that there are justifiers that aren’t reasons. The Composite View we are considering would, however, converge with this view if one restricted justifiers to reasons. Compare with the opening pages of Evans (2013). We have in mind Turri (2010), Goldman (2011), and Sosa (2003). Turri will be discussed at length in §5.1 and Sosa in §5.2. Since Goldman bases his conclusion on the same kind of example and discusses the issue more briefly with a focus on Conee and Feldman’s evidentialism, we will not separately discuss him. Compare with Johnston (2004)’s discussion of Conjunctivism about fully successful perception. Again, compare with Johnston (2004).

Prime Time (for the Basing Relation)

169

9. This claim will receive support later when we argue that our analysis of believing for normative reasons is the only one that can save the view that doxastic justification is belief for sufficient normative reasons. Any composite relation will fail to save this natural view of doxastic justification. 10. Two related notes are in order. First, this terminology is obviously inspired by Williamson (2000). But we do not commit to any further Williamsonian ideas here. The core notion of primeness that played a role in Williamson’s argument does not entail unanalyzability or straightforwardly motivate it, as Brueckner (2002) noted in critiquing Williamson’s case from primeness to unanalyzability. So, our prime view of believing for normative reasons will not entail that this achievement is unanalyzable—just that it is a distinctive achievement. A second note bears on the title of the paper and its relation to the rest of the volume. One might have come in thinking that we were interested in a mere special case of the basing relation, not the basing relation, making our title inapt. As you can see now, the first part of this thought is mistaken if believing for normative reasons is prime. Still, one might suggest that we are just interested in a mere disjunct of a disjunctive relation which includes both the bad-basing relation and our prime relation (cf. Lord (2018b)). But if so, the thing that really carves at the joints will be the prime relation, with the bad-basing relation arguably constituting a mere appearance of this relation. If so, we are indeed interested in the joint-carving basing relation (period) in this paper, not just a special case of basing or a mere disjunct of a disjunctive relation. As we will see later, this is an interpretation of our view that we take very seriously, though it is not strictly entailed by our main claim of primeness. 11. A distinct but equally common view is the view that seeing an F and having an experience as of an F are different things. Indeed, only sense-datum theorists who insist that one sees the same thing in the good and the bad case think otherwise; cf. Millar (1996). Virtually everyone who rejects the sense-datum theory accepts this view, and it would serve our analogy equally well. Because more of the relevant classic literature (including Grice) is about object seeing rather than propositional seeing, it is easier to find explicit acceptance of this view than the view that seeing that p and having an experience as of p are different states (though with something in common). But the rationale is the same. If the relation that one bears to p were the same in the good and the bad case, seeing the fact that p would be a mere composite of the experience as of p and the fact that p. But it obviously isn’t, due to veridical hallucination and deviant causal chains. 12. This view is not necessarily a disjunctivist view. A full-blooded disjunctivist view will entail (i) that there is no common feature in the good and the bad cases (other than indistinguishability) and (ii) that there is a single phenomenon— e.g., perceiving that p—that is disjunctive in nature (like, e.g., jade is disjunctive in nature). A two-relations view need not entail these conclusions (and obviously will not entail (ii)). 13 As we will see later, however, it is still true that believing that p for a normative reason entails believing p for a motivating reason that corresponds to that normative reason, just as seeing that p entails having an experience as of p. 14. Cases like this have been in the epistemology literature for a long time. See (Armstrong, 1973, 98), (Swain, 1988, 467), (Millar, 1991, 57), Turri (2010), and (Goldman, 2012, 7) for similar cases. We certainly do not claim to have discovered these cases. We claim only that the correct moral has never been drawn from them.

170

Errol Lord and Kurt Sylvan

15. Some might dispute whether this stipulation can be made. We will undermine the reasons for thinking so at the end of this subsection. 16. We are understanding sensitivity in its intuitive sense—as a competence resident in the thinker. We are not understanding it in a Nozickian sense. Sensitivity in the counterfactual sense may be evidence of sensitivity in the competence sense. But the competence is not constituted by the truth of any such counterfactual, thanks to the conditional fallacy (etc.). 17. See (Millar, 1991, 57–64). 18. Analyses similar to BSNR have been defended in Arpaly and Schroeder (2014), Wedgwood (2006), and Lord (2018b). 19. Arpaly and Schroeder (2014) refuse to analyze the in virtue of relation, to their detriment. Wedgwood (2006)’s analysis is close to ours, although there are differences in the details. These differences aren’t important for our purposes here. 20. Our view bears some similarities to Evans (2013)’s view. However, as he makes clear at the very beginning, his topic is not believing for normative reasons (or what he calls proper basing). Instead, he is just interested in believing for motivating reasons (or what he calls basing). Further, he provides perhaps the clearest statement of the Composite View when he writes ‘I think many epistemologists have failed to keep this distinction in mind, but we can charitably read them as having sought theories of proper basing. Still, this is putting the cart before the horse. Basing is (at most) necessary for proper basing. It is also necessary for improper basing. It’s best, then, to seek first an account of the basing relation—we can worry about propriety later. That, anyway, is how I shall approach the problem.’ We obviously disagree that this is how the issues should be tackled. Still, Evans’ view is a close cousin of the view of believing for motivating reasons we’ll discuss in the next subsection. 21. See, e.g., Bird (1998), Bird (2007), Martin (1994), and Molnar (2003), to name three among many others. 22. These points are crucial for understanding how our view avoids a particular kind of deviant causal chain case. This kind of case involves the manipulation of someone across modal space. To use a case provided by an anonymous referee, imagine an angel causes one to believe the cat is hungry whenever its meowing is a normative reason to believe that it’s hungry. The angel makes it the case that one wouldn’t have the belief that the cat is hungry were the fact that the cat is meowing not a normative reason to believe it’s hungry. If one had a conditional view of dispositions, this might be sufficient for having the disposition that we appeal to in our analysis later. We deny, however, that one does have the relevant disposition in this case. This is simply one of those cases where the counterfactual test fails. There is a good explanation for why it fails, though; namely, the fact that the angel is manipulating one. We thank an anonymous referee for pushing us to make this explicit. 23. While this statement of the view will work for present purposes, we will need to revise the view once more before this section is out. This will be in response to a worry about dispositions that we think is best saved for our discussion in the next subsection. 24. To be clear, we are just claiming that being in one’s epistemic ken is necessary for possession. It is not sufficient, as we show in §4.2. 25. These three views are defended by Williamson (2000), Schroeder (2009, 2011), and Gibbons (2006) and Lord (2018b) respectively. 26. Certain classical foundationalists think that skepticism is unavoidable without an account of epistemic probability along the lines of Keynes (1921); cf. Fumerton (1995). But this is a rare position, plausible only relative to an extreme variety of epistemological internalism.

Prime Time (for the Basing Relation)

171

27. For more on EL’s views on this (which differ somewhat from the views expressed here), see (Lord, 2018b, ch. 6). 28. See Sylvan (2015) for an appeal to treating in this sense. 29. Since ‘ken’ can be understood externalistically, this thought is not the property of internalists. Cf. Gibbons (2006), Lord (2014, 2018b). 30. For similar arguments, see Gibbons (2001), Hyman (2006), and Hornsby (2007). 31. We here assume with many writers in meta-ethics that the fundamental notion of a normative reason is that of an objective normative reason, which is objective in two senses: it is a fact to which one may lack access, and the reason-relation it bears to one’s attitude may also be one that one fails to appreciate. It is nearly uncontroversial outside epistemology that there are such reasons, even among ‘reasons internalists’ in the philosophy of practical reason like Williams (1981). And this kind of objectivity is consistent with a range of subjectivist views about what makes a fact stand in that relation. A fact might, for example, help to explain why acting in some way would satisfy one’s desires, and hence be ‘desire-dependent’ in a broad sense. But this fact would still give an objective reason in the sense at issue here, since one might fail to appreciate how this fact helps to explain why acting in that way would satisfy one’s desires, as well as simply failing to realize that this fact is a fact. 32. See Schossler (2012) and Sylvan (2015) for more on treating something as a normative reason. For the related notion of conceiving of something as a reason, see Lord (2018b). 33. Thanks to an anonymous reviewer for pressing us to say more here. 34. Each of us has pursued a hypothesis along these lines in other work. See especially Lord (2018b) and Sylvan (2016). 35. (Sosa, 2015, 95) defends this conclusion with respect to the competences constitutive of animal knowledge, which is not essentially reason-based. 36. There is a direct analogy with Sosa (2015), who analyzes intentional action in terms of aptness and thus competence. Just like we will do in the next paragraph, he appeals to a thin standard of correctness (viz., the intention’s success condition) to define the relevant notion of competence. 37. See (Turri, 2010, 313–314) for a list of people who accept this view, including Alston, Cohen, Conee and Feldman, Swain, and Pollock and Cruz. 38. Sosa (2003) pressed the problem against classical foundationalism a la Fumerton (1995) and Bonjour and Sosa (2003). But Pace (2010) rightly notes that the problem generalizes to less radical views, like Pryor (2000)’s nonclassical foundationalism. We have them in mind, though our conclusions extend to classical views. 39. Note that (PJ) doesn’t claim that all possessed reasons for perceptual beliefs are experience-given. It just says that when an experience is present, the experience plays a reason-giving role.

References Armstrong, D. (1973). Belief, Truth, and Knowledge. Cambridge University Press. Arpaly, N. & Schroeder, T. (2014). In Praise of Desire. Oxford University Press. Audi, R. (1993). The Structure of Justification. Cambridge University Press. Bird, A. (1998). Dispositions and antidotes. Philosophical Quarterly, 48(191), 227–234. Bird, A. (2007). Nature’s Metaphysics: Laws and Properties. Oxford University Press. Bonjour, L. & Sosa, E. (2003). Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues. Blackwell Publishing.

172

Errol Lord and Kurt Sylvan

Broome, J. (2013). Rationality through Reasoning. Blackwell. Brueckner, A. (2002). Williamson on the primeness of knowing. Analysis, 62(3), 197–202. Carroll, L. (1895). What the tortoise said to Achilles. Mind, 4(14), 278–280. Cunningham, J. J. (2019). Is believing for a normative reason a composite condition? Synthese, 196(9), 3889–3910. Dancy, J. (2000). Practical Reality. Oxford University Press. Evans, I. (2013). The problem of the basing relation. Synthese, 190(14), 2943–2957. Fumerton, R. (1995). Metaepistemology and Skepticism. Rowman & Littlefield. Gibbons, J. (2001). Knowledge in action. Philosophy and Phenomenological Research, 62(3), 579–600. Gibbons, J. (2006). Access externalism. Mind, 115(457), 19–39. Goldman, A. I. (1999). Internalism exposed. Journal of Philosophy, 96(6), 271–293. Goldman, A. I. (2011). Toward a synthesis of reliabilism and evidentialism? or: Evidentialism’s problems, reliabilism’s rescue package. In T. Dougherty (Ed.), Evidentialism and its Discontents. Oxford University Press. Goldman, A. I. (2012). Reliabilism and Contemporary Epistemology. Oxford University Press. Grice, P. (1961). The causal theory of perception. Proceedings of the Aristotelian Society, Supplementary Volume, 35, 121–153. Hornsby, J. (2007). Knowledge in action. In A. Lesit (Ed.), Action in Context. De Gruyter. Hyman, J. (2006). Knowledge and evidence. Mind, 115(460), 891–916. Johnston, M. (2004). The obscure object of hallucination. Philosophical Studies, 120(1–3), 113–183. Keynes, J. M. (1921). A Treatise on Probability. Dover. Lord, E. (2014). The coherent and the rational. Analytic Philosophy, 55(2), 151–175. Lord, E. (2018a). Epistemic reasons, evidence, and defeaters. In D. Star (Ed.), The Oxford Handbook of Reasons and Normativity. Oxford University Press. Lord, E. (2018b). The Importance of Being Rational. Oxford University Press. Lord, E. (MS). Reasons to suspend judgment and the failure of evidentialism. Unpublished Manuscript, University of Pennsylvania. Lord, E. & Maguire, B. (2016). An opinionated guide to the weight of reasons. In E. Lord & B. Maguire (Eds.), Weighing Reasons. Oxford University Press. Lyons, J. (2009). Perception and Basic Beliefs. Oxford University Press. Mantel, S. (2017). Worldly reasons: An ontological inquiry into motivating considerations and normative reasons. Pacific Philosophical Quarterly, 98(S1), 5–28. Martin, C. B. (1994). Dispositions and conditionals. Philosophical Quarterly, 44(174), 1–8. Millar, A. (1991). Reasons and Experience. Oxford University Press. Millar, A. (1996). The idea of experience. Proceedings of the Aristotelian Society, 96, 75–90. Molnar, G. (2003). Powers: A Study in Metaphysics. Oxford University Press. Pace, M. (2010). Foundationally justified beliefs and the problem of the speckled hen. Pacific Philosophical Quarterly, 91(401–441). Pryor, J. (2000). The skeptic and the dogmatist. Noûs, 34(4), 517–549. Schossler, M. (2012). Taking something as a reason for action. Philosophical Perspectives, 41(2), 267–304.

Prime Time (for the Basing Relation)

173

Schroeder, M. (2007). Slaves of the Passions. Oxford University Press. Schroeder, M. (2009). Means-end coherence, stringency, and subjective reasons. Philosophical Studies, 143(2), 223–248. Schroeder, M. (2011). What does it take to ‘have’ a reason? In A. Reisner & A. Steglich-Petersen (Eds.), Reasons for Belief. Cambridge University Press. Schroeder, M. (2012). Stakes, withholding, and pragmatic encroachment on knowledge. Philosophical Studies, 160(2), 165–185. Schroeder, M. (2015a). Knowledge is belief for sufficient (objective and subjective) reason. In T. S. Gendler & J. Hawthorne (Eds.), Oxford Studies in Epistemology, volume 5. Oxford University Press. Schroeder, M. (2015b). What makes reasons sufficient? American Philosophical Quarterly, 52(2). Shope, R. K. (1983). The Analysis of Knowing: A Decade of Research. Princeton, NJ: Princeton University Press. Sosa, E. (2003). Privileged access. In Q. Smith & A. Jokic (Eds.), Consciousness: New Philosophical Perspectives. Oxford University Press. Sosa, E. (2015). Judgment and Agency. Oxford University Press. Swain, M. (1981). Reasons and Knowledge. Cornell University Press. Swain, M. (1988). Alston’s internalistic externalism. Philosophical Perspectives, (2), 461–473. Sylvan, K. (2015). What apparent reasons appear to be. Philosophical Studies, 172(3), 587–606. Sylvan, K. (2016). Epistemic reasons II: Basing. Philosophy Compass, 11(7), 377–389. Turri, J. (2010). On the relationship between propositional and doxastic justification. Philosophy and Phenomenological Research, 80(2), 312–326. Wedgwood, R. (2006). The normative force of reasoning. Noûs, 40(4), 660–686. Williams, B. (1981). Internal and external reasons. In Moral Luck. Oxford University Press. Williamson, T. (2000). Knowledge and its Limits. Oxford University Press.

Part II

Basing and Its Applications  

9

Hermeneutical Injustice as Basing Failure Mona Simion

1 Introduction Epistemic goods are goods like all others: some of them are ubiquitous (perceptual knowledge of the presence of large dry goods in clear sight), some of them are rare (scientific understanding); some are cheaper (knowledge of one’s pains, likes), some come at more significant costs (knowledge of what’s on the other side of the Moon); some are readily accessible to most of us (knowledge how to walk), some are the prerogative of the powerful few (knowledge of state secrets). If epistemic goods are goods like all others, one would expect questions pertaining to their just or unjust distribution to bring little to nothing of interest to the corresponding discussion in ethics and political philosophy: we should expect, for instance, egalitarianism, utilitarianism, and the like to deal with epistemic goods just as smoothly as they would with housing and potatoes. According to defenders of sui generis epistemic injustice, though, this expectation is mistaken: of course, epistemic goods, like all goods, can be more or less evenly distributed—some people have more access to information than others, some people go to better schools than others, etc., and there is a question to be asked as to how we should go about this in an ethically permissible way. Ethics of redistribution should be able to clear these issues for us. However, it is argued, one can identify several distinctively epistemic forms of injustice, whereby the victim is harmed, specifically, in her capacity as a knower, and which afford specialized theorizing (Fricker 2007, 1). Following Miranda Fricker, several philosophers1 have purported to identify ways in which this may happen. The two main candidates in the literature are testimonial and hermeneutical epistemic injustice. Very roughly, the former happens when the speaker, usually a member of a marginalized group, is not given due credibility by the hearer in virtue of her being a member of said group. Since the quality of our testimony is an expression of our epistemic standing, unjust credibility attribution is taken to be tantamount to unjust assignment of epistemic standing.

178

Mona Simion

Hermeneutical epistemic injustice occurs when the interpretive resources available to a community render a person’s experiences unintelligible to her, due to the epistemic marginalization of that person or members of her social group. More recently, doubts have been expressed in the literature about whether testimonial injustice is correctly categorized as non-distributive, and thus sui generis, to begin with. Jennifer Lackey (2007), most notably, convincingly argues that speakers can be harmed in their capacity as knowers even if they are given due credibility, merely in virtue of more privileged epistemic agents being given more credibility than deserved. This suggests that testimonial injustice requires traditional treatment, in line with other types of unjust goods distribution. This chapter makes the corresponding point concerning hermeneutical epistemic injustice (henceforth also HEI, for short). To this effect, it starts by having a closer look at the meaning dynamics involved in the latter. To begin with, it argues that Fricker’s account is too restrictive: hermeneutical epistemic injustice is more ubiquitous than her account allows. That is because, contra Fricker, conceptual ignorance is not necessary for HEI: hermeneutical epistemic injustice essentially involves a failure in concept application rather than in concept possession (Section #2). Section #3 puts forth a novel view on hermeneutical epistemic injustice as basing failure. Further on, I show that, if this view right, HEI is a form of distributive injustice, and affords the corresponding traditional normative theorizing (Section #4). In the last section I conclude.

2 Hermeneutical Injustice Without Ignorance Consider the following case from (Fricker 2007, c.7): SEXUAL HARASSMENT: Carmita Wood, age forty-four, born and raised in the apple orchard region of Lake Cayuga, and the sole support of two of her children, had worked for eight years in Cornell’s department of nuclear physics, advancing from lab assistant to a desk job handling administrative chores. Wood did not know why she had been singled out, or indeed if she had been singled out, but a distinguished professor seemed unable to keep his hands off her. As Wood told the story, the eminent man would jiggle his crotch when he stood near her desk and looked at his mail, or he’d deliberately brush against her breasts while reaching for some papers. One night as the lab workers were leaving their annual Christmas party, he cornered her in the elevator and planted some unwanted kisses on her mouth. After the Christmas party incident, Carmita Wood went out of her way to use the stairs in the lab building in order to avoid a repeat encounter, but the stress of the furtive molestations and her efforts to keep the scientist at a distance while maintaining cordial

Hermeneutical Injustice as Basing Failure

179

relations with his wife, whom she liked, brought on a host of physical symptoms. Wood developed chronic back and neck pains. Her right thumb tingled and grew numb. She requested a transfer to another department, and when it didn’t come through, she quit. She walked out the door and went to Florida for some rest and recuperation. Upon her return she applied for unemployment insurance. When the claims investigator asked why she had left her job after eight years, Wood was at a loss to describe the hateful episodes. She was ashamed and embarrassed. Under prodding—the blank on the form needed to be filled in—she answered that her reasons had been personal. Her claim for unemployment benefits was denied. (2007, 150) Carmita Wood was the victim of sexual harassment; unfortunately, however, she did not know that she was: she was unable to interpret and understand the meaning of her experience. After she reported her experiences, several women present at the meeting in question had a revelation: they had had similar experiences in the past, and they had never told anyone about them. They decided to hold a speak-out in order to break the silence about this: The ‘this’ they were going to break the silence about had no name. ‘Eight of us were sitting in an office of Human Affairs,’ . . . ‘brainstorming about what we were going to write on the posters for our speak-out. We were referring to it as “sexual intimidation,” “sexual coercion,” “sexual exploitation on the job.” None of those names seemed quite right. We wanted something that embraced a whole range of subtle and unsubtle persistent behaviors. Somebody came up with “harassment.” Sexual harassment! Instantly we agreed. (2007, 150) According to Miranda Fricker, hermeneutical epistemic injustice occurs when the victim’s understanding of her own experiences is impaired in virtue of conceptual ignorance, due to the ways meanings are handled by the socially empowered, at the disadvantage of marginalized groups. Carmita Wood did not know she was sexually harassed because she did not have the concept of sexual harassment; that is why she was unable to understand her own experiences. According to Fricker, Wood suffered ‘an acute cognitive disadvantage due to a gap in the community’s hermeneutical resources’: this particular type of aggression had not been yet fully conceptualized at that time, which rendered Wood epistemically impaired. According to Fricker, in turn, this is due to systematic hermeneutical marginalization of women at the time, that is, their exclusion from social meaning production, as it were: there was no concept of sexual harassment for Wood to access and employ, because social power

180

Mona Simion

relations were such that women’s experiences were not at the forefront of social concept building. Importantly, though, Fricker notes that systematic hermeneutic marginalization, although present in the case of Carmita Wood, is not an essential feature of this form of injustice; hermeneutical epistemic injustice can also be a one-off event, whereby the victim is not a member of an often marginalized group. From what I gather, three features are essential to Fricker’s account of hermeneutical epistemic injustice: 1. Conceptual ignorance on the part of the speaker 2. The fact that feature (1) is brought about through a lack of conceptual resources in the relevant community 3. The fact that the lack in question is brought about unjustly through hermeneutical marginalization (which need not be systematic) I believe condition (3), as stated, is too strong. The reason why I am skeptical about (3) is that it is not the case that injustice, in general, needs to be brought about by marginalization of any sort. We can easily imagine (find) cases in which, contrary to expectation, socially powerful groups are, nevertheless, victims of injustice. It is less likely that this will be the case, of course, but that does not make marginalization an essential source for injustice. What is plausibly essential to hermeneutical epistemic injustice is that the relevant form of epistemic failure is unjustly brought about, no matter how this happens, that is, whether as a result of marginalization or not. I will, however, not press further in this direction, but rather work with a version of (3) simplified accordingly. What I will be concerned with in what follows is arguing that both (1) and (2) are too strong: hermeneutical epistemic injustice is more ubiquitous than Fricker’s account makes it out to be. Let us start with (1). Importantly, even if we accept Fricker’s account of the epistemic trouble present in the Wood case, that is, in terms of conceptual ignorance due to lack of available conceptual resources, it does not follow that the latter is necessary for hermeneutical epistemic injustice to occur. Individual hermeneutical failure, that is, failure to interpret and understand one’s experiences, need not be sourced in a large-scale conceptual gap; individual conceptual impoverishment is enough for HEI, insofar as it is brought about by unjust social structures. To see this, consider a modified story of Carmita Wood (involving Carmita*), happening in recent times, when the concept of sexual harassment is both thoroughly theorized and widely covered by legislation: SEXUAL HARASSMENT*: Carmita* Wood lives in a society characterized by unfair distribution of welfare. She does not have the concept of ‘sexual harassment’ because her parents could not afford

Hermeneutical Injustice as Basing Failure

181

to send her to college. As such, when becoming the victim of sexual harassment, Carmita* is not able to either understand what is going on or communicate her experiences to the relevant authorities. Carmita* suffers from hermeneutic ignorance: she is not able to interpret her own experiences due to the fact that she does not possess the concept of sexual harassment. Furthermore, the relevant ignorance is unjustly brought about: her being a member of an impoverished social group results in her having scarce conceptual resources. What SEXUAL HARASSMENT* shows is that (2) is too strong: hermeneutical epistemic injustice need not be caused by gaps in the collective conceptual resource. A more localized failure to conceptualize is enough, insofar as it is unjustly brought about. To see why (1) is too strong, consider a further modification on the Wood case: SEXUAL HARASSMENT**: In this society, sexual harassment is properly theorized, and Carmita** herself is quite competent with the concept. She is disposed to draw proper inferences employing it, she can offer a more or less precise definition, she can give examples, and, as a matter of fact, she applies the concept very reliably. However, she is not infallible: when it comes to friends and family, Carmita**, like most of us, tends to be epistemically vulnerable. Carmita**’s good old friend John, after being psychologically shaken by an ugly divorce, started acting strangely around Carmita** (in similar ways described in the original Fricker case). Although competent with the concept of sexual harassment, Carmita** is incapable of employing it on this particular occasion: conceptual competence does not imply infallibility in conceptualizing. She does not think John is sexually harassing her, she thinks he is just flirting: ‘after all, we are old friends; he’s just been through a rough time lately, I will just not encourage him, and he will stop,’ she thinks. Carmita** fails to interpret her own experience as sexual harassment due to socially unjust treatment—being taken advantage of—by a close friend. Carmita** is a fictional character, of course, but this phenomenon is fairly ubiquitous, and it often is the outcome of socially unjust treatment of the epistemic victim. Most notably, victims of domestic abuse, although competent with the relevant concepts, fail to apply them in the relevant cases: they are incapable to properly interpret and understand their own experiences when abused by their loved ones. Conceptual competence is, on all remotely plausible extant accounts, at most a matter of reliability in concept application, not of infallibility. When failure in concept application is triggered by social injustice, the

182

Mona Simion

victim suffers hermeneutical epistemic harm: due to unjust social mechanisms or behaviors, she is unable to properly interpret and understand her experiences, in spite of her general competence with the relevant concept.

3 Hermeneutical Epistemic Injustice as Basing Failure If what the previous section argued is correct, hermeneutical epistemic injustice does not essentially involve conceptual ignorance, either at the societal or the individual level. Rather, what is essential to these cases is a failure in concept application. Of course, many—if not most—cases of hermeneutical epistemic injustice will involve failure of concept application in virtue of conceptual ignorance. The victim fails to apply the concept of, say, sexual harassment, because she does not possess the concept of sexual harassment. Furthermore, I am confident that Fricker is right: often, the victim does not possess the relevant concept in virtue of a gap in the conceptual repertoire of her social milieu, likely due to poor theorizing in the relevant area, triggered by marginalization of the affected social group. All of this, however, is not essential to the phenomenon: what hermeneutical epistemic injustice lives and dies with is failure in concept application, triggered by unjust social factors. Interestingly enough, note that, if this is right, hermeneutical epistemic injustice is a completely different type of epistemic beast than we thought. So one important question that arises is: what kind of epistemic failure is involved in hermeneutical epistemic injustice? This chapter’s thesis is that, rather than a type of ignorance (i.e. conceptual), HEI is a failure in basing. In turn, I take a failure in basing to be instantiated when the agent has the epistemic resources needed for proper belief formation available to her but fails to form the corresponding belief. In this, a failure in basing is not a mere absence in basing: it is an impermissible absence in basing. More about this later.2 According to Fricker’s account, it is essential to HEI that victims fail to believe that they are undergoing an experience of type T in virtue of being ignorant about the concept “T”, which, in turn, is triggered by the unjustly brought about general societal ignorance concerning “T”. According to the view defended here, the HEI victim has a particular experience of type T, she is propositionally warranted to believe that she is undergoing T, but she fails to form the relevant belief in virtue of unjustly brought about episodic failure in concept application. In order to understand the notion of propositional warrant I favor, and how it does the work here, it will be useful to talk in terms of reasons to believe. That is because the epistemic basing relation is usually understood as the relation that holds between a reason and a belief when the reason is a reason for which the belief is held. In my view, reasons are facts; they can, however, be facts ‘in the world’ (there being a table in front of me) or mere facts about a subject’s

Hermeneutical Injustice as Basing Failure

183

psychology (my having a perception as of a table). Importantly, my propositional warrant is meant to cover available reasons to believe, whether the availability in question is psychological or not. That is, my propositional warrant covers both reasons the believer has to believe that p, and reasons that are available for the believer to believe that p (that is, available in her epistemic environment), even if they are not psychologically available to her. Here is how to understand the distinction I’m talking about: If I see that there is a table in front of me I have a reason to believe that there is a table in front of me. If there is a table in front of me but I’m not paying attention to it, because I’m concerned with thinking about hermeneutical epistemic injustice, there is a reason available for me to believe that there is a table in front of me. If there is a mathematical truth out there that we haven’t yet discovered, there is a reason for me to believe it, but it’s not available to me. My propositional warrant covers the first two types of reasons (reasons had and available extant reasons), but not the latter. I am propositionally warranted to believe that there is a table in front of me if I see that there is a table in front of me. I am also propositionally warranted to believe that there is a table in front of me if there is a table in front of me, even if I’m not paying attention to it. I am not propositionally warranted to believe not-yet-discovered mathematical truths. Here is some theory about the notion of availability at work here: first, we are cognitively limited creatures. There is only so much information we can access: the fact that there is a table in front of me is something that I can easily access. The fact that X, where X is an undiscovered mathematical truth, is not (usually) something I can easily access. As a first approximation, then, my notion of availability will track a psychological ‘can’ for an average cognizer of the sort exemplified. This psychological ‘can’ will be further restricted by features of the (social, physical) environment: we are supposed to check whether there are crocodiles in the lake, but not in the fridge; that’s because we are too limited to check everywhere, and our physical environment is such that they are more likely to be in the lake than in the fridge. We are supposed to read the newspaper on the table in front of us, but not the letter under the doormat.3 That’s because we are limited creatures—we can’t read everything—and our social environment is such that written testimony is more likely to be present in the newspaper on the table than under the doormat. Availability is easy enough availability. HEI is a failure in basing: the HEI victim has a particular experience of type T, she is propositionally warranted to believe that she is undergoing T—that is, there are reasons available for her, or her social circle, to believe that she is undergoing T, but she fails to form the relevant belief in virtue of unjustly-brought-about episodic failure in concept application: she fails to base her beliefs on available reasons to believe. The view straightforwardly handles two out of the three cases discussed earlier. In SEXUAL HARASSMENT**, Carmita** is propositionally

184

Mona Simion

warranted (in this case, she has reasons) to believe she is the victim of sexual harassment. However, she is epistemically impaired in virtue of being unjustly treated by an old friend, and thus fails to form the relevant belief. In SEXUAL HARASSMENT*, Carmita is propositionally warranted (in the sense that there are available reasons for her) to believe she is being sexually harassed; however, in virtue of her unjustly brought about lack in conceptual repertoire, she fails to form the corresponding belief. One worry that naturally arises at this point concerns Fricker’s original SEXUAL HARASSMENT case: is HEI construed as a basing failure not too restrictive still? After all, again, on all minimally plausible accounts, propositional warrant requires some reasonable degree of availability: again, I am not propositionally warranted to believe all mathematical truths; similarly, if there’s a letter in a bottle somewhere in the Atlantic proving you ate my ice cream, it hardly follows that I am propositionally warranted to believe that you ate my ice cream. More than this seems to be needed, even for the weakest form of warrant.4 The relevant warrantmakers need to be, in some sense, available in the epistemic environment of the subject, even if not psychologically present. But this suggests that in Fricker’s favorite cases, where the relevant conceptualization failure is ultimately due to a gap in the conceptual repertoire of the relevant social group, we don’t get hermeneutical epistemic injustice. I want to suggest that SEXUAL HARASSMENT is a case of indirect hermeneutical epistemic injustice. That is, the victim fails epistemically in virtue of an unjust basing failure at the level of her social group. The relevant social group exhibits conceptual ignorance of ‘sexual harassment’. Importantly, though, the ignorance in question is triggered by a failure in basing on available conceptual resources. The social group is conceptually ignorant in spite of (1) available conceptual resources (the concept of ‘sex’ and the concept of ‘harassment’ are, by stipulation, possessed by the social group), and (2) the ubiquitous nature of the phenomenon, which should trigger interest in concept acquisition. In this, the social group’s ignorance is a failure in basing: they are warranted to believe such episodes constitute sexual harassment, but fail to do so. This failure, in turn, is unjustly brought about by lack of interest in the issue. To see the plausibility of this, note that hermeneutical epistemic injustice co-varies with availability of resources. In line with propositional warrant, it is also plausibly bound by an accessibility constraint: it needs to be, as it were, that the social milieu in question had some access to conceptual resources required for the relevant concept production, for HEI to be instantiated. It is unjust that there was no concept of sexual harassment around in Fricker’s original case because the relevant experiences were fairly ubiquitous—sexual harassment was widely spread—and, as the story goes, all that was needed was for the relevant social group to take interest in the matter. The unfortunate Wood event was merely the

Hermeneutical Injustice as Basing Failure

185

trigger for concept generation. Carmita’s social group had all the needed conceptual resources and incentives to fill the relevant gap in its conceptual repertoire. Importantly, note that, had this not been the case, HEI would have plausibly not been instantiated. In contrast to the case of the concept of ‘sexual harassment,’ not having a concept of ‘electron’ in our society’s conceptual repertoire before we even discovered the existence of subatomic particles cannot have generated HEI, no matter what the effects of this conceptual lack might have been on marginalized groups. Plausibly, that is because the resources needed for concept production were not yet available to us. If that is the case, that is, if propositional warrant and HEI share this availability constraint, it is plausible to think that what happens in HEI cases is that, although propositional warrant exists (either psychologically present or merely available in the relevant epistemic environment), the relevant epistemic subjects (unjustly) fail to form the corresponding beliefs.

4 Basing Failure and Distributive Epistemic Justice I have argued so far that hermeneutical epistemic injustice is much more ubiquitous than Fricker’s account would have it. In particular, neither societal- no individual-level conceptual ignorance is necessary for HEI instantiation. Rather, what is essential to HEI is a (unjustly brought about) failure in basing. The victims of HEI fail to interpret and understand their own experiences, in virtue of failing to form the relevant beliefs, in spite of the fact that the corresponding propositions enjoy propositional warrant. In some cases, this happens because the victims lack the relevant concept; in others, although conceptually competent, they fail to apply it. What I am going to do next is 1) go through some accounts of basing on the market and illustrate how HEI works on these accounts, and 2) argue that, no matter your preferred view on the basing relation, if this chapter is right about HEI amounting to basing failure, HEI is nothing but garden-variety distributive injustice. Causal theories of the basing relation hold that for a belief to be based on a reason, the reason must cause the belief in an appropriate way. As such, according to (Moser 1989, 157): S’s believing or assenting to P is based on his justifying propositional reason Q if and only if S’s believing or assenting to P is causally sustained in a nondeviant manner by his believing or assenting to Q, and by his associating P and Q. On this view on basing, then, my account takes HEI to amount to an unjustly brought about rupture5 in the (nondeviant) causal chain. Although the victim had the relevant causal triggers available to her, the chain is interrupted through some unjust intervention.

186

Mona Simion

The main competitor on the market is the doxastic view (e.g. Tolliver 1982). Doxastic theories of the basing relation hold that having an appropriate meta-belief to the effect that a reason is a good reason to hold a belief is sufficient for the belief’s being based on the reason. On this view, failure in basing will amount to ignorance: the victim of HEI fails to have the corresponding meta-belief in virtue of some unjustlybrought-about epistemic situation—be it conceptual ignorance or failure to note the support relation. There are ups and downs to both these views. The causal view excels in capturing the explanatory aspect of believing based on reasons, but fails to capture its normative aspect: when I base my beliefs on a fact, my belief is not merely causally triggered by the fact in question. It causes it in virtue of my treating the fact in question as normative, as a reason to believe (Simion 2019). The doxastic view captures its normative aspect: the believer forms the corresponding belief in virtue of what she takes to be a good support relation. It does so at a very serious over-intellectualization cost, however. In previous work (Simion 2019), I have suggested a novel view of basing, aiming to capture both the explanatory and the normative dimensions, which I take to be jointly necessary and sufficient for basing, but at no cost for unsophisticated cognizers. I construe basing as indicator following. On the view I favor, knowledge is the norm of belief, and normative reasons are facts that indicate norm compliance. In turn, the basing relation consists in following6 indicators: treating a fact as an indicators of norm compliance. On this view, a belief B is based on a reason R iff, in believing B, the subject is treating R as an indicator of compliance with the norm of belief and thereby follows it. The fact that I see a table in front of me is a reason for me to believe that there is a table in front of me because it (pro tanto, prima facie) indicates that my corresponding belief will comply with the norm of belief—it will be knowledgeable. In turn, my belief that there is a table in front of me is based on the fact that I see a table in front of me iff, in so believing, I treat this fact as an indicator that my belief will comply with the norm of belief and thereby follow it. The basing as following indicators view has a very visual and intuitive way to explain what is happening in cases of HEI: although the corresponding indicators are present, the victim is unjustly prevented from following them. Imagine, as it were, that the victim is driving to hermeneutical success, but fails to take the proper turn, in spite of there being an indicator pointing in the relevant direction, because she is unjustly prevented to follow it. Whatever your preferred view on the basing relation may be, one interesting result to note is that, if HEI is a failure in basing, it is but an instance of good old distributive injustice. In cases of HEI, the victim’s access to the relevant goods is unjustly blocked. Just like people that are

Hermeneutical Injustice as Basing Failure

187

unjustly prevented from accessing extant resources, such as food, shelter, education, health, and so on, HEI victims are unjustly prevented from accessing epistemic goods. On the causal view, that is because the causal chain is unjustly interrupted. We find cases like this in general goods maldistribution all the time. On the doxastic view, what prevents the HEI victim’s access to one type of epistemic good—understanding her experiences—is her (unjustly brought about) lack of a different epistemic good (the relevant metabelief). Again, this picture often obtains in the realm of non-epistemic goods: the poor do not have access to health care in virtue of not having money, which is the case in virtue of not having a job, which, in turn, is the case in virtue of not having access to education and so on. You will not be surprised that my favorite account of basing strongly supports the non-sui generis claim too, that is, the claim that HEI is mere garden-variety distributive injustice: for the simplest analogy, just imagine someone being unjustly kept from reaching their destination by being prevented from following the relevant traffic indicators. Or, to use a more thoroughly theorized analogy, think of opportunity unjust distribution: here, the victims are unjustly blocked from accessing routes to their desired goods. In sum: if HEI is a failure in basing, it amounts to unjustly preventing the victim from accessing the relevant (in this case, epistemic) goods; needless to say, cases of unjustly blocked access to goods are but garden variety cases of distributive injustice. HEI is not a sui generis form of injustice, and there is little reason to believe it affords specialist theorizing. Our results in the literature on general distributive injustice will do just fine. I will end by going back to Carmita Wood, because it is quite interesting to see what particular type of distributive injustice is involved in the original Fricker case. Recall that I proposed that what we have here is a case of indirect hermeneutical epistemic injustice: Carmita’s failure to interpret her own experience is due to a basing failure at the level of her social group: the group is warranted to believe experiences of the type described in Wood’s case are instances of sexual harassment (all conceptual resources for generating the corresponding concept are readily available), but it fails to do so due to unjust marginalization of the interests of the affected group. I want to propose that what we have here is a particularly interesting type of distributive injustice: unjust distribution of production means. The equivalent case for non-epistemic goods would be one whereby we have all the needed resources to produce a particular drug that would save many lives in an impoverished part of the world, but we (unjustly) fail to allocate the needed production resources because the interests of the relevant population don’t figure in our top priorities list. As such, just like Carmita Wood, the victims don’t have access to the relevant goods because we don’t care enough to produce them, in spite of ready availability of resources.

188

Mona Simion

5 Conclusion Epistemic goods are goods like all others: some of us have better access to them than others. Straightforwardly, this is the case with things like information and education. Less obviously, they can be unfairly distributed in testimonial exchanges when we fail to give people the credibility they deserve. This chapter has argued that, in line with general distribution of goods, the access of some of us to hermeneutical epistemic goods such as properly interpreting and understanding our experiences is unjustly blocked through basing failure. Acknowledgements: Many thanks to Pat Bondy, Adam Carter, Chris Kelp, Andrew Moon, and Lee Wilson for extensive comments on this chapter.

Notes 1. See e.g. (Anderson 2012), (Daukas 2006, 2011), (Dotson 2011), (Mason 2011), (Medina 2012). 2. Many thanks to Andrew Moon for pressing me on this point. 3. See Goldberg (2017) for an excellent discussion. 4. Although everything I argue here is also perfectly compatible with a view that construes propositional warrant as broad as to comprise all facts, no matter their availability status. 5. ‘Rupture’ might not be the most intuitive way to describe what is going on in cases where the causal chain is not even initiated. For lack of a better way to put it, here and later I will use ‘rupture’ as a technical term that includes initiation failure. 6. For the purposes of this paper, I will not unpack the following relation. See, however, Simion (2019).

References Anderson, E. (2012). Epistemic Justice as a Virtue of Social Institutions. Social Epistemology: A Journal of Knowledge, Culture and Policy, 26(2): 163–173. Daukas, N. (2006). Epistemic Trust and Social Location. Episteme: A Journal of Social Epistemology, 3(1): 109–124. Daukas, N. (2011). A Virtue-Theoretic Approach to Pluralism in Feminist Epistemology. Feminist Epistemology and Philosophy of Science: Power in Knowledge, Heidi E. Grasswick (ed.), Dordrecht: Springer, 45–67. Dotson, K. (2011). Tracking Epistemic Violence, Tracking Practices of Silencing. Hypatia: A Journal of Feminist Philosophy, 26(2): 236–257. Fricker, M. (2007). Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press. Goldberg, S. (2017). Should Have Known. Synthese, 194 (8): 2863–2894. Lackey, J. (2007). Credibility and the Distribution of Epistemic Goods. Manuscript available here: https://lucian.uchicago.edu/blogs/elucidations/files/2017/08/ Credibility-and-the-Distribution-of-Epistemic-Goods.docx. Mason, R. (2011). Two Kinds of Unknowing. Hypatia: A Journal of Feminist Philosophy, 26(2): 294–307.

Hermeneutical Injustice as Basing Failure

189

Medina, J. (2012). Hermeneutical Injustice and Polyphonic Contextualisms: Social Silences and Shared Hermeneutical Responsibilities. Social Epistemology: A Journal of Knowledge, Culture and Policy, 26(2): 201–220. Moser, P. (1989). Knowledge and Evidence, Cambridge, UK: Cambridge University Press. Simion (2019). Epistemic Norms and Epistemic Functions. Manuscript. Tolliver, J. (1982). Basing Beliefs on Reasons. Grazer Philosophische Studien, 15: 149–161.

10 Agency and the Basing Relation Ram Neta

I Introduction If I offer you enough money to believe that Trump is a great president, you might try to do various things to get yourself to hold that belief. But even if you succeed, you will not thereby have voluntarily believed that Trump is a great president: you will only have voluntarily gotten yourself to believe it. To believe is not to choose, even if some beliefs foreseeably result from choices.1 Given that belief is not choice, how are we to understand the way in which we are responsible, or answerable, for our beliefs? How are we to understand the kind of agency that we exercise in belief? The question is not new, and many different answers have been proposed. According to some philosophers, we are responsible for our beliefs in the same way that we are responsible for our front yard: it is our responsibility to see to their proper maintenance.2 According to other philosophers, we are responsible for our beliefs in the same way that we are responsible for our plans: we exercise our will in forming and sustaining them.3 And according to still other philosophers, we are responsible for our beliefs in the same way that we are responsible for our reactive attitudes: we are responsible for accepting the evaluations expressed by them.4 But though philosophers have offered many different answers to the aforestated question, few have challenged the question as resting on a false presupposition. Kieran Setiya has issued just such a challenge. According to Setiya, the kind of agency that we exercise in belief consists of nothing more substantial than the fact that we not only hold beliefs, but we also hold beliefs about the reasons we have to hold those very beliefs. To believe a proposition P on the basis of a reason R is merely to believe a conjunction of the form P, and the fact that R is a good reason for P. There is nothing more to epistemic agency than that. In particular, epistemic agency differs from practical agency in that the former does not involve causing anything. The present chapter offers a response to Setiya’s argument. I formulate those arguments in Section II, show that they are invalid in Section III, challenge their crucial premise in Section IV, and finally argue that

Agency and the Basing Relation

191

Setiya’s appeal to the metaphysical distinction between the static and the dynamic cannot do the work he wants it to do.

II Setiya’s Argument for Deflationism About Epistemic Agency Setiya’s argument is an inference to the best explanation. It begins by noting three features of belief, and then proposes an explanation of those three features. First: suppose I am prone to wishful thinking, and I would continue to believe that I will win the lottery even if I had no evidence. As it happens, I know that the lottery is rigged in my favour and regard this as proof that I will win. Although the belief that I will win is not sustained by my belief that the lottery is rigged and is counterfactually independent of it, that does not prevent me from believing that I will win on the ground that the lottery is rigged, or from having a justified belief that I will win. Asked ‘why do you believe that you will win the lottery?’ I can cite conclusive proof. What more could knowledge demand?5 Setiya generalizes this point from cases in which we recognize ourselves to possess conclusive evidence to cases in which we take ourselves to possess less than conclusive evidence. In short, Setiya says, believing for a reason doesn’t require that our belief be causally sustained by, or counterfactually sensitive to, our having that reason: merely recognizing the reason for our belief as such suffices for it to be a reason for which we believe. Second, it is incoherent to assert ‘p and the fact that q is evidence that p, but I don’t believe that p even partly because I believe that q.’ . . . the paradox rests on the fact that being in a position to assert the first claim entails the falsehood of the second. One cannot believe that p, and that the fact that q is evidence p, without believing that p because one believes that q, in the epistemic sense. . . . For instance, I believe that I was born in Hull and that the fact that my passport says so is evidence for this claim. Do I believe that I was born in Hull because I know that my passport says so? In some sense of ‘because,’ surely not. I did not form this belief by looking at my passport, nor would I revise it if I discovered that my passport says something else. What my passport says is not the first or most important evidence of my place of birth, and it would be pragmatically odd to cite it as the ground of my belief. . . . Still, once we acknowledge these facts, there is no reason to deny that the words in my passport are among the grounds on which I believe that I was born in Hull.6 Third, our knowledge of our own reasons for belief is privileged in the same way that our knowledge of our own beliefs is privileged. If you believe that p, you are in a position to know that you believe that p. Likewise, if you believe that p on the ground that q, you are in a position to know that you believe that p and that the fact that q is evidence that p.

192

Ram Neta

But to believe that p because one believes that q just is to believe that p and that the fact that q is evidence that p. So you are in a position to know that you believe that p because you believe that q. At the same time, the reductive theory explains how your answer to the question ‘Why?’ plays a constitutive role in believing for a reason. When you believe that p, it follows from your beliefs about the evidence that p that you believe it on the corresponding grounds. What accounts for these phenomena if believing for a reason is not a mere conjunction of beliefs? These three facts—the fact that I can believe on grounds to which my belief is not sensitive, that I cannot coherently assert that I have good evidence that p even though it is not my grounds for believing that p, and that I have privileged knowledge of my own reasons for belief—are all straightforwardly explained, Setiya claims, by the following simple account of the epistemic basing relation, which I will henceforth call “Basing as Believing a Conjunction”, or BBC: (BBC) To believe that p on the ground that q is to believe that p and that the fact that q is evidence that p.7 But if BBC is true, then there are two very basic differences between practical agency and so-called “epistemic agency”. First, practical agency involves a causal relation between one’s motives or intentions, on the one hand, and one’s intentional action, on the other—but no such causal relation is necessary for epistemic agency, since the latter sort of agency involves nothing more than believing certain conjunctions of the form P, and the fact that Q is evidence that P. And second, though it may be typically true that we act for reasons that we take to be good reasons for so acting, our taking them to be good reasons for so acting is not merely insufficient, but also unnecessary, for our acting for those reasons. I can be perfectly well aware of why I performed some action (e.g. doing something deliberately to annoy my partner when I find myself annoyed by something she does), even while recognizing that my reason was not a good reason for so acting. Thus, believing a conjunction about our reasons is neither necessary nor sufficient for practical agency, which involves causation, but it is both necessary and sufficient for epistemic agency, which doesn’t involve causation. In this sense, Setiya claims to defend a kind of “deflationism” about epistemic agency: (Deflationism) Epistemic agency does not involve our exercise of a capacity to cause anything—it involves nothing over and above our having certain kinds of belief. This completes Setiya’s argument for Deflationism about epistemic agency. In the remainder of this chapter, I will attempt to undermine both his argument from BBC to Deflationism, and also his argument for BBC.

Agency and the Basing Relation

193

III Setiya’s Inference From BBC to Deflationism, Rebutted Let’s grant, for the sake of argument, that BBC is true. Does Deflationism follow? In this section, I argue that it does not. The inference from BBC to Deflationism may seem plausible. After all, BBC is an identity statement. It says that believing p on the grounds that q is identical to believing the conjunction: p, and the fact that q is evidence that p. If this identity statement is true, then it follows that epistemic agency involves nothing over and above our having certain kinds of belief (viz., beliefs in conjunction of the form just mentioned). And isn’t this just to say that Deflationism is true? No. Deflationism says not only that epistemic agency involves nothing over and above our having certain kinds of belief. Deflationism also says that epistemic agency does not involve the exercise of a capacity to cause anything. But this further claim about what epistemic agency does not involve doesn’t follow from the claim that epistemic agency involves nothing over and above our having certain kinds of belief. For Deflationism is consistent with the claim that our having the relevant kinds of belief involves the exercise of a capacity to cause certain kinds of things. And Setiya says nothing to rule out that latter claim. For all that Setiya argues, our believing conjunctions of the form p, and the fact that q is evidence that p involves the exercise of our capacity to cause certain kinds of things to happen. It is plausible that many of our beliefs are beliefs that we can have only if we exercise our capacity to cause, or to be affected by, certain kinds of thing. When playing tennis, I might believe that this is my most effective underhand return: my having a belief of this kind is possible only if I exercise one of my athletic capacities while having the belief. In the absence of such an exercise, there is nothing for the demonstrative element in my belief to pick out, and so there are no truth-conditions that I believe to obtain. When trying to decide which color to paint my walls, I might believe that this shade of blue is more appealing than that one: my having a belief of this kind is possible only if I exercise some of my perceptual capacities while having the belief. In the absence of such an exercise, there is nothing for the demonstrative elements in my belief to pick out, and so there are no truth-conditions that I believe to obtain. In such cases, it’s a condition of the possibility of my holding a certain kind of belief that I exercise a capacity to cause certain things to occur, or to be affected by certain perceptible objects. The previous cases involve demonstrative reference, and so each such belief involves the exercise of a particular practical or perceptual capacity for such reference. But even some beliefs that do not involve demonstrative reference might nonetheless be such that our having them involves our exercise of a capacity to cause certain kinds of things. Obvious cases of this sort are our beliefs concerning which games are most challenging to play, or which foods are most enjoyable to eat. Even when such beliefs

194

Ram Neta

involve no demonstrative reference, our having them requires our use of the relevant concepts, which in turn requires our exercise of our imaginative, practical, and perceptual capacities. Finally, consider our beliefs concerning which things are right or wrong, better or worse, appropriate or inappropriate, and so on. We can employ those evaluative concepts only if, and only because, those concepts can play some role in our lives. Concepts that can play no such role—e.g., such concepts as (for me at least) witch, upstart, vermin, divine—are concepts that can be mentioned in our beliefs, but not used in them. In other words, we can attribute beliefs involving such concepts to others whom we understand only incompletely, but we cannot form beliefs involving such concepts ourselves. Clearly, the same is true of the concept having evidence: this is a concept that we can use only because the concept can play some role in our lives. And, as I have argued elsewhere,8 that concept does play a role in our lives, guiding our beliefs and our credal states, as well as guiding our responses to the beliefs and credal states of others. But the concept can play such a role only if, and only because, we exercise a capacity to adjust our beliefs and credal states in response to the evidence we have, and we exercise a capacity to evaluate the beliefs and credal states of others by appeal to the evidence they have. Had we no such capacity—were we in no better position to adjust our beliefs to the evidence we have than we are, say, to adjust our beliefs to the number of neutrons in the nearest galaxy—we would not possess, and could not use, the concept evidence; we could, at best, describe agents who do use such a concept. In sum, it is a condition of the possibility of our having beliefs about our evidence as such that we have the concept evidence, and having that concept requires us to have the capacity to adjust our beliefs and credal states to our evidence. The latter capacity is a capacity to cause various changes in our overall mental state to occur. Thus, forming beliefs about evidence requires a capacity to cause various changes to occur. Even if BBC is true, Deflationism doesn’t follow from it, and is anyway plausibly false. In the next section, however, I will argue that BBC is not even true.

IV Setiya’s Argument for BBC, Rebutted In the preceding section, I granted for the sake of argument that BBC is true, and then I showed that Deflationism doesn’t follow. In the present section, I argue that Setiya’s argument for BBC fails: even if we grant that the three explananda that BBC is supposed to explain are all true, BBC still is not the best explanation of them. Let’s recall the explananda: (a) Q can be a reason for which we believe that P even if our belief that P is not counterfactually sensitive to Q.

Agency and the Basing Relation

195

(b) It is incoherent to assert something of the form ‘p and the fact that q is evidence that p, but I don’t believe that p even partly because I believe that q’. (c) Our knowledge of our own reasons for belief is privileged in the same way that our knowledge of our own beliefs is privileged. Some philosophers would challenge one or more of these explananda, but I won’t do so here—not because I take them to be obviously true (indeed, I take (b) to be not quite true), but rather because their truth won’t matter for the challenge I issue here. My challenge to Setiya’s argument for BBC is that, even if (a)—(c) were true, there is still a better explanation of them than BBC. To see what this explanation could be, let’s start by noticing that (c) is an insufficiently general statement concerning our privileged access. For if we have privileged access to our own reasons for belief, then we also have the same sort of privileged access to our own reasons for action. My ability to say why I believe, for example, that I was born in Holon is no different in kind than my ability to say why I decided not to return to Israel. Of course, I am not infallible with respect to either question, but first-person privilege does not entail infallibility. Whatever my firstperson privilege consists in with respect to the question of what I believe and why I believe it, it consists in precisely the same thing with respect to the question of what I intend and why I intend it, or the question of what I am doing and why I am doing it. And BBC doesn’t begin to explain my first-person privilege with respect to these questions about my intentions or my intentional actions. What could explain such first-person privilege is that all of the answers to all of these questions consist in representations that I have, and to which I have privileged access. Not only do my beliefs and intentions consist at least partly in such representations, but also my holding these beliefs or intentions for reasons consist in such representations, and even my performing certain intentional actions for reasons consist in such representations. This may at first seem like a bizarre suggestion: how could my performing certain intentional actions for certain reasons consist in my representing something? What kind of representation could it consist in? To see how the performance of an intentional action for a reason could consist in my representing something, let’s consider an example. I’m looking at a blueprint of a house that has not yet been built, and I’m performing various intentional actions (e.g., putting bricks in various places) in order to create something depicted by the blueprint. In such a case, I’m performing various intentional actions, but what unites all these various actions into a single process of building a house is that they are all done as part of my effort to implement the blueprint. The blueprint is a representation of what I’m trying to create, and my various intentional actions

196

Ram Neta

add up to the act of building a house by virtue of the fact that they are all performed in an effort to make it the case that the representation is accurate. The blueprint is thus what unifies my various actions into a single intentional action of building a house. The example I’ve given is one in which an agential performance has the form (or the particular kind of unity) it has by virtue of that performance being guided by my representation of that very form. But I propose that what goes for the agential performance in this case also goes for any agential performance done for any reason: to think or do or feel anything for a reason is for the reason to guide one’s thinking or doing or feeling, and to guide it by virtue of the agent’s representing this guidance as justifying the thought or deed or feeling at issue. This representation is similar to, but also different in several ways from, the representation involved in what Boghossian (2013) calls the “taking condition” on inference. The taking condition involves the agent representing a particular reason as a good reason for a particular kind of response, whether or not the agent issues that response, and whether or not she issues that response for that reason. But the kind of representation I’m describing here involves a case of token-reflexive reference to a particular guidance relation between R, on the one hand, and the agent’s response to R, on the other, and it refers to that guidance relation under the guise of justifying the agent’s response. So the kind of representation I’m describing here is a representation that an agent cannot have unless the guidance relation to which the representation token-reflexively refers is actual. Without the obtaining of that guidance relation between R and the agent’s response, the representation in question cannot so much as exist—never mind its accuracy. Thus, a necessary condition of the agent’s having the relevant representation is that she has the reason R, she has the attitude or performs the action that is responsive to R, and these two things are explanatorily related in a way to which the agent can refer. Only if all of these necessary conditions obtain can the agent form the representation that constitutes the agent’s believing or intending or doing something for the reason R. How can an agent’s believing or intending or intentionally doing something for a reason R consist in her having this representation? The blueprint example is intended to provide an illustration of how this is possible, but perhaps another illustration will be helpful here. Consider the various explanatory relations that can obtain between someone’s committing a crime and their conviction on the charge of committing that crime. Of course there may be many different explanatory relations running between an agent’s committing a crime, on the one hand, and their conviction on the charge of committing that crime, on the other. But a particular one of these explanatory relations is constituted by the fact that conviction itself is a conviction for committing that very crime. The conviction represents itself as appropriately responsive to the agent’s commission

Agency and the Basing Relation

197

of that crime, and by so representing itself, it unifies a sundry collection of processes, states, and events as together constituting the explanatory relation between the agent’s crime, on the one hand, and the conviction, on the other. It is by virtue of the conviction’s so representing itself in this way that it counts as a conviction for committing that crime. The representation unifies a bunch of processes, states, and events into a single explanatory relation—the relation in virtue of which the conviction is for that crime. More generally, the guidance relation between R, on the one hand, and the agent’s response to R, on the other, is an explanatory relation that is individuated by the agent’s representation of that relation—and in particular, it can be individuated by the agent’s representation of that relation as involving the agent’s appropriate responsiveness to R. There may be lots of different explanatory relations running between R, on the one hand, and the agent’s response, on the other, but the explanatory relation that is identical to the agent’s basing her response on R—that is, having that response for the reason that R—is the explanatory relation that is individuated by the agent’s representing her response as an appropriate response to R. Since the explanatory relation is individuated by that representation, the agent will have the same privileged access to the explanatory relation as she has to the representation that individuates it. Now that we’ve proposed an explanation of (c) that generalizes to intentions and intentional actions, notice that this same explanation can also explain (a) and (b). It can explain (a) because the explanatory relation between R and the agent’s response to R may overdetermine the agent’s response. And explanatorily overdetermined phenomena are typically not counterfactually sensitive to any particular one of their explainers: the same point is true of intentions as well as belief. And it can explain (b) because, when the agent asserts something of the form ‘p and the fact that q is evidence that p, but I don’t believe that p even partly because I believe that q’, the first two conjuncts of her assertion represent her own belief that p as appropriately responsive to her evidence q. But her representation of that belief as an appropriate response to her evidence can unify an otherwise sundry collection of processes, states, and events causally connecting her evidence q and her belief that p, and unify these various things under the category of her having that belief on the basis of, and so because of, q: again, the same point is true of intentions as well as beliefs. Of course, her representation can unify these various things in this way only if these things exist, that is, only if there actually are some processes, events, and states causally connecting her evidence q and her belief that p: but she could assert the first two conjuncts in a single assertion only if there are at least some such processes, events, and states. Thus, the present explanation of the generalized version of (c) also explains (a) and (b), whereas BBC cannot even begin to explain the generalized version of (c). I take this to show that BBC is not the best

198

Ram Neta

explanation of (a)—(c): a still better explanation is that the basing relation always consists in the agent’s token-reflexive representation of a particular response of hers as an appropriate response to some reason.9 I conclude that, even if we grant Setiya that (a)—(c) are all true, his argument for BBC still fails. Nonetheless, putting aside the defects of Setiya’s argument for Deflationism, there may nonetheless seem to be something plausible about Deflationism. Part of the plausibility that Setiya lends to Deflationism comes from the fact that there is a metaphysical contrast between belief, on the one hand, and certain kinds of intentional action, on the other. Setiya tries to make out this contrast in terms of aspect. In the next section, I say why the metaphysical distinction marked by aspect doesn’t do the work that Setiya takes it to do.

V The Metaphysics of Agency Setiya distinguishes conditions as either static or dynamic. Dynamic conditions are those expressed by verbs that admit of a distinction between progressive and perfective aspect, and static conditions are those expressed by verbs that do not admit of such a distinction. The progressive aspect indicates that something is in progress but not yet complete, whereas the perfective aspect indicates that it is complete: thus, “he was walking” is progressive, whereas “he walked” is perfective. The example just given shows that “walking” is a verb that expresses a dynamic condition. In contrast, “believes” and “owns” are verbs that express static conditions, since there is no progressive form of either. In Setiya’s terminology then, believing is a static condition, while walking is a dynamic condition. This is not controversial. But what is controversial is the idea that Setiya finds to be suggested by means of these labels—the idea that dynamic and static conditions differ with respect to the kind of agency they involve.10 We can begin to weaken the force of this suggestion if we notice that, by Setiya’s own definition, intending, desiring and resenting (all conditions involved in motivating our exercises of practical agency) are every bit as static as believing, while tripping, aging, and dying (all conditions that can befall us as organisms, and not in our capacity as practical agents) are every bit as dynamic as walking. The distinction between static and dynamic conditions seems to crosscut the distinction between those conditions involved in our practical agency, and those conditions not so involved. In that case, how should we understand the distinction between the conditions that are involved in our practical agency and conditions that are not so involved? To answer this question, I need to make two preliminary distinctions. First, there is what I will call the distinction between disposition-exercises and other properties. Some properties of an object, for example, its height or location, are not exercises of any disposition the

Agency and the Basing Relation

199

thing has, whereas other properties, for example, its dissolving in water on a particular occasion, or its moving toward the magnet on a particular occasion, are. Now, within the category of disposition-exercises, we can distinguish a subcategory of capacity-exercises: there are exercises of a special kind of disposition, a disposition to achieve some aim or objective. A metallic object may exercise its disposition to move toward a magnet on a particular occasion, but it doesn’t do so in order to achieve any aim or objective. But an object exercises a capacity only when it exercises a disposition to achieve some aim, or objective. And, of course, an object may, on some occasion, exercise a capacity to achieve some aim, even if it fails to achieve that aim. Indeed, our explanatory appeal to capacities is useful precisely because it helps us to understand what’s in common to capacity-exercises that are successful in achieving their aim and capacityexercises that are unsuccessful in achieving their aim. Agency, most broadly conceived, consists in the exercises of a creature’s capacities. On this broad conception, even plants and microorganisms have a kind of agency—we may call it “nutritive” agency, since their capacities have the aim of nourishing the possessor of the exercised capacity. Animals have a more specific kind of agency—we may call it “appetitive” agency, since their capacities have the aim of getting the animal to move around in order to satisfy its appetite. And humans have a still more specific kind of agency—we may call it “rational” agency, since their capacities have the aim of enabling the human to form a coherent picture of itself and its relation to the world, and strive to live in accordance with that picture.

VI Conclusion: Agency as the Exercise of an Organism’s Capacity to Achieve Its Objective Belief is a state, not an act. But agency, at least as it’s been most commonly understood, consists in the exercise of certain sorts of capacities, and not all exercises of those capacities are acts. Indeed, many of them—including not just our beliefs, but also our intentions, our reactive attitudes, our skills, and our evaluations—are states. What distinguishes those states from the many other states we can occupy independently of our agency is that the former are simply the exercises of agentive capacities whereas the latter are not. The metaphysical distinction between states and acts does not mark a distinction between two kinds of agency.

Notes 1. 2. 3. 4. 5.

Williams (1970), Hieronymi (2011). Chrisman (2008). Peacocke (1999). Hieronymi (2006). Setiya (2013, 191).

200

Ram Neta

6. Setiya (2013, 191–192). 7. BBC is one way of specifying what is sometimes called a “Doxastic Theory of Basing”, but it is not the only way: doxastic theorists might take the epistemic basing relation not to involve belief in any particular conjunction, but rather some synchronic or diachronic pattern of beliefs. 8. Neta (2008). 9. I’ve argued in detail for this account of the basing relation in Neta (forth coming). 10. Chrisman (2012) defends this same controversial idea.

Works Cited Boghossian, P. 2013. What Is Inference? Philosophical Studies 169: 1–18. Chrisman, M. 2008. Ought to Believe. Journal of Philosophy 105: 346–370. Chrisman, M. 2012. The Normative Evaluation of Belief and the Aspectual Classification of Belief and Knowledge Attributions. Journal of Philosophy 109: 588–612. Hieronymi, P. 2006. Controlling Attitudes. Pacific Philosophical Quarterly 87: 45–74. Hieronymi, P. 2011. Believing at Will. In D. Hunter, ed., Belief and Agency. Calgary: University of Calgary Press: 149–187. Neta, R. 2008. What Evidence Do You Have? British Journal for the Philosophy of Science 59: 89–119. Neta, R. Forthcoming. The Basing Relation. Philosophical Review. Peacocke, C. 1999. Being Known. Oxford: Oxford University Press. Setiya, K. 2013. Epistemic agency: Some doubts. Philosophical Issues 23: 179–198. Williams, B. 1970. Deciding to Believe. In Problems of the Self. Cambridge: Cambridge University Press, 1981: 136–151.

11 Epistemic Conservatism and the Basing Relation Kevin McCain

It is widely recognized that merely having strong evidence for a belief is not sufficient for justified belief.1 At a minimum one must also believe on the basis of that evidence in order for the belief to be justified. How does one believe on the basis of evidence? The most common answer is that one believes on the basis of some evidence when one believes because of that evidence. That is to say, one believes on the basis of some evidence only when that evidence causes the belief in question. This sort of causal account of the basing relation is both very popular and plausible.2 Another epistemological view that is plausible, but far from popular, is epistemic conservatism. Epistemic conservatism is the view that the mere having of a belief confers some positive epistemic status on the content of that belief. Crudely put, S’s believing that p makes p (defeasibly) justified for S to some extent.3 Those who defend epistemic conservatism often point out its various benefits in dealing with issues like external world skepticism and the justification of memory beliefs as reasons to accept the view.4 Despite these purported benefits, epistemic conservatism has never been widely endorsed. Additionally, there have been several objections to the view.5 In fact, many philosophers are apt to agree with Richard Fumerton’s (2007: 85) assessment that “epistemic conservatism shows all the signs of theft over honest toil.” While I think that epistemic conservatism deserves a lot more credit than it’s typically given, my goal here isn’t to mount a general defense of epistemic conservatism (though I will provide some defense of it along the way). Rather, my goal is to explore a largely unexamined challenge to epistemic conservatism: it apparently conflicts with causal accounts of the basing relation.6,7 After all, epistemic conservatism seems to say that a belief can be justified simply by being held, but the causal view of the basing relation says that a belief can be justified only if it is caused by evidence that justifies it. How can a belief cause itself though? If a belief cannot cause itself, then it seems there is a problem with thinking that a belief that p can provide the sort of evidence that can lead to a justified belief that p. Hence, it seems that one cannot accept both epistemic conservatism and a causal account of the basing relation.

202

Kevin McCain

Of course, one easy response to this apparent inconsistency would be to reject epistemic conservatism. After all, the causal view of basing is about as well respected as epistemic conservatism is disparaged. This easy response would be a mistake though. Initial appearances notwithstanding, when epistemic conservatism is properly understood it is perfectly consistent with a causal account of the basing relation. So, the truth of the causal view of the basing relation isn’t a reason to think that epistemic conservatism is false. Furthermore, when properly understood, epistemic conservatism is a plausible view that shouldn’t be lightly dismissed.

1 Understanding Epistemic Conservatism The first step in reconciling epistemic conservatism and the causal account of the basing relation is to properly understand epistemic conservatism. As noted earlier, epistemic conservatism is the idea that merely having a belief confers some positive epistemic status on that belief’s content. More precisely, the various ways that one might understand epistemic conservatism fit this schema: If S believes p at t, then p has some positive epistemic status for S at t.8 The key question concerning epistemic conservatism is how to understand “some positive epistemic status.” There are a variety of ways that this might be understood. However, here I will limit the focus to just two: the most common way that critics of epistemic conservatism have fleshed out the view and the best way to flesh out epistemic conservatism.9 Let’s begin with the first way. The simplest way to understand epistemic conservatism is to construe “some positive epistemic status” as amounting to enough epistemic support to fully justify belief. This gives us: EC-J: If S believes p at t, then (in the absence of undefeated defeaters) S’s belief that p is justified at t. Although EC-J is a simple way of understanding epistemic conservatism, it immediately runs into problems. The two most prominent objections to epistemic conservatism are devastating for EC-J. The first problem is what might be called the “boost” problem.10 The gist of this problem is that it seems that EC-J allows that someone can boost her justification for believing p simply by forming the belief. In such a case S has evidence that supports p to degree n at t1, and the only thing about S’s epistemic situation that changes at t2 is that she comes to form the belief that p. Yet, it seems that at t2 EC-J is committed to the claim that p is supported to degree n + x (where both “n” and “x” take positive values). The second problem is what may be called the “conversion” problem.11 This is the idea that it could be that believing that p is not justified for S at t1, but

Epistemic Conservatism

203

by simply coming to believe p at t2 it becomes justified for her. To see this objection, consider a situation where S has equally good evidence for two contrary hypotheses, H1 and H2. Since her evidence is equal for these, and she is aware that at most one is true, the justified position for her to take is suspension of judgment. However, if S where to form a belief in H1, say, despite her evidential situation, according to EC-J her belief would be justified—S’s believing H1 would convert it from unjustified to justified. But, of course, that doesn’t seem right.12 Now, in response to the boost problem and the conversion problem, epistemic conservatives might stick to the idea that the relevant “positive epistemic status” is justified belief but restrict the situations in which EC-J is applicable.13 A better way to go would be to flesh out epistemic conservatism differently—in a way that is more in line with how prominent epistemic conservatives such as Roderick Chisholm (1982, 1989), Catherine Elgin (1996, 2014), and William Lycan (1988, 2013) have understood the view.14 These conservatives understand the positive epistemic status conferred by belief in roughly the following way:15 EC: If S believes p at t, then S thereby has minimal positive evidence for p at t.16 At this point the obvious question is: exactly how strong is “minimal positive evidence for p”? It is the amount of evidence that, if undefeated, is sufficient to provide S with some presumption in favor of p, but no more than that. What exactly is “some presumption in favor of p”? Fortunately, this is a question that Chisholm (1977: 8) answers explicitly: h has some presumption in its favor for S = Df Accepting h is more reasonable for S than accepting not-h. Importantly, a proposition’s having some presumption in its favor doesn’t mean that accepting/believing that proposition is justified. Far from it. Presumption delineates a strength of epistemic support that, while positive, is “vanishingly close to zero” (Lycan, 1988: 171). Consequently, p may have some presumption in its favor for S even though the justified doxastic attitude for S to take is withholding judgment (neither believing p nor ~p). Cases where p has some presumption in its favor for S, but withholding is the justified doxastic attitude aren’t hard to imagine. Consider the following sort of situation. You know with certainty that a particular coin is slightly biased toward heads—vanishingly close to fair but yet biased toward heads (i.e. the odds of heads on a fair toss are the slightest amount above .5), and you know with certainty that this coin was just tossed fairly. You can’t see the results of the toss and have no other information. Intuitively, in this situation you should withhold belief concerning whether the coin landed heads or tails. However, “the coin landed heads” has some presumption in its favor for you. If you were

204

Kevin McCain

forced to pick heads or tails, the rational choice would be to pick heads. This is a case where p has some presumption in its favor for you even though believing p isn’t justified. Consideration of the slightly biased coin example helps to illustrate why EC is intuitively plausible. In a situation where S finds herself with the belief that p, she thereby has minimal positive evidence for p. If this is all of the evidence that S has, continuing to believe that p has some presumption in its favor. In the sort of case we’re imagining, S isn’t aware of any additional evidence for p, nor is she aware that she lacks evidence for p—she simply has the belief (e.g., S may simply have not reflected on her evidence, or lack thereof). What is the most reasonable doxastic attitude for S to take here? Supporters of EC can readily admit that it may be most reasonable for S to withhold belief. However, it seems that believing that p is more reasonable for S than disbelieving that p. It would be very strange to say that S finds herself believing that p, but this has absolutely no bearing on what it’s reasonable for her to do. If S were forced to believe that p or believe that ~p in this situation (if withholding were somehow not an option), the fact that she already believes p gives her minimal positive evidence to continue to believe p.17 Admittedly, this evidence is not strong—in fact, it is the minimal amount of evidence that one can have for believing a proposition rather than its denial. But, this minimal amount of evidence is enough for some presumption. Recall that in the slightly biased coin case, the most reasonable thing to do is to withhold belief concerning how the coin landed. Nevertheless, if one had to choose between heads or tails, choosing heads is the most reasonable option of the two because one has a bit of evidence in support of heads that she doesn’t have for tails. The same applies when S believes that p. The classic objections to epistemic conservatism are ineffective against EC. It’s clear that EC isn’t as bad off as an unrestricted version of EC-J when it comes to the boost and conversion objections. If nothing else, EC is better off because p’s having some presumption in its favor for S is significantly weaker than the justification that EC-J says mere belief provides. While this is true it’s not all that comforting to epistemic conservatives. Pointing out that your view isn’t as bad off as some other view doesn’t provide much support for your view. Fortunately, there is a straightforward response to both of these objections available to EC conservatives.18 Both the boost objection and the conversion objection depend on the truth of what Richard Feldman (2014: 296) calls the “additivity of evidence principle”:19 AE If S acquires new evidence that supports P without losing any old evidence or acquiring any defeating evidence for P, then S becomes better justified in believing P.20

Epistemic Conservatism

205

Recall, the problem in the boost objection is that S’s justification for believing that p illegitimately increases when she believes that p. This assumes that the evidence believing that p yields according to epistemic conservatism is simply added to the evidence that S already had for p resulting in increased justification. Similarly, the conversion objection is thought to highlight a problem because a belief that wasn’t justified prior to S’s believing it becomes justified once she believes it. In order for this to be a genuine problem for EC it has to be the case that the minimal positive evidence that believing provides for p can be added to the evidence that S already had for p in a way that increases S’s overall justification for p. There are good reasons for thinking that AE is false though. Feldman argues against AE by appealing to epistemic certainty.21 His idea is that in a case where you are already epistemically certain that p is true any additional evidence will be redundant—it won’t increase your justification for believing that p. Consider a case of epistemic certainty—your belief that you are in pain when you are consciously experiencing an excruciating pain is a good candidate. Plausibly, your belief that you are in pain in this situation is epistemically certain—it can’t get any better justified. If someone were to come along and say to you, “you are in pain,” this wouldn’t increase your justification. Why not? Once you are epistemically certain that you are in pain, your justification for believing this is at the maximum possible level, so adding more evidence won’t make you more justified. There’s no higher level of justification to which you can ascend. Hence, the testimony you receive that you are in pain is simply redundant evidence. Nevertheless, as Feldman (2014: 296) notes, “redundant evidence is evidence, even if it makes no difference in levels of justification.” Since it’s possible that there is redundant evidence, there can be cases where S acquires new evidence that supports p without losing any old evidence or acquiring any defeating evidence for p, and yet S doesn’t become better justified in believing p.22 Thus, AE is false. Redundant evidence doesn’t just arise in cases of epistemic certainty though. You can generate a lot of redundant evidence for yourself right now for beliefs that aren’t epistemically certain for you. Presumably, your belief that you are reading about epistemic conservatism is justified, but not epistemically certain. Call this justified belief of yours (that you are reading about epistemic conservatism) “r.” r is evidence for believing that w (“someone has written about epistemic conservatism”). You are justified in believing r in this case, and it seems that when you think about it you can quickly become justified in believing that r v w by way of disjunction introduction. Given that your justified belief that r is evidence for believing that w, it seems that your new justified belief r v w is also evidence for believing that w. However, in this case it seems intuitive that your epistemic position relative to w has not improved at all. Your justified belief that r v w, though evidence for w, is redundant evidence. Consequently, adding this justified belief to your evidence doesn’t improve

206

Kevin McCain

your justification for w even though you didn’t lose any of your previous evidence for w or gain any defeaters. Hence, redundant evidence isn’t limited to cases of epistemic certainty. And, the possibility of redundant evidence means that AE is false. Since AE is false, the boost objection and the conversion objection fail. It is perfectly consistent for epistemic conservatives to maintain that the minimal positive evidence that S gains by believing p doesn’t add to the evidence that she already has in support of p. In which case coming to believe that p doesn’t give a boost to S’s justification for believing p, nor does it convert an unjustified belief to a justified one. What is more, when p already has some presumption in its favor for S because S believes that p, the evidence that yields this presumption won’t add to the evidence that she later gains in support of p; that is, supporters of EC can reasonably maintain that the minimal positive evidence that S has by way of believing that p is “swamped” by other evidence that she gains.23 An analogy might help illustrate how the presumption that is provided by believing that p can be swamped by other evidence. When S is on trial for a crime there is a presumption of innocence until there is reason to believe otherwise. If during the course of the trial convincing evidence is produced that S couldn’t have committed the crime because she was in another location, the initial presumption of S’s innocence is irrelevant in light of her rock-solid alibi. The presumption of innocence in place at the beginning of the trial is swamped by the evidence presented in S’s favor. Similarly, the minimal positive evidence that S’s believing provides makes it so that p has some presumption in its favor for S. However, after S gains additional evidence in support of p the presumption generated by her believing is swamped. At this point, one might wonder why the truth of EC matters given that the amount of evidence that it says believing confers is so minimal and, on top of that, is swamped by other evidence. There are several reasons why EC is still an important principle. EC allows for beliefs to be regress stoppers (not the certain foundations of classical foundationalism, but a fallible variety that can end the regress of reasons). After all, given EC, beliefs can have some minimal justification without gaining that justification from some further justifier. EC also helps to make sense of the role of background evidence in justification. It is plausible that whenever we reason, background beliefs are playing a significant role in fixing what’s justified. These beliefs need to have at least some presumption in their favor in order to help justify. After all, if they don’t have a positive epistemic status, how can they justify or even constrain what we justifiably come to believe?24 Essentially, EC can accommodate any of the benefits that have been plausibly ascribed to epistemic conservatism. The key point is that although believing that p only provides minimal positive evidence for p, it’s something. The something that minimal positive evidence supplies is enough to give us presumption, which is the inch that in some cases can

Epistemic Conservatism

207

allow us to take a mile. Much more could be said to motivate EC; however, since the present purpose isn’t to provide a full defense of epistemic conservatism, this is sufficient for now. We have good reason to think that EC is how epistemic conservatism should be understood, and we have good reason to think that it is worthwhile to attempt to reconcile epistemic conservatism and the causal view of basing rather than simply deny epistemic conservatism.

2 EC and Causal Basing With a proper understanding of epistemic conservatism in hand it is relatively straightforward to see that the apparent conflict between it and a causal account of the basing relation is only apparent. Causal accounts of the basing relation require that a belief be caused by justifying evidence in order for the belief to be justified. This is thought to pose a problem for epistemic conservatism because “a belief cannot cause itself at a time, just as a fire cannot cause itself at a time” (Frise 2017: 289). As a result, beliefs that epistemic conservatism claims are justified fail to satisfy the causal requirements for basing. Hence, the beliefs that epistemic conservatism claims are justified can’t be so. Although this worry seems challenging if we understand epistemic conservatism along the lines of EC-J, there’s no difficulty for EC. According to EC, when S believes that p she has minimal positive evidence for p. This isn’t enough for S’s belief that p to be justified. If undefeated, S’s minimal positive evidence is only sufficient for S’s believing that p to be more reasonable than her believing that ~p. EC captures the intuitive idea that, in Elgin’s terms, our deliverances are our starting points for epistemology and reasoning more generally. Nowhere does EC say that the minimal positive evidence that is generated by belief is enough to make a belief justified. Since EC doesn’t claim that S’s belief is justified (or unjustified), there’s no conflict between EC and causal accounts of the basing relation. Nonetheless, one might worry that even if we understand epistemic conservatism as EC there is still a conflict with causal accounts of the basing relation. After all, if we are to move beyond our starting points, at least some of these beliefs that have presumption in their favor will have to either transition to fully justified beliefs or provide part of the justification for other fully justified beliefs. In light of this, one might think that we have only pushed the problem back a step without solving it. The minimal positive evidence that S’s believing yields will end up as part of S’s overall evidence, and so on causal accounts of the basing relation, will have to be a cause of S’s justified beliefs. There are two ways to understand the concern here. The first is that S’s belief that p, which has presumption in its favor, will be used to justify some other belief q. The second is that the minimal positive evidence for

208

Kevin McCain

p that S has in virtue of her believing that p will constitute part of her evidence for p when she transitions from having minimal positive evidence for p to a justified belief that p. So, the worry is that EC will conflict with causal accounts of the basing relation either by allowing beliefs that have minimal positive evidence in their favor to justify fully justified beliefs or by allowing that the evidence S has in support of p because of her believing it will have to be a cause of her justified belief that p. Neither worry constitutes a genuine problem though. Let’s consider the first worry. There simply isn’t a problem with S’s belief that p (for which S’s believing that p provides minimal positive evidence) being a cause of some other justified belief, q. In order for there to be a conflict with causal accounts of the basing relation it would have to be that a belief that p couldn’t be a cause of a belief that q. But, clearly this isn’t the case! As long as beliefs can be causes of other beliefs (which they can), S’s belief that p can be a cause of her belief that q. Now, one might worry that there is a problem here because a belief like p, which only has minimal positive evidence in its favor according to EC, cannot be a justifier for q. Importantly, if this were a problem (it isn’t), it wouldn’t be because it conflicts with causal accounts of the basing relation. If this were a problem, it’d be a problem because such a belief can’t provide justification. There’s no reason to think such a belief can’t provide justification though. Admittedly, S’s belief that p doesn’t have much going for it epistemically, but it has enough so that it has minimal positive evidence that could allow its coherence with other beliefs and experiences to provide support for q.25 What about the second worry? In this case the problem is thought to be that the evidence S has in support of p because of her believing it will have to be a cause of her justified belief that p. The idea is that in a case where S has minimal positive evidence for p at t1, when at t2 S has sufficient evidence for her belief that p to be fully justified her minimal positive evidence (her belief at t1) will be part of the evidence for her justified belief at t2. As a result, her belief that p at t1 will have to be a cause of her belief that p at t2. This worry involves a misunderstanding of epistemic conservatism. The epistemic conservative isn’t committed to claiming that when S moves from a belief that p, which has minimal positive evidence in its favor, to a fully justified belief that p her initial minimal positive evidence has to be part of the evidence that justifies her belief. Recall from earlier that AE is false—evidence doesn’t always sum. It’s perfectly consistent to claim that when S has the belief that p at t1 (and it thereby has minimal positive evidence in its favor) and she transitions to a justified belief that p at t2, the minimal positive evidence S started with at t1 is swamped by the justifying evidence that S has for believing that p at t2. That is, the minimal positive evidence that S has in favor of p because of her believing

Epistemic Conservatism

209

that p at t1 isn’t a necessary component of S’s justification for her justified belief that p at t2. As a result, there is no conflict with causal accounts of the basing relation because S’s belief at t1 doesn’t need to be a cause of her belief at t2. The additional evidence that S has gained after t1 is what justifies her belief, and this evidence is what has to cause her belief in order to satisfy causal accounts of the basing relation. And, there is no reason to think that the evidence S gained after t1 can’t cause S’s belief at t2.26 What if S’s belief that p at t1 is a cause of her belief that p at t2 though? After all, it does seem plausible that part of the cause of S’s believing that p at t2 could be her believing that p at t1. One tokening of a mental state can cause another tokening of a mental state—even a tokening of the same type of mental state with the same content. Won’t this amount to S’s belief that p at t1 being part of the basis of S’s belief that p at t2? Yes, but that’s not a problem. Any plausible causal account of the basing relation will allow that some causes of our justified beliefs aren’t part of the evidence necessary for justifying those beliefs. If nothing but the necessary justifying evidence could be part of the basis of a justified belief, we wouldn’t have many (if any) justified beliefs at all. It is very likely that a variety of factors in addition to justifying evidence play causal roles in our having most (perhaps all) of our beliefs. The key consideration for causal accounts of the basing relation is whether the evidence that makes S justified in believing a proposition plays a sufficiently strong causal role in S’s believing that proposition.27 The fact that other things play a causal role is irrelevant. Thus, even if it turns out that S’s belief that p at t1 is a cause of her belief that p at t2, it doesn’t follow that the minimal positive evidence for p that S has because of her believing p at t1 must be part of the justification of S’s belief that p at t2. As a result, S’s belief that p at t1 being a cause of her belief that p at t2 poses no problem for combining epistemic conservatism with a causal view of the basing relation. In light of these considerations it is reasonable to conclude that there is no genuine conflict between epistemic conservatism and causal accounts of the basing relation once epistemic conservatism is properly understood. This is a happy result since both views are independently plausible and have several theoretical benefits.28

Notes 1. Since I take Evidentialism to be true, I use the term “evidence” in this discussion because I understand justification in terms of evidence. However, “good epistemic reasons” can be substituted for “evidence” throughout the discussion without harm. 2. Korcz (2015) notes that causal accounts of the basing relation are most prominent in the philosophical literature. I will not defend a causal account of the basing relation here or go into further detail describing the best way to understand such a view. It is enough for present purposes to have a general

210

3.

4.

5. 6. 7.

8.

9. 10. 11. 12.

13.

14. 15.

Kevin McCain idea of the causal view of basing. For a more precise formulation and defense of a causal account of the basing relation see McCain (2012b) and (2014). It is important to recognize that this is different from other principles that are sometimes referred to as “conservatism.” For instance, phenomenal conservatism (see Huemer (2001)) and dogmatism about perceptual justification (see Pryor (2000)) both hold that, roughly, S’s having a seeming that p, rather than S’s having a belief that p, confers some positive epistemic status on p for S. Plausibly, these views are distinct from epistemic conservatism—proponents of phenomenal conservatism and dogmatism often tend to think they are defending something other than epistemic conservatism. There is some controversy about this, though. Some, such as Hanna (2011), argue that phenomenal conservatism reduces to epistemic conservatism. This, however, is mistaken (see McCain 2012a). See Chisholm (1989), Lycan (1988), McCain (2008), and Sklar (1975) on epistemic conservatism and skepticism. See McGrath (2007), (2016) on epistemic conservatism and memory. See Poston (2014) for a brief summary of several additional benefits of epistemic conservatism. See Christensen (1994), (2000), Feldman (2003), Foley (1983), Fumerton (2007), and Vahid (2004). It’s worth noting that epistemic conservatism is perfectly consistent with doxastic and the most plausible hybrid views of the basing relation. For more on these views see Korcz (2015) and McCain (2016). This objection first came to my attention during a conversation with Andrew Moon and Philip Swenson in 2008. Subsequently, Moon (2012) has suggested that this purported problem prevents epistemic conservatism from helping internalists account for a subject’s knowledge while she sleeps, and Frise (2017) has used this objection to argue that epistemic conservatism can’t account for memory justification. A notable exception to this schema is Podgorski’s (2016) “dynamic conservatism” which is a principle about the epistemic status of initiating mental processes rather than the epistemic status of beliefs at a time or the epistemic status of beliefs’ retention over time. See Vahid (2004) for a nice discussion of several versions of epistemic conservatism proposed in the literature. See Foley (1983) and Huemer (1999). See Christensen (2000), Foley (1983), Feldman (2003), and Huemer (1999). The evidence S has for H2 would still provide some defeat for believing H1 in this case. However, the thought behind this objection is that the support that H1 gets from being believed would be enough to make it so that S’s total evidence justifies her belief that H1 is true. McCain (2008), McGrath (2007), (2016), and Poston (2014) all attempt this strategy. Though Poston (2014) may ultimately opt for the second sort of understanding of epistemic conservatism that we will discuss shortly. In places his discussion seems to move in this direction. Quine and Ullian (1978) seem to endorse a similar view. Chisholm construes epistemic conservatism as applying to things that we accept/believe. Lycan originally applies epistemic conservatism to “spontaneous beliefs.” Subsequently, Lycan (2013) allows that the motivations for epistemic conservatism (what he calls the “principle of credulity”) may also support phenomenal conservatism (see Huemer 2001). Elgin (2014: 245) holds that all “deliverances”—“perceptual inputs, fixed or transient beliefs, passing thoughts, and so forth”—have some minimal evidence in their favor. So, Elgin’s view is perhaps the most expansive version of epistemic conservatism defended in the literature.

Epistemic Conservatism

211

16. It is not uncommon for epistemic conservativism to be construed as a principle of when a belief can be justifiably retained rather than, as it is here, a principle concerning the status of a belief at a particular time (see Harman (1986), McCain (2008), McGrath (2007), and Poston (2014)). This alternative way of understanding epistemic conservatism would yield a principle like the following: If S believes p at t1, then S thereby has minimal positive evidence for continuing to believe p at t2. Although the merits of epistemic conservatism as a principle of the dynamics of belief are worth exploring in their own right, the focus here is on the static EC. Intuitively, it is easier to square a dynamic epistemic conservative principle than a static one with a causal account of the basing relation (see Frise 2017). So, if EC can be shown to be consistent with causal basing, the same should be true of its dynamic counterpart. 17. There are various ways that one might try to ground the evidence provided by S’s believing that p. One way to do this is to maintain, as Sklar (1975) does, that epistemic conservatism arises from our need for cognitive efficiency. This idea could be fleshed out in terms of cognitive efficiency being necessary for best achieving our epistemic goals, and hence providing an epistemic reason rather than merely a pragmatic one for retaining our beliefs. Another approach would be to appeal to self-trust. Foley (2001), Lehrer (1997), and Zagzebski (2012) have all argued that self-trust is a key aspect of our epistemic agency. In fact, Zagzebski (2012: 51) claims “Self-trust is the foundation of what we take rationality to be.” Epistemic conservatives might pick up on this idea and argue that when S finds herself believing that p what it means to trust herself is that her believing itself gives her some evidence in favor of p. 18. Since I am only concerned with defending EC and showing that it is consistent with a causal account of the basing relation, I won’t explore whether this response will work for EC-J as well. 19. Poston (2014: 35) suggests something similar concerning these two objections though his discussion isn’t explicitly in terms of AE or a similar principle. 20. “Better justified” here refers to p being supported to a higher degree. An example using rational credences can help illustrate. Assume that S has evidence e and Pr(p|e) = .9. AE says that if S gains new evidence for p, e*, without losing e or gaining defeating evidence, then Pr(p|e + e*) > .9. So, upon gaining e* S should raise her credence in p, according to AE. 21. See Lord (2018), McGrath (2016), and Poston (2014) for additional reasons to doubt AE. Interestingly, although Lord argues that this sort of principle is false, he claims that redundant evidence can make one better off rationally. 22. This is especially clear in cases where additional evidence doesn’t affect the balance of evidence (e.g., it doesn’t lead to the probability of p being higher or lower), but it does add to the weight of evidence (e.g., it makes probability functions that include p more resilient in the face of additional evidence). For discussion see Joyce (2005) and McCain & Poston (2014), (2017). 23. The point here is somewhat similar to Poston’s (2014) restriction of a principle like EC-J to instances of empty symmetrical evidence. 24. This point is made by Koons (2006), Poston (2014), and Sklar (1975). Koons (2006: 208) goes so far as to claim that “As Sklar and Wittgenstein recognize, all inferential justification takes place against a background of beliefs which the individual is not capable of justifying.” 25. See Lycan (1988) and Van Cleve (2014) for arguments in support of the idea that in order for coherence to increase justification the inputs in the coherent system must have initial credibility. See BonJour (1985) for reasons to think that this initial credibility doesn’t have to be strong enough to make the

212

Kevin McCain

propositions in question more likely true than false in order for coherence to provide justification. 26. It is important to recognize that causal accounts of the basing relation do not require that S’s justifying evidence be the initial cause of her belief. As long as S’s belief is causally sustained by some evidence at the time in question S’s belief is based on that evidence. For more on this see McCain (2012b) and (2014). 27. See McCain (2012b) and (2014) for explanation and defense of this point. 28. Thanks to Pat Bondy, J. Adam Carter, Earl Conee, Bill FitzPatrick, Matt Frise, John Komdat, Jennifer Lackey, Errol Lord, Andrew Moon, Luca Moretti, Tommaso Piazza, Ted Poston, Mike Veber, Ed Wierenga, and audiences at the 2018 Southeastern Epistemology Conference and the 10th Biennial Rochester Graduate Epistemology Conference for helpful comments and discussion.

References BonJour, Laurence. 1985. The structure of empirical knowledge. Cambridge, MA: Harvard University Press. Chisholm, Roderick. 1977. Theory of knowledge, 2nd Edition. Englewood Cliffs, NJ: Prentice Hall. Chisholm, Roderick. 1982. The foundations of knowing. Minneapolis, MN: University of Minnesota Press. Chisholm, Roderick. 1989. Theory of knowledge, 3rd Edition. Englewood Cliffs, NJ: Prentice Hall. Christensen, David. 1994. Conservatism in epistemology. Nous, 28: 69–89. Christensen, David. 2000. Diachronic coherence vs. epistemic impartiality. Philosophical Review, 109: 349–371. Elgin, Catherine. 1996. Considered Judgment. Princeton, NJ: Princeton University Press. Elgin, Catherine. 2014. Non-foundational epistemology: Holism, coherence, and tenability. In Matthias Steup, John Turri, and Ernest Sosa (eds.) Contemporary Debates in Epistemology, 2nd Edition. Malden, MA: Wiley Blackwell, 244–255. Feldman, Richard. 2003. Epistemology. Englewood Cliffs, NJ: Prentice Hall. Feldman, Richard. 2014. Evidence of evidence is evidence. In Jonathan Matheson and Rico Vitz (eds.) The Ethics of Belief. New York: Oxford University Press, 284–300. Foley, Richard. 1983. Epistemic conservatism. Philosophical Studies, 43: 165–182. Foley, Richard. 2001. Intellectual trust in oneself and others. New York, NY: Cambridge University Press. Frise, Matthew. 2017. Internalism and the problem of stored beliefs. Erkenntnis, 82: 285–304. Fumerton, Richard. 2007. Epistemic conservatism: Theft or honest toil? Oxford Studies in Epistemology, 2: 63–86. Hanna, Nathan. 2011. Against phenomenal conservatism. Acta Analytica, 26: 213–221. Harman, Gilbert. 1986. Change in view. Cambridge, MA: MIT Press. Huemer, Michael. 1999. The problem of memory knowledge. Pacific Philosophical Quarterly, 80: 346–357.

Epistemic Conservatism

213

Huemer, Michael. 2001. Skepticism and the veil of perception. Lanham, MD: Rowman & Littlefield. Joyce, James. 2005. How probabilities reflect evidence. Philosophical Perspectives, 19: 153–178. Koons, Jeremy. 2006. Conservatism, basic beliefs, and the diachronic and social nature of epistemic justification. Episteme, 2: 203–218. Korcz, Keith Allen. 2015. The epistemic basing relation. In Edward N. Zalta (ed.) The Stanford Encyclopedia of Philosophy, Fall 2015 Edition. URL = https:// plato.stanford.edu/archives/fall2015/entries/basing-epistemic/ Lehrer, Keith. 1997. Self-trust: A study of reason, knowledge, and autonomy. Oxford: Oxford University Press. Lord, Errol. 2018. The importance of being rational. Oxford: Oxford University Press. Lycan, William. 1988. Judgement and justification. New York, NY: Cambridge University Press. Lycan, William. 2013. Phenomenal conservatism and the principle of credulity. In C. Tucker (ed.) Seemings and justification: New essays on dogmatism and phenomenal conservatism. New York: Oxford University Press, 293–305. McCain, Kevin. 2008. The virtues of epistemic conservatism. Synthese, 164: 185–200. McCain, Kevin. 2012a. Against Hanna on phenomenal conservatism. Acta Analytica, 27: 45–54. McCain, Kevin. 2012b. The interventionist account of causation and the basing relation. Philosophical Studies, 159: 357–382. McCain, Kevin. 2014. Evidentialism and epistemic justification. New York, NY: Routledge. McCain, Kevin. 2016. The nature of scientific knowledge: An explanatory approach. Switzerland: Springer. McCain, Kevin and Poston, Ted. 2014. Why explanatoriness is evidentially relevant. Thought, 3: 145–153. McCain, Kevin and Poston, Ted. 2017. The evidential impact of explanatory considerations. In Kevin McCain and Ted Poston (eds.) Best explanations: New essays on inference to the best explanation. Oxford: Oxford University Press, 121–129. McGrath, Matthew. 2007. Memory and epistemic conservatism. Synthese, 157: 1–24. McGrath, Matthew. 2016. The justification of memory beliefs: Evidentialism, reliabilism, and conservatism. In Brian P. McLaughlin and Hilary Kornblith (eds.) Goldman and his critics. Malden, MA: Wiley Blackwell, 69–84. Moon, Andrew. 2012. Knowing without evidence. Mind, 121: 309–331. Podgorski, Abelard. 2016. Dynamic conservatism. Ergo, 3: 349–376. Poston, Ted. 2014. Reason and explanation: A defense of explanatory coherentism. New York, NY: Palgrave Macmillan. Pryor, James. 2000. The skeptic and the dogmatist. Nous, 34: 517–549. Quine, Willard V. and Joseph S. Ullian. 1978. The web of belief, 2nd Edition. New York, NY: Random House. Sklar, Lawrence. 1975. Methodological conservatism. Philosophical Review, 84: 374–400.

214

Kevin McCain

Vahid, Hamid. 2004. Varieties of epistemic conservatism. Synthese, 141: 97–122. Van Cleve, James. 2014. Why coherence is not enough: A defense of moderate foundationalism. In Matthias Steup, John Turri, and Ernest Sosa (eds.) Contemporary debates in epistemology, 2nd Edition. Malden, MA: Wiley Blackwell, 255–267. Zagzebski, Linda Trinkaus. 2012. Epistemic authority: A theory of trust, authority, and autonomy in belief. New York, NY: Oxford University Press.

12 Can Beliefs Be Based on Practical Reasons? Miriam Schleifer McCormick

The claim that there are no practical reasons for belief is one commonly made by theorists of reasons and belief. In thinking about different kinds of reasons, some will say there are those that pertain to belief (called epistemic or theoretical) and those pertaining to action (called practical), and that these are completely exclusive domains.1 To get a grasp of the difference between an epistemic reason and a non-epistemic reason for belief, consider two different kinds of reasons I can have for the belief that my child will awaken from the general anesthetic he was given. The first appeals to the statistics and probabilities that support the truth of the proposition related to the belief in a straightforward way. The second appeals to the value of having the belief; having this belief provides me with peace of mind, allows me to care for my other children and for my child to have his required surgery. When the claim that there are no practical reasons for belief is developed or defended, it is usually modified so that what is being denied is the possibility of a quite specific phenomenon. All will admit that practical considerations, in fact, can contribute causally to what one believes. Many will even say that such considerations can count as reasons for these subjects to believe and, again, such reasons may partially cause the beliefs. What they deny, however, is that these non-evidential reasons are reasons for which these subjects believe; beliefs, they say, cannot be based on such reasons.2 Exactly what relation between reason and belief is captured by the expression “for which” or “basis” that goes beyond the mere causal is controversial and much of the work in this volume is devoted to clarifying this relation. To get the idea of a consideration that is the cause of a belief but not one that the belief is based on consider the difference between the following cases: Sam’s reason for believing that Billy is kind is that he gives to Oxfam. Lois’s reasons for believing that the paper should be accepted are the paper’s excellent ideas and examples. Sam and Lois differ from Jean, whose belief that Abuja is the capital of Nigeria was caused by a blow to the head. While there is a reason why Jean believes what she

216

Miriam Schleifer McCormick does (i.e., an explanatory reason), there are no reasons for which she believes what she does (i.e., no motivating reasons).3

Given that this volume concerns the “epistemic basing relation,” if these domains really are completely distinct then one may think it follows trivially that beliefs cannot be epistemically based on practical reasons. Here are two examples representing the view that beliefs cannot be based on practical reasons: Although practical considerations can make a difference to what one believes, they do not do so by constituting grounds on which beliefs are based. . . . And rational beliefs, like rational actions, are rationalized by those considerations on which they are based. (Kelly, 2002, 174) Necessarily, if S takes r to be a practical reason S has to believe p, then S does not believe that p on the basis of r. (Comesaña, 2015, 195) To answer the title question fully, it needs to be divided into two separate questions. The first is: Can one take oneself to believe for practical reasons? I will argue that the most plausible characterizations of the basing relation do not rule this out. What is centrally involved in basing a belief on a reason is that one treats the relation as a justificatory one, that one endorses the connection between the reason and the belief. The further question is: Can this endorsement or representation be correct, proper or rational? Since many take the answer to the first question to be “No,” the second question never gets asked. My answer to the second question is also “yes”; one can distinguish between good and bad cases of believing for practical reasons just as one can distinguish between good and bad cases as believing for evidential ones. To make room for the idea that practical reasons can justify, however, requires de-linking doxastic justification from propositional, taking seriously a point made by Derek Piller (2001, 200) that “logic is only indirectly related to theories of reasoning.”

1 Can Beliefs Be Based on Practical Reasons? To assess whether it is the case that there is a whole class of reasons on which beliefs cannot be based, we need some understanding of the basing relation. But providing a characterization of this relation has proved extremely difficult, though there is much recent (and current) work being done trying to clarify it. It should be noted that, for epistemologists, the motivation to gain a clearer understanding of the basing relation is usually that doing so will help us understand what kind of relationship is required between a belief and a reason so that one is doxastically justified in holding the belief. Often these discussions assume that there are

Can Beliefs Be Based on Practical Reasons?

217

reasons that count in favor of the proposition believed (and this is termed “propositional justification”) but for the attitude of belief to be justified by this reason, it must be based on this reason, but what does it mean to be so based?4 As noted, the relationship cannot be simply causal as many causes of beliefs may not be reasons at all, let alone reasons for which one believes. As Keith Korcz puts it, “given that in principle anything can cause anything, a causal account of the basing relation will allow beliefs to be based on reasons which seem completely unrelated to them. For instance, one’s belief about having ridden a zebra once might, in principle, cause one to believe that Queen Elizabeth was a member of the Mafia” (2000, 545). In attempts to provide an account of the relation, some stick to a general causal story but try to articulate the appropriate kind of causation so as to rule out deviant causal chains while others have abandoned that approach in favor of what are sometimes termed “doxastic” accounts which state that a belief is only based on a reason if one has a metabelief that the belief is so based.5 Still others are searching for an alternative which avoids the pitfalls and retains the virtues of both the causal approach and the doxastic ones.6 These alternative or hybrid approaches attempt to articulate what the agent’s attitude who is basing the belief on the reason must be. The problem with simple causal accounts is they say nothing about the agent’s perspective or attitude. Doxastic accounts try to remedy this but in doing so overly intellectualize the requirement. If one needs to believe that a particular consideration is the normative reason for which one believes, this would seem to imply that children or animals could not base their beliefs on reasons. It seems to require that one have quite sophisticated concepts like “normative reason.” Further, one may wonder if one needs to have meta-beliefs about those meta-beliefs, and further meta-meta beliefs, leading to an infinite regress of higher order beliefs.7 Stephanie Leary (2017, 8) suggests that we label the cognitive attitude one adopts toward the reason and its relation to one’s belief as “conceiving.” So, let’s use the phrase conceiving of r as a reason to  as a stipulative term that refers to whatever cognitive element is involved in  ng for a reason, and simply claim that in order for R to be a reason for which one  s one must be disposed to  because one conceives R as reason to  . . . And I will assume here that believing R is a reason to  is sufficient, but may not be necessary for conceiving R as a reason to . She does not offer any precision about what it takes to conceive in this way, and she does not need to for the main point of her argument, but the examples she provides suggest that it requires a conscious endorsement of r being a reason. Believing for a reason, or basing a belief on reason

218

Miriam Schleifer McCormick

does not, however, always seem to require that one has such awareness. It is common in accounts of the basing relation, to talk about “representing” a consideration as reason which, depending on the view, can allow for such representations to be less than fully conscious or explicitly endorsed. Ram Neta (2019, 204) is quite explicit that one can represent a reason as justifying in the way required for it to be the reason for which one has a particular attitude without conscious recognition: This account is consistent with the obvious fact that it is possible for an agent to C for the reason R even when she doesn’t know what her reason for C’ing is: this is quite common for mature humans, and even more common for the less mature. There might be reasons for which I am angry at my parents, but I might not know what those reasons are: I can represent an explanatory relation even if I fail to represent some of its relata, just as I can represent a whole even if I fail to represent its parts. Also, my account of the basing relation is consistent with an agent’s C’ing for the reason R even when she also believes that R is not a good reason for C’ing: my account explains why such cases involve a kind of incoherence on the part of the agent, who is both committed (by her basing relation) to R being a good reason for C’ing, but also committed (by her beliefs) to R not being a good reason for C’ing. Kurt Sylvan (2016, 382) characterizes this agential attitude required in the basing relation as “treating”: Treating something as a reason does not require having the concept of a reason. This is an instance of the more general thought that treating Xs as Fs doesn’t require the concept of an F. Cats, for example, can treat entities as prey while lacking the concept PREY. Inferring p from q entails treating q as a normative reason to believe p, but does not require having or using the concept normative reason. Others use the language of “taking” to describe the attitude needed in basing a belief on a reason, or inferring one belief from another. In trying to articulate what it means to take something to be evidence or to be a reason which distinguishes it from believing it to be evidence or a reason, Comesaña asks us to consider the question of what it means to take someone to be a scoundrel. He says “in a very clear way, believing that S is a scoundrel is neither necessary nor sufficient for taking S to be a scoundrel” (200). You can act in ways which reveal that you take someone is a scoundrel and actually discover this to be the case without having the belief that he is a scoundrel. He proposes we “interpret takings not as full-fledged doxastic attitudes, but rather as motivational states that would be rationalized by the corresponding doxastic attitudes” (200).

Can Beliefs Be Based on Practical Reasons?

219

According to these views, then, for a belief to be based on reason requires that one “treat” or “take” the consideration as reason. Depending on the details of these accounts, treating or taking can be understood as a kind of complex disposition or representation, or a combination of both.8 And what is important is that someone treating a consideration as reason for believing figures in the explanation of why that person so believes. Most such accounts do not rule out that one could treat a practical consideration as a reason.9 This is most obvious for those who allow that the treating condition need not even be consciously endorsed. Consider someone who believes that humans all have the capacity for good. On what reason would such a belief be based? Having such a belief is good for the person who has it and for those with whom the person interacts. The benefits of believing this way is one of the factors that sustains it and it is hard to see what would rule out treating in this way, as tacitly endorsing it as the answer to the question why do you believe? Thinking about tacit treating of evidential considerations as reasons for belief, ones that may even be explicitly disavowed, makes it even more clear that practical considerations cannot be ruled out as reasons for which one believes. We saw earlier that Neta’s view clearly allows for this. He considers the case of Nyambi who claims that his reason for believing that the Russians bombed civilian targets in Syria is that he heard a news report of Al Jazeera when, in fact, he actually has “a disposition to believe everything he hears reported on CNN” and that he “(unbeknownst to him) exercises this disposition and thereby comes to believe that Russian forces have bombed civilian targets in Syria.” In such a case Nyambi would represent the relation between “I heard it on Al Jazeera” and “I believe the Russian forces bombed civilian targets” (178) in the way needed for it to be an example of basing his belief on that reason, but he would be mistaken in doing so: “Just as it is possible to represent the visible distance between two objects even when one is ignorant or mistaken about what those two objects are, so too is it possible to represent the explanatory relation between R and one’s C’ing, even when one is mistaken about what R and C are” (2019, 208). In recent work Paul Boghossian (forthcoming) is clear that to “take” something as reason does not require conscious awareness, and that the “taking condition” which he argues is necessary for one to infer q from p can sometimes be tacit, and may not even be accessible to the agent. In quick, effortless reasoning it may seem that one is drawing inferences, or basing a belief on a reason without needing to take p to be a reason for q. Susannah Siegel, for example says of the following that it is an inference without a taking condition, or what she calls inference without reckoning. Pepperoni: Usually you eat three slices of pizza when it comes with pepperoni. But tonight, after eating one slice, you suddenly don’t

220

Miriam Schleifer McCormick want any more. Struck by your own uncharacteristic aversion, you form the belief that the pizza is yucky. Though you don’t know it, the factors include the facts that (i) the pepperoni tastes very salty to you, (ii) it looks greasy, (iii) it reminds you of someone you don’t like, who you recently learned loves pepperoni, and (iv) you have suddenly felt the force of moral arguments against eating meat. If the next bites of pepperoni were less salty, the greasy appearance turned out to be glare from the lights, you learned that your nemesis now avoids pepperoni, and the moral arguments didn’t move you, the conclusion of your inference would weaken, and so would your aversion. You haven’t classified what you see and taste as: too greasy, too salty, reminiscent of your nemesis, or the sad product of immoral practices. Nor are you consciously thinking right now about any of these things. (2019, 18)10

Siegel argues that cases like this are inferences since your conclusion is epistemically dependent on the factors i–iv, and “you could have better or worse reasons for the conclusion . . . and that would make the conclusion better or worse.” Boghossian replies that if it is a genuine inference then there must be a taking condition, however tacit. He thinks that there are number of good theoretical reasons for insisting on there being such a condition. I will mention two of them. First, he argues that without such a condition there is no way to differentiate between mere association and inference. He considers the example of the depressive who, on thinking “I am having so much fun” always then thinks “But there is so much suffering in the world.” Such a person “is having an association of judgments on the basis of their content. But he is not thereby inferring from the one proposition to the other”(forthcoming). Now one of the differences between association and inference is that inference, or reasoning involves something you do and so allows you do be held responsible and appropriately criticized. He says that the taking condition offers a clear way of making “your reasoning count as agential” and for “your reasoning to be a process for which you could intelligibly be held rationally responsible.” We do not here need to adjudicate on who is right or about what makes something an inference or reasoning, as opposed to mere association. What is agreed upon is that sometimes inferences are made without a full understanding or endorsement of how one made them. In cases of implicit bias this seems to be what it going on. I may well take the fact that the candidate is male as a reason for my belief that he is better qualified for the job without any kind of awareness or endorsement that I believe for that reason.11 Once cases like these are brought to mind, nothing seems to rule out that one can represent practical considerations as justifying my belief. I can, for example, see my need for peace of mind as justifying my belief

Can Beliefs Be Based on Practical Reasons?

221

that my child will awaken from surgery. It may turn out that in doing so I am always mistaken in some way, but remember now we are only concerned with whether a practical reason can be one for which I believe, not whether in doing so I am necessarily doing something faulty or irrational. Perhaps beliefs based on such reasons are always improperly based. But an account of the basing relation should not only explain when beliefs are properly based, or when reasons actually justify. In Neta’s recent discussion he argues that one of the conditions that a theory of the basing relation should meet is that it can explain the difference about when the reason one represents as justifying a belief actually does. On his view, we have seen, one can be mistaken in one’s representation of a consideration as reason. In a similar vein, Singh (2018) argues that it is a constraint on an account of motivating reasons that they can fail to be good normative reasons. And, Boghossian is clear that an account of reasoning or inference needs to make room for explaining when one reasons poorly or makes a faulty inference: “Sometimes I reason from a p to a q where p does not support q. That makes the reasoning bad, but it is reasoning nonetheless. Indeed, it is precisely because it is reasoning that we can say it’s bad. The very same transition would be just fine, or at any rate, acceptable, if it were a mere association” (forthcoming). Before moving on to the question of whether beliefs based on practical reasons could be rational, we should note that some argue that the nature of belief rules out the possibility of basing beliefs on practical reasons. Jonathan Adler, for example, was explicit in taking this approach in his Beliefs’ Own Ethics. One of his central contentions is that it is a mistake to appeal to “normative notions” in assessing what to believe. He refers to such approaches as “extrinsic,” and he argues that this notion is based on a faulty assumption, namely that the concept of belief alone does not fix the ethics of belief. Beliefs, he maintains, have their own “ethics,” discovered by a clear analysis of the concept of belief. And such an analysis, he claims, shows that we must believe according to what we take to be evidence and that any mental state based on practical reasons is not really a belief. In arguing in support of his claim quoted earlier which says that practical considerations cannot be the reasons on which beliefs are based, Kelly makes a similar claim. He says “it is part of the nature of belief that beliefs are states which can be based on epistemic considerations but not on practical considerations” (2002, 177). The weaknesses in these conceptual arguments have been pointed out in a number of recent discussions. (See Rinard 2015, 2017, forthcoming; Leary 2017; McCormick 2015, 2017, forthcoming; Sullivan-Bissett forthcoming; Reisner 2009, 2014; Way 2016). I do not have space here to reconstruct the various arguments opposing the view that the nature of belief precludes that one can have practical reasons for believing. Common threads in these criticisms are that this “normativist” view requires a revisionist explanation of common practice, that it has an overly narrow

222

Miriam Schleifer McCormick

view of what counts as a belief, and an overly demanding view of what counts as a reason. Both Susanna Rinard and Jonathan Way, for example, argue that if one takes seriously what counts as reason according to such constitutive views, we would rarely act for practical reasons either. If I am right that there is no way, in principle, to rule out basing beliefs on practical reasons, then the argument between those who think some beliefs based on practical reasons can be permissible and those who do not must be conducted at the level which we find in the classic debate between Wilfred K. Clifford and William James. Clifford never denied that it is possible to believe for a practical reason, but he thought it was always wrong to do so. James, on the other hand, argued that under certain conditions believing without evidence is the right thing to do. I now turn to the question of whether it can ever be rational to base a belief on a practical reason.

2 Distinguishing Between Good and Bad Practical Reasons If there were a way to rule out practical reasons as being reasons which for which one believes then there would not be any need to ask the further question as to whether it is ever rational to do so. Many have claimed that it is a condition on something being a normative reason that it can be a motivating reason.12 They have then argued that practical reasons cannot be motivating reasons and therefore cannot be normative ones. I take it that my first section has shown that on many plausible accounts of what it means to base one’s belief on a reason that practical reasons can be motivating reasons. But others will say it is a condition on something being a genuine reason, and so a reason upon which a belief can be based, that it can be a good normative one.13 This is, for example Jonathan Way’s view. He says “Reasons for you to  must be considerations from which you could reason well to -ing” (2016). He then argues that one could never reason well from a practical consideration to a belief. Many agree that even if one could take a practical consideration to be a reason for believing that it is always a mistake to do so; one could never rightly (correctly, properly, aptly) base a belief on it. What are some features that make a reason a good one, or as some would put it, a genuine “normative” one? While I have no intent to list sufficient, or even necessary conditions of what makes something a good reason, I will discuss three features that are frequently invoked by those who are skeptical about practical reasons being able to be good reasons for which one can believe. These are that good reasons can provide guidance,14 that they apply universally, and they can rationalize or justify. (i) Guidance It seems there are two levels at which ordinary evidential considerations can guide. Perhaps there is a sense that I am guided by what I see when

Can Beliefs Be Based on Practical Reasons?

223

I look out my window to the belief that there is a gentle breeze blowing. One might think in such a case my cognitive system is being guided much the way an airplane is guided to the runway; if the guidance succeeds it is brought to the right place. But one might think that reasons should provide guidance in a more robust sense, more like when we turn to a friend or mentor for guidance; they should help advise us on how we ought to believe. When appeals to evidential or alethic considerations offer such guidance it is at the level of offering general rules of belief maintenance such as: “Only believe true propositions,” “Do not believe things for which you have no evidence,” “Withhold belief without sufficient evidence,” or “If one’s current evidence is against a proposition, one ought not believe it.” When such rules are internalized and we are guided by them, we comply with them automatically. This kind of guidance, or compliance, allows us to communicate, reason, survive, and even flourish. Most of the time believing in accordance with the evidence will be the way to have the best beliefs one can—the beliefs that are the most helpful to oneself and others, the beliefs that reflect the kind of person one wants to be. Most of the time I do not have to think about how I ought to believe, or why I believe as I do. But most of our actions also require little assessment or deliberation; we often operate almost automatically, and we often manage not to violate the rules of prudence or morality. It is when the right course to take is not obvious that deliberation comes in. And the same is true of belief. When I find myself questioning whether or not I should believe something and so deliberating about it, it is precisely because it is not obvious these evidential rules are the only ones that are relevant. In such cases, broadly practical considerations can help guide us. Take the case of Robert, an alcoholic, who is thinking about whether to commit to the program of Alcoholics Anonymous. Robert is highly educated and has done extensive research on AA. He knows that some of the central claims made by the program are epistemically suspect; there is a lot of evidence showing they are likely to be false. He also knows that AA is the most successful treatment program. He desperately wants to regain control of his life and knows that the most promising way of doing so is to commit to AA which requires believing some of claims lacking evidential support.15 In deliberating about whether to believe, for example, that “being an alcoholic means I have an overpowering chronic disease over which I cannot gain control and that the only way to recover is to turn to God” it is plausible that some of the considerations guiding him would be practical. This seems even more likely if a belief can be based on a reason that is not decisive, that is if it can be partly based on a reason in conjunction with others. In both Rinard and Leary’s discussion of how beliefs can be based on practical reasons they point to cases where practical considerations lead you to respond to evidence or epistemic reasons in a way that you would not if such a practical consideration was not present.

224

Miriam Schleifer McCormick

Here is an example from Leary: (Mary) Mary is usually sceptical of other people’s testimony and arguments. She didn’t give religion much thought growing up, but at college Mary comes to believe that she would be happier if she were to believe that God exists, and that this is a strong reason to do so. While taking a philosophy class, she reads Aquinas’s and Anselm’s arguments for the existence of God and she befriends a student who tells her about his experiences of divine revelation. While this would usually not be enough to convince Mary (given her sceptical nature), because she recognizes the practical benefit of believing in God as a reason to do so, this causes her to be more swayed by those arguments, and she ends up believing that God exists. Many will respond to such cases by saying that the belief is actually based on evidential and not practical considerations. But that this evidence would not convince Mary but for such a practical consideration speaks in favor of it being at least one of her motivating reasons. Leary points out that such a case is just like one where one only finds a current piece of evidence convincing because of some other evidential consideration that causes one to see it as convincing. She considers the case of Scully who believes that a suspect isn’t human based on a test result showing it is 80% likely that it is not. Ordinarily, this test result would not be enough to convince her but the fact that she had come across an old X-file stating the subject’s tissue sample had non-human DNA caused her to view the current evidence as sufficient reason for belief. In this case it seems one of the reasons for which Scully believes the suspect is not human is that it is stated in the old X-file report. Similarly, the fact that believing in God would make Mary happier is one of the reasons for which Mary believes. It certainly seems like it is one of the considerations guiding her. (ii) Universality Now let’s consider the second feature of good reasons for belief, that they apply universally, some even might say categorically. If I have evidence that supports the truth of a proposition, it seems I have reason to believe it regardless of my desires, inclinations or motivations. And if you find yourself in identical epistemic circumstances, you have the same epistemic reasons that I do, even if our desires and interests vastly differ. I know that you have a reason to believe the earth revolves around the sun without knowing any of your specific aims or goals. But purported practical reasons for belief seem very different. In his discussion of what excludes practical considerations from being reasons for belief Mark Schroeder (2010) notes that such reasons are almost always relativized to a particular subject and context (often involving evil demons or eccentric

Can Beliefs Be Based on Practical Reasons?

225

billionaires) and so are “idiosyncratic” whereas evidential reasons are universal, applying to all believers. Why is it, though, that evidential reasons seem to have this universality? In his argument for why epistemic rationality and reasons are not instrumental Kelly (2003) recognizes that we talk and act as if reasons that are tied to the general desire of self-preservation are categorical. I say, for example that you have a reason not to consume a poisonous substance without considering whether you have the goal of living longer or not. I may have no desire to take the medicine, but I still have a reason to do so which is connected to my more general commitment to life. That we can talk about reasons for belief applying to agents independently of their specific circumstances and goals may be because of a similar assumption we make about agents in general. Kelly argues, however, that when it comes to beliefs, we cannot make similar assumptions: “there is simply no cognitive goal or goals, which is plausible to attribute to people generally, which is sufficient to account for the relative phenomena. Individuals do not typically have this goal: believing the truth.” I agree that if one tries to find a specific desire or goal like the goal of acquiring true beliefs, that one will not find that in all agents. But what if the system of epistemic normativity that provides us with reasons to believe is ultimately dependent on its extreme practical value? In a footnote, Kelly considers a wider conception of instrumental rationality based on Williams’s (1981) discussion of internal and external reasons which counts a reason as instrumental if it will “advance not only goals which I actually hold but also goals which I might reach by a process of sound deliberation from my present ‘subjective motivational set.’” While it seems clear that I can have a reason to believe a particular proposition even if I have no desire to believe (and perhaps even a desire not to believe) that particular proposition, it may well be that in my subjective motivational set, as well as every other agent, one can find a commitment to practices that contribute to human flourishing. I view our system of epistemic norms and reasons as one of those practices, and so these reasons are practical though not narrowly instrumental. When making the case that practical reasons cannot be reasons for belief, “practical” is often used in a narrow sense to mean prudential, in my interest, or advantageous. Invariably monetary incentives are invoked. That we are unable to form a belief against (or without) the evidence when offered money or other incentives to do so is often taken to show decisively that we cannot believe for non-evidential reasons. While I think there are some beliefs that one cannot believe for some non-evidential reasons, I do not think we can generalize from examples of this kind to the conclusion that non-evidential reasons are never reasons for belief. It is quite likely that there are many actions one could not perform no matter how high the monetary incentive like, for example, killing an innocent person or jumping out the window, but this would not

226

Miriam Schleifer McCormick

tell us that one can never act for reasons of this kind. To object that one could perform these actions but one chooses not to begs the question. In both cases—that of believing and that of acting—one is being asked to do something that goes against a deeply entrenched view of who one is and what one values. We can see then that the broad category that can be termed “practical” or “pragmatic” goes beyond the narrowly instrumental. If part of your reason for believing something is that it will contribute to the good in general then this counts as a practical reason. And so we see on this view, that evidential reasons are also practical. And that ultimately their universality is explained by appeal to this their broadly practical value. (iii) Justification Way claims that one cannot reason well from a practical consideration to a belief because good reasoning must correspond to good arguments and that an argument with a practical consideration as a premise will always be a bad argument: “Since the argument ‘Believing in God would make me happy, so, God exists’ is plainly a bad argument, it follows that the corresponding reasoning is bad”(2016, 815). Further, it seems one appropriately bases a belief on a reason if the relation between the two is justifying (Neta 2019, 186–189). Here I think we arrive at the core of why it is thought that we cannot properly base our beliefs on practical reasons: it seems obvious to many that only truth-related considerations can justify or support a belief. But why do such considerations make a belief justified or rational, why are they good reasons? This is a question rarely posed and, to some, it may not sound like a coherent question. I have argued (2015, 2018) that such reasons are good ones because they provide us with true beliefs, and that the value of true beliefs is broadly practical.16 If this is the case then other broadly practical reasons can also rationalize, ones that contribute to flourishing despite their not being evidential. To try to get a handle on when we consider someone as reasoning well as opposed to reasoning poorly, let’s begin by thinking about cases where only evidential considerations are at play. I believe that the ground is wet and this belief is based on my belief that it rained last night. I take it that the fact that it rained last night supports my belief that the ground is wet. Although my belief may turn out to be false, it is likely that I reasoned in an acceptable manner, being guided in some sense by modus ponens. To take an example from David Hume, if you come to believe that “Irishmen have no wit,” and the reason you believe this is that all the Irishmen you have met are dull-witted, but you have only met two, then you fail to reason well. This is not, however, because the conclusion is not entailed by the premises. If one’s sample size is large enough, we would think it perfectly good reasoning to come to a belief that cannot be construed as

Can Beliefs Be Based on Practical Reasons?

227

a conclusion of a deductively valid argument. If my belief that it never snows in Virginia in July is based on the fact that it never has, this is perfectly acceptable reasoning. I have good reasons to believe it; the reasons offer support for my belief. Now can practical reasons offer support for my belief, and can we offer a parallel contrast of when they do and when they do not? I’ll begin by describing an actual case from my life. One of my very best friends, Susan, will be dying soon from an aggressive sarcoma. She has approached her illness and dying with acceptance despite being one of the heathiest and most active people I have known. Susan is not religious, would not claim to believe in God. She is also highly intelligent and educated; I remember her taking her L-SATs on a whim and achieving such a high score that she had top law schools courting her, though she was never really going to pursue that path. She believes that when she dies that some aspect of her being will still exist. Of course, people tell her all the time about stories and dreams that provide evidence for such a belief. But none of these form the basis of her belief; they are insufficient support and she would not believe it if these were the only considerations in support. She believes it because it helps to mitigate the concern she has for the pain of her loved ones. The reasons for which she believes it are broadly practical, connected to love and comfort rather than evidence and truth. Reasons of this kind, and the beliefs they support help provide coherence and meaning.17 Susan’s belief is quite different from the examples often cited in discussions about practical reasons to belief. As I mentioned earlier, these almost always appeal to incentives to believe, and the kind of practical reason is one of narrow prudential interest. Here is a standard example taken from Alston (1988) “Can you, at this moment, start to believe that the United States is still a colony of Great Britain, just by deciding to do so? If you find it too incredible that you should be sufficiently motivated to try to believe this, suppose that someone offers you $500,000,000 to believe it, and you are much more interested in the money than in believing the truth.” Alston’s main point in introducing such examples was to show we lack direct control over beliefs, but they have also been taken to show that we cannot believe for such reasons. Such monetary incentives cannot support beliefs in the way that considerations about the value of the belief can, the same value associated with following rules that tend to provide us with true beliefs, namely that they allow for both individual and collective flourishing. The difference between these two cases is similar to the difference between beliefs formed on the basis of good inductive inferences and those that result from prejudice. In one case I am being guided by a rule of belief maintenance that, if everyone internalizes and follows, will lead to overall flourishing. As noted earlier this does not mean that coming to have any particular belief

228

Miriam Schleifer McCormick

will be good for a particular person. A particular belief supported by the evidence can be very painful. And holding on to a particular prejudiced belief may serve someone’s self-interest, but doing so requires violating rules that lead to true beliefs, or following rules that lead away from them. One can make a similar distinction between beliefs based on practical reasons. If the reasons are such that they do not require one to violate or ignore the rules that tend to lead to truth then they can be justifying. In Susan’s case, there is no evidence in support of the falsity of her belief and so no evidence that requires ignoring or suppressing.18 Way’s argument against practical considerations being reasons for belief assumes that there is a link between good reasoning and good arguments, and he points out that one place where a pragmatist (meaning someone who thinks practical reason can be reasons for belief) could push back is by denying that link. What I am suggesting here is that these should be de-linked, particularly if arguments are understood as needed to be in propositional form, containing premises and conclusions. I mentioned earlier that many epistemologists take it that the basing relation is trying to capture the relationship that an agent’s belief must have to a reason where that reason when stated as a proposition offers justification for the belief when stated as a proposition. That some propositions offer support for others does not tell us anything about what is going on in the mental movements of the agent. Through our brief tour of different ways of thinking about basing, and reasoning, we can see that it is far from clear that what is going on with these mental movements is best captured by an argument with premises and conclusions. As we saw earlier, coming to believe, even when such beliefs are inferred from or based on other considerations (which may be other beliefs), can be done in an automatic manner and that to reconstruct what takes place in such mental transitions as something akin to an explicit argument with premises and conclusions is misleading. To ask whether an inference is valid when engaged in a logical proof and to ask whether someone is justified in coming to believe what she believes on the basis of the reasons she does are usually two very different kind of questions.19 Think about this piece of reasoning: I have to decide whether I should stay home and grade, or go see my friend’s band play. What goes on when I reason about this? It seems I make a kind of list of considerations in favor and opposed to each course of action. Some people even transfer this mental list on to actual paper to assist in their deliberation. If, in the end, I decide to stay home and grade, it seems anything that came up in that list can be a reason for my staying home and grading. But did it function as premise? Would it make sense to think of my deliberative process along these lines: If I don’t grade tonight it will just make things worse for me tomorrow. Things being worse for me tomorrow is something I should avoid. Therefore, I should grade.

Can Beliefs Be Based on Practical Reasons?

229

I suppose one can reconstruct my reasoning in such a way, though it bears little resemblance to what I think actually goes on in such deliberation. This way of thinking about deliberation fails to capture, for example, all the considerations that were rejected that supported another course of action. Arpaly and Schroeder, like Siegel, argue that much reasoning is done without deliberation, and that what makes something reasoning “require(s) nothing more than mental transitions that occur because of the logical relations (theoretical entailment, practical entailment, statistical relevance, or the like) between the different attitudes involved” (2012, 238). What kind of logical relation is left quite open. Given that they are concerned with both practical and theoretical reasoning, this cannot mean that there must be logical connection holding between propositions. I contend that certain kind of practical reasons can support beliefs while others cannot. And just like in the case of evidential reasons, the details matter. While Way imagines a pragmatist might want to deny the link he proposes, he offers some points to support the idea that it is “very natural indeed to think there is a tight connection between good reasoning and good arguments.” One is that reasoning is expressible, and that it may even be a shared activity. Even if we accept that reasoning is expressible, and I think some of the cases Siegel and others point to suggest otherwise, why think the form of expression must be by stating an argument with premises and a conclusion? When I ask Susan to describe why she believes as she does, she does not need to be silent. But what she says will not be of the form Way imagined about someone believing that God exists because it makes him happy. That is, she would not say “Believing that some aspect of my being will continue after I die is comforting and so some aspect of my being will continue after I die.” Instead, she would express different feelings she has when contemplating her impending death and the various possibilities of what will occur after. The feeling associated with the belief that some aspect of her being will continue to exist is supportive; it is part of what the belief is based on. This feeling is not that different from what some have deemed an intellectual seeming, or an intuition of truth. Some have referred to such seemings or intuitions as feelings of “rightness.”20 Such feelings can be part of what justifies one believing as one does. To most such a feeling includes feeling it true. But if a belief can be right, independent of its truth value, then a feeling like the one Susan has can also justify. Something can feel right because it helps one make sense of life, providing coherence and meaning.

3 Conclusion: Addressing Some Possible Worries While I can imagine many objections to the view I put forth, I will not have space here to address most of them. An important assumption for the view I set out in Section 2 is that the value of true beliefs is not

230

Miriam Schleifer McCormick

intrinsic, but ultimately grounded in its practical value. I have argued for this claim at length elsewhere.21 One who wants to resist that idea as well as the conclusions I come to here may say that even if there is some sense in which practical reasons can offer practical justification for beliefs, they cannot offer epistemic justification for belief. As mentioned earlier, this volume is concerned with the epistemic basing relation and so what I am saying here may seem to be largely beside the point. First, it is important to note that many who are concerned with understanding the basing relation are trying to understand it quite generally; what is the relationship between a particular mental state (intention, belief, hope, etc.) and a reason such that we say that one is based on the other. If one is thinking about this more general question then, once the relation is identified it may well turn out that beliefs and practical reasons can be related in that way. To rule out that possibility before we have identified the relationship is unmotivated. Second, and relatedly, as Way puts it “it would surely beg the question for the evidentialist to insist that reasons to believe be premises of epistemically good reasoning” where such reasoning is defined as those which evidentially support the beliefs they are based on. One thing epistemology should be concerned with is doxastic justification. If it turns out that sometimes one can be doxastically justified when one’s belief is supported by a practical reason what would it mean to say that one is not epistemically justified? Perhaps it means that such reasons could not be the kinds of reasons that determine whether one has knowledge. But what rules out that such considerations could bear on knowledge in a significant way? Any theory which allows for pragmatic encroachment on knowledge is allowing that practical considerations are not wholly irrelevant to whether one knows.22 I want to end by considering a recent challenge Arpaly has raised for anyone who thinks that there are practical reasons to believe. She provides this example: Imagine this: you have cancer and you do not yet know if the course of chemotherapy you have undergone will save you or not. You sit down at your doctor’s desk, trying to brace yourself for news, aware that at this point there might be only interim news—indications that a good or a bad outcome is likely. The doctor says there are reasons to be optimistic. Though you are still very tense, you perk up and you feel warm and light all over. You ask what the reasons are. In response, the doctor tells you about ironclad scientific results showing that optimism is good for the health of cancer patients. In such a scenario, Arpaly rightly points out, your heart would sink. And she says “A good theory of epistemic and practical reasons should

Can Beliefs Be Based on Practical Reasons?

231

account for the sense of being cheated or mocked that accompanies Sinking Heart cases.” Her answer is that the doctor is telling you a falsehood because there are no practical reasons to believe. When it comes to figuring out what to believe our concern and attention is almost always on whether it is true. And context matters. In the scenario described, it is evident that this is what the patient was asking about, and a doctor who did not get this would be either extremely insensitive or cruel. In another context, hearing about the practical reasons to believe would be entirely appropriate. Think again about Robert and AA. When he reads about the success of the program this provides him with practical reasons to believe AA’s tenets; the beliefs (whether true or not) can help him. Here, no sinking feeling need accompany someone telling him there are reasons to believe them even if they are practical. Whether or not he is justified in believing them for those reasons is another matter. If maintaining the belief would require him to ignore or suppress evidence then it may well be unjustified but perhaps, depending on the details, excusable.

Notes 1. See Derek Parfit’s (2001) very influential discussion on this point. He says “Practical and epistemic reasons are always quite different” though they can be related in important ways. Nomy Arpaly (ms) has recently argued that we have no practical reasons to believe: “It is a category mistake to talk about practical reasons to believe” (6). 2. In some discussions of the basing relation, reasons for which we X are also referred to as motivating reasons. For example, Keshav Singh (2018) begins his recent paper by saying “Motivating reasons are the reasons for which we . These reasons are also sometimes said to be the reasons in light of which, or the reasons on the basis of which, we act, believe and so on.” Other theorists have some qualms talking about motivating reasons for belief because it makes belief seem to be too action-like. Can I have a motivating reason to be in a state? I will mostly avoid talking in terms of motivating reasons, but when I do, I will, like Singh, take them to be equivalent to the reasons for which one believes. 3. Sylvan and Lord begin their paper (this volume) by presenting these cases. They argue that to understand what it means for a belief to be based on a sufficient normative reason requires more than the reason being both motivating and a good normative one. It is possible for each of those conditions to met, they claim, without the appropriate relation holding between it being motivating and being a good reason. They argue instead that this relation cannot be reduced to parts that obtain independently of each other. 4. The distinction is also sometimes made in terms of a belief being justifiable and beliefs being justified. This is, for example, how Korcz (2000) introduces his discussion. For helpful discussion of what the relation is trying to identify see also McCain (2012) and Neta (2019). 5. For a useful overview of more complex causal accounts see Sylvan (2016). 6. Ram Neta (2019), Kurt Sylvan (2016), and Keshav Singh (2018) are among those who are trying to articulate an alternative theory. 7. McCain (2012) brings up this objection, as well as many other problems with doxastic accounts in Chapter 7.

232

Miriam Schleifer McCormick

8. Boghossian (2014) canvasses different ways that we can understand the taking condition which he argues is necessary for reasoning from p to q, or basing q on p. He ultimately concludes that it is best understood as “following a rule in one’s thought” but that the notion of following a rule is “an unanalyzable primitive” (17). 9. Jonathan Way (2016) comes to the same conclusion. None of these accounts rule out that one can base one’s belief on practical considerations if one allows that one can reason badly from such considerations. I discuss his argument for why practical reasons cannot be reasons for belief later. 10. Nomy Arpaly and Timothy Schroeder (2012) have also brought to attention examples of reasoning without awareness or deliberation. 11. I have been referring to discussions of inference which is not directly on the topic of the basing relation but the two are very closely connected. In introducing the topic of inference Boghossian (2014) says “My inferring (3) from (1) and (2), it may be thought, is for me to judge (3) on the basis of (1) and (2). Our question about inference, then may be seen as a special topic that is discussed in the epistemological literature under the label the ‘basing relation.’” In a more recent discussion he says, “inferring from the judgment that p to the judgment that q establishes p as the epistemic basis for judging q, whereas associating q with p does not” (2019). 12. See Rinard (forthcoming) for a discussion of those who argue for this claim. 13. There are many for whom the idea of a bad reason does not make sense. If something is genuinely a reason this means that it something like a unit of normativity, counting in favor of some action or belief. According to such views, when one is motivated by a consideration they take to be a reason but that does not support the belief or action, this is not bad reason but no reason at all. I think it is important to maintain the idea that there can be bad reasons. See my for further discussion on this point. Others who agree are Singh (2018) and Hieronymi (2013). 14. See Way (2016) for a discussion of why reasons should guide. Littlejohn (2019) questions whether reasons need to offer guidance. 15. One may wonder if belief is really needed here. Would something like acceptance be enough? In Jonathan Cohen’s (1992) influential discussion of this distinction he is clear that one of the key differences between belief and acceptance is acceptance lacks what he call “credal feelings.” Even if one is skeptical about there being any phenomenology associated with believing, given that one can accept for the sake of argument, or because one needs to get something done, one does not need to commit to the truth of the proposition. And it would seem that it is exactly such a commitment that would be needed in the case of AA. For a very interesting discussion of the AA case, and where I found the example see Jenkin (ms). 16. I argue (especially in Chapter 2, 2015) that one cannot make sense of a point of view that is distinctly and exclusively epistemic while at the same time retaining the normative force of epistemic reasons. For epistemic goods— such as truth, knowledge, and rationality—to be goods to be promoted, or to provide us with norms to follow, they must be grounded in the practical or moral. Thinking about why we value truth and knowledge reveals that the norms guiding us in what is called the epistemic realm are not isolated from other normative domains. 17. For a discussion on how the adaptive purpose of belief is much more than to track truths see McKay and Dennett (2009). 18. In Chapter three of (2015), I discuss, at length, how to distinguish permissible non-evidentially based beliefs from permissible ones. 19. Boghossian clearly distinguishes between inference as reasoning and inferences as arguments “I am not here talking about inference as argument:

Can Beliefs Be Based on Practical Reasons?

233

that is as a set of propositions, with some designated as ‘premises’ and one designated as the ‘conclusion’. I am talking about inference as reasoning, as the psychological transition from one (for example) belief to another” (2019). 20. See de Sousa (2008). 21. See footnote 15. I address some worries about this view in responses to Trevor Hedberg’s comments here: https://syndicate.network/symposia/philosophy/ believing-against-the-evidence/ 22. For arguments supporting the view that practical considerations matter for whether one counts as knowing, see Fantl and McGrath (2007, 2009). For a recent overview and critique of arguments in favor of pragmatic encroachment see Roeber (2018), and Gardiner (2018).

References Adler, Jonathan Belief’s Own Ethics. Cambridge: MIT Press, 2002. Alston, William P. “The Deontological Conception of Epistemic Justification” Philosophical Perspectives 2 (1988), 257–299. Arpaly, Nomy “Epistemology and the Baffled Action Theorist” (ms). Arpaly, Nomy, and Timothy Schroder “Deliberation and Acting for Reasons” Philosophical Review 121 (2) (2012). Boghossian, Paul “Inference, Agency and Responsibility” in M. Balcerak-Jackson and B. Balcerak-Jackson, eds., Reasoning: Essays on Theoretical and Practical Thinking, 2019, 101–128. ——— “What Is Inference?” Philosophical Studies 169 (1) (2014). Cohen, Jonathan L. An Essay on Belief and Acceptance. Oxford: Oxford University Press, 1992. Comesaña, Juan “Can We Believe for Practical Reasons?” Philosophical Issues 25 (2015), Normativity, 189–207. de Sousa, Ronald. “Epistemic Feelings” in G. Brun, U, Doguoglu, and D. Kuenzle, eds., Epistemology and Emotions. Farnham, UK: Ashgate Publishing, 2008. Fantl, Jeremy and Matthew McGrath “On Pragmatic Encroachment in Epistemology” Philosophy and Phenomenological Research 75 (3) (2007), 558–589. ——— Knowledge in an Uncertain World. Oxford: Oxford University Press, 2009. Gardiner, Georgi “Evidentialism and Moral Encroachment” in Kevin McCain, ed., Believing in Accordance with the Evidence: New Essays on Evidentialism. Cham: Springer, 2018, 169–195. Hieronymi, Pamela. “The use of reasons in thought (and the use of earmarks in arguments).” Ethics 124 (1) (2013): 114–127. Jenkin, Robert “Alcoholics Anonymous & Normative Epistemology: Can Believing Useful False Beliefs Be Justified?” (ms., masters thesis). Kelly, Thomss “The Rationality of Belief and Some Other Propositional Attitudes” Philosophical Studies 110 (2002), 163–196. ——— “Epistemic Rationality as Instrumental Rationality: A Critique” Philosophy and Phenomenological Research 66 (3) (May 2003), 612–640. Korcz, Keith Allen “The Causal-Doxastic Theory of the Basing Relation” Canadian Journal of Philosophy 30 (2000), 525–550. Leary, Stephanie “In Defense of Practical Reasons for Belief” Australasian Journal of Philosophy 95 (3) (2017), 529–542. Littlejohn, Clayton. “Being More Realistic About Reasons: On Rationality and Reasons Perspectivism.” Philosophy and Phenomenological Research 99 (3) (2019), 605-627.

234

Miriam Schleifer McCormick

McCain, Kevin “The Interventionist Account of Causation and the Basing Relation” Philosophical Studies 159 (3) (2012), 357–382. McCormick, Miriam Schleifer Believing against the Evidence: Agency and the Ethics of Belief. New York. Routledge, 2015. ——— “Responding to Skepticism about Doxastic Agency” Erkenntnis (2017), DOI: 10.1007/s10670-017-9906-2 ——— “No Kind of Reason Is the Wrong Kind of Reason” in Kevin McCain, ed., Believing in Accordance with the Evidence: New Essays on Evidentialism. Springer, 2018, 261–276. McKay, Ryan T. and Daniel C. Dennett “The Evolution of Misbelief” Behavioral and Brain Sciences 32 (2009), 493–561. Neta, Ram “The Basing Relation” The Philosophical Review 128 (2) (2019), 179–217. Parfit, Derek “Rationality and Reasons” in Dan Egonsson et al., eds., Exploring Practical Philosophy. Burlington, VT: Ashgate, 2001, 17–39. Piller, Christian “Normative Practical Reasoning” Proceedings of the Aristotelian Society Suppl. 25 (1) (2001), 195–216. Reisner, Andrew. “The Possibility of Pragmatic Reasons for Belief and the Wrong Kind of Reasons Problem.” Philosophical Studies 145 (2) (2009): 257–272. ——— “No Exception for Belief”. Philosophy and Phenomenological Research 94 (1) (2017): 121-143. ——— “A Short Refutation of Strict Evidentialism”. Inquiry (5) (2014): 1–9. Rinard, Susanna “Against the New Evidentialists” Philosophical Issues 25 (2015), DOI: 10.1111/phis. 12061 ——— “Believing for Practical Reasons” Nous 53 (4) (2019), 763–784. DOI: 10.1111/nous.12253 Roeber, Blake “The Pragmatic Encroachment Debate” Nous 52 (1) (2018), 171–195. Schroeder, Mark “Value and the Right Kind of Reason” in R. Shafer-Landau, ed., Oxford Studies in Metaethics 5. Oxford: Oxford University Press, 2010, 25–55. Siegel, Susanna “Inference as Reckoning” in M. Balcerak-Jackson and B. Balcerak-Jackson, eds., Reasoning: Essays on Theoretical and Practical Thinking. Oxford: Oxford University Press, 2019, 15-31. Singh, Keshav “Acting and Believing under the Guise of Normative Reasons” Philosophy and Phenomenological Research (2018). Sullivan-Bissett, Ema “Explaining Doxastic Transparency: Aim, Norm, or Function?” Synthese 195 (8) (2018), 3453–3476, DOI: 10.1007/s11229-017-1377-0 Sylvan, Kurt “Epistemic Reasons II: Basing” Philosophy Compass 11 (7) (2016), 377–389. Way, Jonathan “Two Arguments for Evidentialism” The Philosophical Quarterly 66 (265) (2016), 805–818. Williams, Bernard “Internal and External Reasons” in Bernard Williams, eds., Moral Luck. Cambridge: Cambridge University Press, 1981, 101–113.

13 Epistemological Disjunctivism and Factive Bases for Belief Duncan Pritchard

1 Epistemological Disjunctivism According to epistemological disjunctivism, in paradigmatic epistemic conditions one’s rational basis for a perceptual belief that p can be the factive reason that one sees that p. Seeing that p is factive in the sense that one can only see that p if p is true. What makes epistemological disjunctivism so distinctive is the further claim that the rational support one’s belief enjoys is reflectively accessible, such that there can be reflectively accessible factive rational support.1 Note that epistemological disjunctivism is not a theory of knowledge in general, but rather a thesis about a particular kind of perceptual knowledge.2 Call any case where the epistemic conditions for perception are paradigmatic—both subjectively and objectively speaking (the reason for this qualification will become apparent later)—a good case. Now compare this with a corresponding bad case where the epistemic conditions are systematically bad but which is indistinguishable from the good case (at least from the subject’s perspective).3 So consider, for example, a case where one is a brain-in-a-vat being stimulated by supercomputers to have experiences which are indistinguishable to the subject from those had by one’s non-envatted counterpart. In such a scenario, a large body of one’s beliefs are false and yet one has no inkling at all of this fact. According to epistemological disjunctivism, one’s reflectively accessible rational support in the good case is the factive reason that one sees that p. In contrast, in the bad case, one’s reflectively accessible rational support in the bad case cannot be that one sees that p since this is not the case (we will consider in a moment just what the rational support, if any, that is in play amounts to). Thus even though good and bad cases are ex hypothesi indistinguishable, one’s reflectively accessible rational support in the good case can be significantly different from one’s reflectively accessible rational support in the bad case. This is the sense in which this is an epistemological disjunctivism: either one is in the good case, with reflectively accessible rational support of one kind, or one is in the bad case, with reflectively accessible rational support of another (much lesser) kind.

236

Duncan Pritchard

Epistemological disjunctivism is a controversial view. I’ve discussed some of the key problems that face the position in a number of places, so I won’t be making a comprehensive case for the position as a whole here. My concern is rather the more specific task of exploring how the notion of a factive rational base for belief plays out when it comes to epistemological disjunctivism. Along the way we will be seeing how epistemological disjunctivism responds to two particular challenges: the basis problem and the access problem.

2 Good/Bad Cases and Bases for Belief So what is one’s rational basis for belief in the bad case if it isn’t the factive rational basis of seeing that p? In order to answer this question, we need to distinguish between three types of epistemic reasons: normative, motivational, and explanatory. Normative epistemic reasons are objectively good reasons for regarding the target proposition as true, regardless of whether that proposition is believed by the subject, and regardless of whether the subject is aware of these reasons. That one sees that p is of course an excellent normative epistemic reason for regarding p as true, in that being factive it entails p. Call a factive normative epistemic reason a decisive normative epistemic reason. Not all normative epistemic reasons are decisive, however. Compare being presented with a pig in paradigmatic epistemic conditions with seeing a glimpse of the pig’s curly tail as it disappears from view. The former would be naturally characterised as seeing that there is a pig; a decisive normative epistemic reason. But not the latter. This is, rather, what I will call a suggestive normative epistemic reason. It offers objectively good reasons for regarding the target proposition as true, but not decisively (factively) so. Next, consider motivational epistemic reasons, which are one’s reason for believing that p.4 We’ve already noted that a normative epistemic reason for believing that p can co-exist with one not believing that p, so clearly there can be normative epistemic reasons without the corresponding motivational epistemic reasons. Indeed, even if one does believe that p, one’s motivational epistemic reason for believing that p might be distinct from the normative epistemic reason for believing that p. For example, one’s motivational reason for believing that p could be testimony that in fact provides no support for p at all, and which disregards a genuine non-testimonial normative epistemic reason to believe that p. Indeed, we can also formulate variants on this case where the normative epistemic reason is completely absent (i.e., there is no objectively good reason for regarding p as true), but where the subject believes that p regardless. In such scenarios the motivational epistemic reason obtains without there being any normative epistemic reason for the target belief. Finally, consider explanatory epistemic reasons, which are considerations that explain why one believed that p. Such reasons need not line up

Epistemological Disjunctivism

237

with normative or motivational epistemic reasons. That they don’t line up with normative epistemic reasons should be straightforward. If there are objectively good reasons for treating p as true, but these reasons are not the subject’s motivational epistemic reasons, then one would hardly try to explain why the subject believed that p by appeal to the normative epistemic reasons. Putting the point this way, however, might lead one to suspect that one’s motivational epistemic reasons and one’s explanatory epistemic reasons are just the same thing, but this would be a mistake.5 In order to see this, it is useful to set out how one compares to one’s envatted counterpart with this threefold distinction in place. All three strata of reasons line up quite straightforwardly in the good case. That one sees that p is a (decisive) normative epistemic reason for believing that p, and it is also one’s motivational epistemic reason for believing that p and the explanatory epistemic reason for why one believes that p. But this alignment doesn’t hold in the bad case. To begin with, since it isn’t true that one sees that p, this can’t be a normative epistemic reason for believing that p. When envatted it does seem to one that p, however, and one might plausibly regard this as a (suggestive) normative epistemic reason for believing that p. Accordingly, one might imagine that in the bad case one’s motivational epistemic reason is that it seems to one that p rather than that one sees that p. A moment’s reflection reveals that this can’t be quite right, however. For remember that one is completely unaware that one is deceived, and so one’s reason for believing that p is not that it seems to one that p but rather that one takes oneself to see that p. This is where the distinction between motivational and explanatory epistemic reasons becomes important, since what does seem right is that the explanation for why one believes that p in the bad case is that it seems to one that p, even though this is not the reason for which one believes that p. Aside from the difference in normative epistemic reasons available in the good and bad cases (i.e., seeing that p versus it seeming to one that p), there is thus also a disconnect between one’s motivational and explanatory reasons in the bad case.6 We thus have three kinds of reason in play, reasons which subjects can stand in very different relations to depending on whether they are forming their beliefs in good or bad scenarios. With the foregoing discussion in hand, we are now in a position to explain what epistemological disjunctivism refers to when it describes the subject’s rational basis for belief. Obviously it cannot be the normative epistemic reason, since this needn’t be one’s basis for belief at all, not least because it can co-exist with one failing to believe the target proposition. Moreover, it cannot be the motivational epistemic reason either, since it is obviously important to epistemological disjunctivism that the subject has different rational bases for belief in the good and the bad case, and yet we have just seen that this need not be the case when it comes to motivational epistemic reasons. The relevant rational basis is thus the subject’s explanatory

238

Duncan Pritchard

epistemic reasons, which as we have seen can come apart across good and bad cases. In particular, the epistemological disjunctivist claim is that in good cases one’s rational basis for believing that p is the explanatory epistemic reason that one sees that p, while in corresponding bad cases it is the distinct explanatory epistemic reason that it merely seems to one that p. This is the sense in which, in the good case, one has a factive rational basis for belief, as opposed to a merely non-factive rational basis in the corresponding bad case, even though the good and bad cases are ex hypothesi indistinguishable from the subject’s perspective.7 I take the epistemic basing relation in play here to be straightforwardly causal.8 It would take us too far afield to digress into the complex philosophical issues raised by this relation, but let me at least offer some general supporting remarks.9 First off, notice that the familiar non-causal accounts of the basing relation are not going to be a good fit with epistemological disjunctivism, given how we have described the view. Take doxastic accounts, for example, which characterise the basis relation in terms of certain meta-beliefs held by the subject regarding the basis for her belief—e.g., that such-and-such is a consideration in support of the truth of the believed proposition.10 But given the distinction we have drawn between one’s motivational and explanatory epistemic reasons, and how they come apart in bad cases, it is hard to see why the subject would have the meta-beliefs in question in the bad case.11 (Of course, proponents of the doxastic account of the epistemic relation might well take this as a reason to be suspicious of epistemological disjunctivism, but that’s a separate issue). In any case, thinking of the epistemic basis for belief in terms of explanatory epistemic reasons not only naturally suggests a causal reading of the basing relation but also a particular way of thinking about the notion of causation in play. The key is to notice how one would go about picking out the explanatory reasons. I take it that this would most naturally be done in causal terms—i.e., by picking out the most salient part of the causal explanation for why the subject holds her belief. But since the explanatory reasons are the subject’s basis for belief, then that is just to say that one is thereby picking out the basis in causal-explanatory terms too. Such a causal-explanatory account of the epistemic basis relation fits particularly well with epistemological disjunctivism. Proponents of such a view maintain that our natural way of thinking about perceptual knowledge in epistemically paradigmatic conditions is in terms of factive epistemic reasons. Put in causal-explanatory terms, this means that where conditions are epistemically optimal then the causal explanation naturally traces the basis for belief back to the factive explanatory epistemic reason of the subject’s seeing that p. In contrast, where conditions are not epistemically optimal, then the sub-optimality will prompt one to focus

Epistemological Disjunctivism

239

on more specific bases for belief. So, for example, rather than treating the subject’s explanatory epistemic reason as being that she sees that there is a pig in front of her, it might instead be that, say, she can see something resembling a pig’s tail. Where we are dealing with epistemically suboptimal situations that are indistinguishable from the subject’s perspective from epistemically paradigmatic conditions then the sub-optimality of the conditions will again be salient. Unlike other situations involving suboptimal epistemic conditions, however, the type of sub-optimality in question doesn’t prompt us to focus upon a specific concrete basis for belief, but rather picks out the generic basis that it merely seems to the subject that p. Moreover, notice that while there are familiar problems with causal accounts of the epistemic basing relation, they don’t seem to be relevant here. In particular, insofar as they arise at all, they do so only in the bad case, and yet it is in the good case that epistemological disjunctivism is offering a distinctive account of rational support. Accordingly, these problems, such as they are, do not pose any specific challenge to epistemological disjunctivism. Consider the familiar problem regarding basing and deviant causal chains, for example. Here, for instance, is Alvin Plantinga’s (1993, 69) presentation of such a case: Suddenly seeing Sylvia, I form the belief that I see her; as a result, I become rattled and drop my cup of tea, scalding my leg. I then form the belief that my leg hurts; but though the former belief is a (part) cause of the latter, it is not the case that I accept the latter on the evidential basis of the former. In general, such scenarios offer us a deviant causal chain such that the most natural causal basis for the belief (in this case, the prior belief that one sees Syliva) does not seem to be the epistemic basis for that belief. Crucially, however, it is in the very nature of the kind of scenario required for such a deviant causal chain that the subject is not in epistemically paradigm conditions for perceptual knowledge. Plantinga’s case doesn’t even concern perceptual knowledge, paradigmatic or otherwise, but we can straightforwardly adapt the case to make the target belief perceptual. As Keith Korcz (2015, §1) puts it, we can easily construct parallel cases involving deviant causal chains such as ‘glitches in the brain, wandering thoughts, wishful thinking, strong emotions, etc.’, and of course all of these could causally result in a perceptual belief. But insofar as the subject is in conditions where the perceptual belief is causally produced via deviant mechanisms of this kind, however, then it follows that she is not in the good case. This is no accident, since it is characteristic of the good case as epistemological disjunctivism understands it that the

240

Duncan Pritchard

subject’s normative, motivational and explanatory reasons line-up, as this is a consequence of such cases being epistemically paradigmatic. Where there are deviant causal chains present, however, then clearly the subject’s normative epistemic reason will not align with her explanatory (and possibly also motivational) epistemic reason, as this will concern the deviant causal basis for the belief. Since the subject is not in the good case, she is not in possession of the factive epistemic support that is distinctive of epistemological disjunctivism. Accordingly, insofar as there is a problem posed by deviant causal chains for causal accounts of the epistemic basing relation, it is not a problem that is in any way specific to the combination of epistemological disjunctivism and the causal account. Or consider so-called ‘Gypsy-Lawyer’ cases, as famously proposed by Keith Lehrer (1971). The details of the case aren’t important for our purposes. What is salient is just that we have a subject who believes the target proposition in an epistemically inferior way, but who as a result subsequently comes to further reflect on the evidence and in doing so formulates a sound epistemic rationale for that belief. Nonetheless, the psychology of the subject is such that the motivation for the belief remains the original one with the inferior epistemic pedigree. Lehrer argues that the subject’s belief amounts to knowledge, and hence that this is a problem for the causal account of the basing relation, since it would be required to treat the subject’s basis for belief as being the original epistemically inferior ground. I am somewhat sceptical about Gypsy-Lawyer cases, as they do not strike me a psychologically plausible.12 But we can in any case set them to one side as, like deviant causal chains, they do not pose a difficulty that is specific to epistemological disjunctivism—i.e., which would specifically impact upon the kind of factive epistemic bases that epistemological disjunctivism claims are reflectively available in the good case. This is because it is crucial to Gypsy-Lawyer scenarios that there is a disconnect between the causal basis for belief and the available epistemic basis. In terms of our tripartite division of previously mentioned reasons, these scenarios would thus require the subject’s explanatory and motivational reasons (presumably aligned in these cases) to come apart from the normative epistemic reason (which would be the bona fide epistemic support available for the target belief). But that just reminds us that the subject is not in the good case, since such cases essentially involve, as we saw earlier, an alignment between these three kinds of reason. Indeed, this is just a consequence of what epistemological disjunctivism has in mind when it talks of the epistemically paradigmatic conditions of the good case. Hence the distinctive kind of epistemic support proposed by epistemological disjunctivism is not even in play in Gypsy-Lawyer scenarios. Accordingly, as with deviant causal chains, even if Gypsy-Lawyer cases do pose a problem for causal accounts of the epistemic basis relation, they don’t create any difficulties that are specifically relevant to the

Epistemological Disjunctivism

241

combination of epistemological disjunctivism and causal accounts of the epistemic basing relation.

3 A Dilemma for Epistemological Disjunctivism On the face of it, we can pose a dilemma for epistemological disjunctivism, at least as described. As we will see, however, being clear about the epistemic basing relation in play helps us to resolve this dilemma. It is common to hold that seeing that p is just a specific way of knowing that p. If that’s right, then seeing that p entails knowing that p. Call this the entailment thesis.13 If one accepts the entailment thesis, then it seems that according to epistemological disjunctivism one’s epistemic basis for knowing that p can entail that one knows that p. But that sounds very odd. Shouldn’t one’s epistemic basis for knowledge be distinct from the knowledge in question? Elsewhere, I have called this the basis problem for epistemological disjunctivism.14 This is the first horn of the dilemma. It seems that avoiding the basis problem requires denying the entailment thesis, but the problem with this strategy, however, is that it seems to exacerbate another problem that epistemological disjunctivism faces. We thus encounter the second horn of the dilemma. This is because combining epistemological disjunctivism with the entailment thesis seems to offer us a straightforward way out of another difficulty that one can pose for the view. This worry arises out of the very idea that one can have a rational basis for belief that is both reflectively accessible and factive. The concern is that this seems to suggest that one ought to be able to reflect on the nature of one’s rational support and thereby come to recognise, since it is factive, that the target proposition must be true. But doesn’t that mean that one can come to know, via an entirely nonempirical route, that a specific empirical proposition is true? That sounds somewhat miraculous. Elsewhere I have called this the access problem for epistemological disjunctivism.15 The entailment thesis appears to offer a painless way of dealing with this problem. This is because if seeing that p is just a way of knowing that p, then this factive reason is only reflectively accessible to one if one already knows that p. Moreover, since seeing that p is an empirical way of knowing that p, this means that one only has reflective access to the factive reason when one already has empirical knowledge that p. It follows that reflecting on one’s factive reasons is not going to provide one with a nonempirical route to knowledge that p. The access problem thus disappears. If one does not ally epistemological disjunctivism to the entailment thesis, however, then this way out of the access problem is no longer available. So it seems that epistemological disjunctivism is presented with two unpalatable options. On the one hand, it can accept the entailment thesis, and thereby evade the access problem, but then it faces the basis problem. On the other hand, it can reject the entailment thesis, and thereby evade

242

Duncan Pritchard

the basis problem, but then it loses its straightforward response to the access problem. As I will explain in the next section, the way out of this dilemma is to recognise that epistemological disjunctivism can consistently embrace the second horn of the dilemma.16

4 Resolving the Dilemma The key to resolving this dilemma is to recognise that the entailment thesis is false: seeing that p doesn’t entail knowing that p. What is undoubtedly true—and it is this that drives the entailment thesis—is that seeing that p is closely related to knowing that p in that it puts one in a robust epistemic relationship to p. One doesn’t count as seeing that there is a pig before one merely by being visually presented with a pig in poor epistemic conditions (e.g., where the pig is not clearly in view).17 But while there is a close relationship between knowing that p and seeing that p (we will specify what this relationship is in a moment), I contend that it is not a relationship of entailment. I think this point is apt to be lost because we tend to work with a Manichaean conception of good versus bad cases which is far too crude. Instead, we need to recognise that between these two extremes are a spectrum of intermediate cases of epistemic import. In particular, while I would grant that in the good case as described earlier, where conditions are epistemically paradigmatic, one both sees that p and knows that p, there are intermediate scenarios in the vicinity of the good case where the latter comes apart from the former. Let us distinguish two ways in which a scenario can be epistemically paradigmatic, objectively and subjectively. As the name suggests, the former concerns how the epistemic conditions in fact are. So, for example, the subject’s perception is veridical, her cognitive faculties are working appropriately, she is presented with the target objects in clear view and in good cognitive conditions for perception, and so on. There are also no defeaters present, where this includes normative defeaters of which the subject might be unaware. In contrast, a scenario is subjectively epistemically paradigmatic where from the subject’s point of view all indications suggest that it is epistemically paradigmatic, regardless of whether it in fact is epistemically paradigmatic. So it seems to the subject as if her perception is veridical, that her cognitive faculties are working appropriately, that cognitive conditions are good, that there are no defeaters present, and so on. Clearly a scenario can be subjectively epistemically paradigmatic without being objectively epistemically paradigmatic. Indeed, the bad case as it is usually presented would fit the bill on this score. While the subject is in epistemic conditions that are about as bad as they can be from an objective point of view, since there is nothing to indicate to the subject

Epistemological Disjunctivism

243

that she is the victim of such a scenario, the prevailing epistemic conditions are also nonetheless subjectively epistemically paradigmatic. The good case, at least as it is understood by epistemological disjunctivism, treats the scenario in play as being both objectively and subjectively epistemically paradigmatic. This is important because there is a noteworthy kind of quasi-good case which is objectively epistemically paradigmatic while not being subjectively epistemically paradigmatic. In particular, imagine that a subject is in fact in paradigmatic epistemic conditions for perception, but is nonetheless aware of a misleading epistemic defeater. Let’s say, for example, that someone has convinced her that the perceptual scene before her is not as it appears (even though there is in fact nothing amiss at all). This would suffice to make the scenario subjectively epistemically sub-optimal, even though it is in fact objectively epistemically paradigmatic. This kind of case is interesting precisely because it demonstrates how seeing that p and knowing that p can come apart. I take the claim that the agent lacks knowledge that p in this case to be uncontroversial. Given the presence of the misleading defeater, the subject is not in a position to form the belief that p in an epistemically responsible fashion, and that will prevent her from having knowledge. That the agent nonetheless sees that p is less straightforward, but no less defensible. This follows from the fact that the subject is in objectively paradigmatic epistemic conditions. We can see this by imagining that the subject subsequently becomes aware that the misleading defeater is just that: misleading. While there is no temptation to regard her former self as knowing that p—she wouldn’t have even believed that p, for one thing, at least not in an epistemically responsible manner—she would regard her former self as seeing that p. (Though she wouldn’t, of course, regard her former self as believing that p on this basis, even if her former self happened to believe that p, albeit in an epistemically irresponsible fashion.) After all, she was objectively in paradigmatically good epistemic conditions for seeing that p, and the subjective factors that prevented her from responsibly believing that p were objectively inaccurate. This is the sense in which seeing that p is robustly epistemic, in that one needs to be in objectively good epistemic conditions in order to satisfy this relation. In particular, when one sees that p then one is objectively in good epistemic conditions for knowing that p. This is why seeing that p and knowing that p tend to go together. Where they come apart, however, is in those specific scenarios where the epistemic conditions, while objectively epistemically paradigm, are subjectively sub-optimal. So seeing that p comes apart from knowing that p, and hence the entailment thesis is false. The second horn of the dilemma, however, was the charge that if one denies the entailment thesis then one is faced with the access problem. Put in terms of our new terminology, if one can see that p

244

Duncan Pritchard

without knowing that p in the quasi-good case, and if one’s rational support for believing that p in this case can be that one sees that p, then what is to prevent someone’s belief that p in the quasi-good case from enjoying this factive rational support? Moreover, since the factive rational support in play is meant to be reflectively accessible, then what is to prevent the subject from becoming aware that she enjoys this factive epistemic support for her belief and hence coming to have knowledge that p as a result purely by reflecting on the rational support she has for her beliefs? The worry, of course, is that one is thereby acquiring new knowledge of a specific empirical proposition in the quasi-good case via a reflective route, and that looks very much like a reductio of epistemological disjunctivism. If that were a consequence of the denial of the entailment thesis, then it would clearly be bad news for epistemological disjunctivism. But this result is not in the offing. Remember that the distinctive claim made by epistemological disjunctivism is that one’s rational basis for believing that p in the good case is that one sees that p, and that one then knows that p on this basis. Accordingly, that one satisfies the relation of seeing that p in the quasi-good case does not entail that seeing that p can be one’s rational support for believing that p. Even so, one might insist that this is by-the-by, since so long as the subject believes that p in the quasi-good case, regardless of the epistemic impropriety of doing so, then why doesn’t she count as knowing that p in virtue of enjoying factive rational support? It is undeniable, of course, that agents can form beliefs in epistemically irresponsible ways, so that is not the issue here. Given the foregoing, it is also true in this case that there is a decisive normative epistemic reason for believing that p in play, in form of the subject’s seeing that p. Crucially, however, notice that the subject’s motivational and explanatory epistemic reasons do not line up with the normative epistemic reason in this scenario. The subject may believe that p, but her awareness of the misleading defeater entails that her motivational epistemic reason won’t be that she sees that p, since she is aware that she doesn’t see that p. One’s motivational epistemic reasons cannot be decided by fiat.18 In any case, even if one wishes to insist on an account of motivational epistemic reasons such that one can in principle elect any motivational epistemic reason that one wishes, it remains that the basing relation here concerns the explanatory epistemic reason anyway, and that clearly cannot be that the subject sees that p in this case. We would explain the subject’s belief, rather, in very different terms. Indeed, I don’t think we would naturally describe the subject’s mental state as one of belief in the first place, given that it is contrary to the manifest counterevidence, but rather as something rather different, such as wishful thinking.19 But even if we granted that it was a belief, what would explain the formation of the belief would be whatever story accounted for why the subject formed this belief despite the presence of the misleading defeater, such as that she

Epistemological Disjunctivism

245

was strongly motivated to regard p as true.20 However the explanatory epistemic reason is individuated, it will not be that the subject sees that p. It thus cannot be the subject’s basis for belief in the quasi-good case, and hence even though the subject genuinely does see that p in this case, there is nonetheless no route to knowledge that p via the (epistemically irresponsible) formation of a belief that p. We can also see this point in action by reconsidering the case as involving a misleading normative defeater. A normative defeater is a defeater that the subject is unaware of but which she epistemically ought to be aware of. Imagine, for example, that there is a sign in clear view above the pig pen that states that the creatures therein are not real pigs. It could be that our subject is so inattentive that she fails to clock this sign. Moreover, we can stipulate that the sign is intended as a joke, and hence that the normative defeater is misleading. Nonetheless, our subject’s inattention hardly suffices to enable her to have knowledge that there are pigs in the pen regardless.21 Misleading normative defeaters are interesting in that they are not easily captured by the objectively/subjectively epistemically paradigmatic distinction that we drew earlier. As misleading defeaters, they certainly don’t fall on the objective side of the contrast. But as normative defeaters of which the subject is unaware, they don’t obviously fall on the subjective side of the contrast either. What is important is that they are clearly inimical to being in generally paradigmatic epistemic conditions, and so they are excluded by the conception of the good case that the epistemological disjunctivist has in mind. In any case, if our subject is in conditions where a normative misleading defeater is present, then she will fail to know that p. Since the misleading defeater is normative, however, there are no issues here about what the subject believes, epistemically responsibly or otherwise. Given that she is unaware of the misleading defeater, she will naturally believe that p regardless. Moreover, unlike the case involving the non-normative misleading defeater offered earlier, the subject’s motivational epistemic reason will be that she sees that p. Since the subject is in objectively paradigmatic epistemic conditions, she will also genuinely see that p, and hence the motivational and the normative epistemic reasons will be in alignment. And yet the subject’s belief is not rationally supported by her seeing that p since this is not her explanatory epistemic reason. After all, prompted by her failure to attend to the clearly displayed sign, we would explain why the subject believes what she does not by appealing to the factive epistemic reason but rather by opting for something much weaker, such as that it seems to her as if there are pigs before her. The sub-optimality of the epistemic conditions, which is what prevents it from being a good case as the epistemological disjunctivist understands that notion (however this sub-optimality is to be analysed), is also what ensures that the causal explanatory story about the subject’s epistemic basis naturally flows not

246

Duncan Pritchard

towards the factive epistemic reason but rather to a different epistemic basis entirely. The upshot of the foregoing is that one can deny the entailment thesis without succumbing to the access problem. In all of the cases where the subject continues to see that p while failing to have knowledge that p, her epistemic basis for belief is not the factive epistemic reason. Accordingly, she will not be in a position to reflect on the factive rational basis for her belief and thereby acquire new empirical knowledge that she previously lacked. So long as we are clear about the nature of what epistemological disjunctivism is claiming about the subject’s rational support in the good case, and so long as we are in addition clear about what the nature of her epistemic basis is, then the dilemma we have posed for epistemological disjunctivism disappears.22

Notes 1. I have defended such a proposal in a number of works. See, for example, Pritchard (2007, 2008, 2009, 2011a, 2011b, 2012b, 2015). See also Neta & Pritchard (2007 ). The view is rooted in work by McDowell (e.g., 1995, 2002). 2. That said, one might be tempted to apply such an epistemological model to other kinds of knowledge, in paradigmatic epistemic conditions, beside the perceptual. McDowell (1994, 1995) certainly seems tempted along these lines. For my own preferred account of knowledge—anti-luck (/anti-risk) virtue epistemology—see Pritchard, Millar & Haddock (2010, chs. 1–4) and Pritchard (2012a, 2016a, forthcominga). I explain how epistemological disjunctivism relates to this broader theory of knowledge in Pritchard (forthcomingb). 3. There will be a spectrum of cases between these two extremes of good and bad cases, as detailed in Pritchard (2011a, 2012b, part 1). Since most of these intermediate cases won’t be relevant for our purposes, we won’t discuss them all, but some of them will become salient later. 4. Note that in what follows I will take it as given that one is aware of one’s motivating epistemic reasons. There are no doubt cases where one’s motivational epistemic reason is not what one thinks it is, but such cases raise complications that are not our concern here, so we can set them to one side without loss. 5. I’m grateful to Clayton Littlejohn for convincing me, in discussion, that this would be a mistake. This issue crops up briefly in a recent exchange we had on Pritchard (2012b)—see Littlejohn (2016) and Pritchard (2016b). 6. I introduce and develop this threefold account of epistemic reasons in Pritchard (forthcomingc). Note that one doesn’t need to embrace epistemological disjunctivism in order to grant that one’s motivational epistemic reason in the bad case is that one sees that p. Given how the good case is described, everyone should grant that one’s motivational epistemic reason is naturally understood along these factive lines. What marks the difference between epistemological disjunctivism and competing proposals rather concerns the explanatory epistemic reason, since according to such proposals these reasons can never be understood along factive lines (i.e., on these views our normal practices involving factive epistemic reasons should not be taken at face-value).

Epistemological Disjunctivism

247

7. Notice that one consequence of this way of thinking about epistemic basing in the context of epistemological disjunctivism is that the subject’s rational basis for belief in the bad case is not reflectively accessible. After all, she is mistaken about what her explanatory basis for belief is. This point is significant, since on traditional epistemically internalist ways of thinking about reflectively accessible rational support, the reflectively accessible rational support one’s belief enjoys in the good case cannot be any better than the reflectively accessible rational support one’s corresponding belief enjoys in the bad case, the presumption being that there is a non-factive rational basis for belief that is reflectively accessible in the latter case. Epistemological disjunctivism not only rejects this general way of thinking about reflectively accessible rational support—what is known as the new evil demon intuition—but also, at least as I develop the view at any rate, the more specific idea that in the bad case one’s (non-factive) rational basis for belief is reflectively accessible. For the locus classici with regard to the new evil demon intuition, see Lehrer & Cohen (1983) and Cohen (1984). For a very useful contemporary survey of work on this topic, see Littlejohn (2009). For further discussion of how epistemological disjunctivism rejects this intuition, see Neta & Pritchard (2007) and Pritchard (2012a, 2015, forthcomingc). 8. For two key defences of the causal account of the epistemic basing relation, see Moser (1989) and McCain (2012). Note that counterfactual or dispositional accounts of the epistemic basis relation can count as causal in the sense that I am using the term, to the extent that they are offering particular ways of understanding the causal relation in play. For examples of the former, see Swain (1981) and Bondy (2015). For examples of the latter, see Turri (2011) and Evans (2013). 9. For some useful general discussions of the epistemic basing relation, see Neta (2010) and Korcz (2015). 10. For some key defences of doxastic accounts of the epistemic basing relation, see Tolliver (1982) and Leite (2008). 11. Note that this issue will arise for versions of the doxastic account that only demand that the meta-belief is necessary for the epistemic basing relation. As such, it will likely also affect ‘mixed’ accounts of the epistemic basing relation, such as causal-doxastic views like that offered by Korcz (2000). 12. For scepticism about the Gypsy-Lawyer case, see Audi (1983) and Wallbridge (2018). For a (fairly) recent defence of the case, see Kvanvig (2003). 13. The entailment thesis is defended by, amongst others, Dretske (1969), Williamson (2000), and Cassam (2007). 14. See especially Pritchard (2011a, 2012b). 15. See especially Pritchard (2012b). 16. In fact, I also think that one can consistently embrace the first horn of the dilemma too, in that I’m not altogether convinced by the challenge posed by the basis problem. (Is it really so obvious that one’s knowledge cannot be epistemically based on a way of knowing?) In any case, since one only needs to embrace one horn of a dilemma in order to evade it, we can set this point to one side here. Moreover, as we will see, it is only the denial of the second horn of the dilemma that is of interest to our current concerns. 17. One would count as seeing a pig, of course. Seeing an object is an objectdependent relation, in that one cannot see a pig that isn’t there, but it isn’t an otherwise robustly epistemic relation like seeing that p (i.e., it is not the kind of epistemic relation that bears a close relationship to knowledge that p). 18. Indeed, this is in part of the reason why I am so sceptical about Gypsy-Lawyer cases, since they do seem to presuppose that one’s motivational epistemic reasons can be settled in this way.

248

Duncan Pritchard

19. At least in the specific sense of ‘belief’ that is relevant to epistemologists anyway (roughly: that propositional attitude which is a constituent part of knowledge). I discuss this notion of belief and some of its implications in Pritchard (2015, part 2, 2018b). 20. Note, by the way, that it wouldn’t be relevant here that the subject thought that they might have a defeater for the defeater, as that would remove the misleading defeater, and thereby change the structure of the case. (Whether a defeated misleading defeater can return one to a scenario that is subjectively epistemically paradigmatic is a further issue, which it would take us too far afield to engage with here.) I discuss defeaters in more detail in Pritchard (2018a). 21. I offer a virtue-theoretic explanation of why such defeaters are knowledgeprecluding in Pritchard (2018a). 22. I am grateful to Pat Bondy, Adam Carter, and Ram Neta for detailed comments on an earlier version of this paper.

References Audi, R. (1983). ‘The Causal Structure of Indirect Justification’, Journal of Philosophy 80, 398–415. Bondy, P. (2015). ‘Counterfactuals and Epistemic Basing Relations’, Pacific Philosophical Quarterly 96, 542–569. Cassam, Q. (2007). ‘Ways of Knowing’, Proceedings of the Aristotelian Society 107, 339–358. Cohen, S. (1984). ‘Justification and Truth’, Philosophical Studies 46, 279–295. Dretske, F. (1969). Seeing and Knowing, London: Routledge & Kegan Paul. Evans, I. (2013). ‘The Problem of the Basing Relation’, Synthese 190, 2943–2957. Korcz, K. A. (2000). ‘The Causal-Doxastic Theory of the Basing Relation’, Canadian Journal of Philosophy 30, 525–550. ———. (2015). ‘The Epistemic Basing Relation’, Stanford Encyclopedia of Philosophy, (ed.) E. Zalta, https://plato.stanford.edu/entries/basing-epistemic/ #CauTheBasRel. Kvanvig, J. L. (2003). ‘Justification and Proper Basing’, The Epistemology of Keith Lehrer, (ed.) E. Olsson, 43–62, Dordrecht, Holland: Kluwer Publishing Co. Lehrer, K. (1971). ‘How Reasons Give Us Knowledge, or the Case of the Gypsy Lawyer’, Journal of Philosophy 68, 311–313. Lehrer, K., & Cohen, S. (1983). ‘Justification, Truth, and Coherence’, Synthese 55, 191–207. Leite, A. (2008). ‘Believing One’s Reasons Are Good’, Synthese 161, 419–441. Littlejohn, C. (2009). ‘The New Evil Demon Problem’, Internet Encyclopaedia of Philosophy, (eds.) B. Dowden & J. Fieser, www.iep.utm.edu/evil-new/. ———. (2016). ‘Pritchard’s Reasons’, Journal of Philosophical Research 41, 201–219. McCain, K. (2012). ‘The Interventionist Account of Causation and the Basing Relation’, Philosophical Studies 159, 357–382. McDowell, J. (1994). ‘Knowledge by Hearsay’, Knowing from Words: Western and Indian Philosophical Analysis of Understanding and Testimony, (eds.) B. K. Matilal & A. Chakrabarti, 195–224, Dordrecht, Holland: Kluwer.

Epistemological Disjunctivism

249

———. (1995). ‘Knowledge and the Internal’, Philosophy and Phenomenological Research 55, 877–893. ———. (2002). ‘Knowledge and the Internal Revisited’, Philosophy and Phenomenological Research 64, 97–105. Moser, P. (1989). Knowledge and Evidence, Cambridge, Cambridge University Press. Neta, R. (2010). ‘The Basing Relation’, Routledge Companion to Epistemology, (eds.) S. Bernecker & D. H. Pritchard, 109–118, New York: Routledge. Neta, R., & Pritchard, D. H. (2007). ‘McDowell and the New Evil Genius’, Philosophy and Phenomenological Research 74, 381–396. Plantinga, A. (1993). Warrant: The Current Debate, Oxford: Oxford University Press. Pritchard, D. H. (2007). ‘How to Be a Neo-Moorean’, Internalism and Externalism in Semantics and Epistemology, (ed.) S. Goldberg, 68–99, Oxford: Oxford University Press. ———. (2008). ‘McDowellian Neo-Mooreanism’, Disjunctivism: Perception, Action, Knowledge, (eds.) A. Haddock & F. Macpherson, 283–310, Oxford: Oxford University Press. ———. (2009). ‘Wright Contra McDowell on Perceptual Knowledge and Scepticism’, Synthese 171, 467–479. ———. (2011a). ‘Epistemological Disjunctivism and the Basis Problem’, Philosophical Issues 21, 434–455. ———. (2011b). ‘Evidentialism, Internalism, Disjunctivism’, Evidentialism and its Discontents, (ed.) T. Dougherty, 362–392, Oxford: Oxford University Press. ———. (2012a). ‘Anti-Luck Virtue Epistemology’, Journal of Philosophy 109, 247–279. ———. (2012b). Epistemological Disjunctivism, Oxford: Oxford University Press. ———. (2015). Epistemic Angst: Radical Skepticism and the Groundlessness of Our Believing, Princeton, NJ: Princeton University Press. ———. (2016a). ‘Epistemic Risk’, Journal of Philosophy 113, 550–571. ———. (2016b). ‘Responses to My Critics’, Journal of Philosophical Research 41, 221–238. ———. (2018a). ‘Anti-Luck Virtue Epistemology and Epistemic Defeat’, Synthese 195, 3065–3077. ———. (2018b). ‘Disagreement, of Belief and Otherwise’, Voicing Dissent: The Ethics and Epistemology of Making Disagreement Public, (ed.) C. Johnson, 22–39, London: Routledge. ———. (Forthcominga). ‘Anti-Risk Virtue Epistemology’, Virtue Epistemology, (ed.) J. Greco & C. Kelp, Cambridge: Cambridge University Press. ———. (Forthcomingb). ‘Epistemological Disjunctivism and Anti-Luck Virtue Epistemology’, Epistemological Disjunctivism, (eds.) C. Doyle, J. Milburn & D. H. Pritchard, London: Routledge. ———. (Forthcomingc). ‘Shadowlands’, The New Evil Demon: New Essays on Knowledge, Justification and Rationality, (ed.) J. Dutant, Oxford: Oxford University Press. Pritchard, D. H., Millar, A., & Haddock, A. (2010). The Nature and Value of Knowledge: Three Investigations, Oxford: Oxford University Press. Swain, M. (1981). Reasons and Knowledge, Ithaca, NJ: Cornell University Press.

250

Duncan Pritchard

Tolliver, J. (1982). ‘Basing Beliefs on Reasons’, Grazer Philosophische Studien 15, 149–161. Turri, J. (2011). ‘Believing for a Reason’, Erkenntnis 74, 383–397. Wallbridge, K. (2018). ‘The Peculiar Case of Lehrer’s Lawyer’, Synthese 195, 1615–1630. Williamson, T. (2000). Knowledge and Its Limits, Oxford: Oxford University Press.

14 From Epistemic Basing to Epistemic Grounding Jesper Kallestrup

I Introductory Remarks The orthodox view of epistemic basing is that agent S’s belief that p is justified when S has a reason r (or evidence e) to believe p and S believes p on the basis of r (or e). In that case, S’s belief is properly based. While there are competing accounts, most prominently in terms of causal conditions, of what basing amounts to, most parties accept that basing is non-primitive.1 Moreover, if justified belief is necessary for knowledge, we can take r to be the basis for S’s knowledge that p, which, given such conditions, thus causally depends on r. In epistemology, proper basing backs causal explanation of knowledge: to causally explain why S knows p involves citing the basis for that state. Epistemic basing is responsible for the conversion of propositional into doxastic justification, but even if any reductive analysis of what converts the latter into knowledge is futile, metaphysical explanations of knowledge remain feasible. Epistemic grounding is a relation of metaphysical dependence of knowledge on more fundamental epistemic states, that is, a non-primitive, partial order with a specific structure that backs such explanation in epistemology: to metaphysically explain how S knows p involves citing the ground for that state.2 The determination relation is a case in point, and a paradigm example of an epistemic ground is the state of epistemic seeing qua way of knowing. Knowing that p and seeing that p are related as determinable to determinate in that they share relevant determination dimensions. So, while a reliably produced, visual experience as of p may count as the basis for knowledge that p, seeing that p constitutes a ground for such knowledge. The chapter proceeds as follows. Section II presents epistemic basing chiefly as a causal relation between a belief and a justificatory reason for which the belief is held. Section III turns to the metaphysics of grounding, arguing that ‘grounded in’ picks out a plurality of specific relations of metaphysical dependency, rather than a single, primitive relation. Section IV then develops the logic and metaphysics of determination as one such dependency relation. Section V demonstrates how epistemic grounding can be understood in terms of this suitably adapted, determination relation, which in turn is apt to underwrite metaphysical explanation of knowledge. Finally, Section VI summarises our findings.

252

Jesper Kallestrup

II Epistemic Basing Epistemic basing is a relation between S’s belief that p and the reason r for which S believes p, where r is some mental state or extra-mental fact that justifies the belief by providing justificatory support for p.3 More specifically, basing is the relation whose obtaining makes r the (or a) reason for which S believes p, and when basing is epistemic r justifies the belief; that is compatible with r also being the (or a) motivating reason for which S believes p. The relation is inferential if r itself is a belief; if instead r is a perceptual experience, the belief that p can be based on r without being formed by inference from r. Most accounts of epistemic basing involve two components: r must be the reason why S believes p, by being a salient cause of S‘s belief, and r must justify the belief, by providing epistemic support for p.4 If r does not justify the belief that p, then r is not the reason for which S believes p, even if r is the reason why S believes p. Similarly, if r is not the reason why S believes p, then r is not the reason for which S believes p, even if r justifies that belief. Epistemic basing is typically regarded as what converts propositional into doxastic justification.5 S’s belief is propositionally justified just in case S has an undefeated reason to believe p, and S’s belief is doxastically justified just in case S believes p for that reason. If S fails to believe p for the reason S has, then the belief is merely propositionally justified for S. Pretty much everyone agrees that such conversion is non-primitive. The received view appeals to a suitable causal connection between belief and justifying reason; roughly: (CAUSAL BASING) S’s believing p is based on justifying reason r =df S’s believing p is caused by r Two preliminary observations are in order: (i) r is on this view typically understood as a mental state, but if r is taken to be a propositional reason then S’s believing p should instead be caused by S believing r, so that, in either case, basing involves mental-to-mental causation; (ii) Moser (1989) imposed a stronger requirement that believing p be causally sustained by believing r, but, as Goldman (2001) noted, this rules out the possibility of justifiably believing p on the basis of forgotten reasons. A familiar problem with (CAUSAL BASING) concerns deviant causation, which of course also plagued causal theories of knowledge. More precisely, deviant causal chains of beliefs show that causation is insufficient for basing.6 Here’s Plantinga (1993: 69, fn. 8): (SYLVIA) Suddenly seeing Sylvia, I form the belief that I see her; as a result, I become rattled and drop my cup of tea, scalding my leg. I then form the belief that my leg hurts; but though the former belief is a (part) cause of the latter, it is not the case that I accept the latter on the evidential basis of the former.

From Epistemic Basing to Epistemic Grounding

253

One way to fix the problem about deviant causes is to restrict r to direct causes of belief.7 The person’s belief that their leg hurts is based only on the felt pain, which is the direct cause of the belief that their leg hurts. In contrast, forming the belief that they see Sylvia, becoming rattled, etc., are at best indirect causes of that belief. To insist that in a causal chain of events or states basing only pertains to the cause that is immediately before the belief is to deny that basing is transitive. That may be unproblematic; after all, the belief can still causally depend on links in the chain that are prior to the direct cause.8 Nor is the concern that some direct causes of beliefs fail to constitute reasons, as when one belief accidentally causes another belief with a completely unrelated content. For (CAUSAL BASING) is couched in terms of justifying reasons. Anyway, a deviant cause of a belief need not be indirect: (FIONA) Fiona comes to believe that tomorrow will be a bright sunny day as a result of reading the BBC forecast. Due to a glitch in her brain this belief then accidentally causes the further belief that she need not bring an umbrella to work. The worry is that Fiona has no inkling of what directly caused her umbrella belief, let alone how the weather belief provides epistemic support for it. Inspired by Moser (op. cit.), consider a refined formulation: (CAUSAL BASING*) S’s believing p is based on justifying reason r =df S’s believing p is directly caused by r and by S having associated p and r For S to associate p and r is for S to be de re aware of r’s providing epistemic support for p, which again should be limited to the point at which the belief is formed. Such awareness is not for S to merely seem as if r supports p, as such appearance could be misleading, nor for S to have a justified belief that r supports the belief that p, since such meta-belief would then itself need to be properly based. Rather, the awareness involves S recognising that r supports p, and the claim is that such awareness must be part of what directly caused S’s belief in p. Imposing such additional doxastic constraint, while thus still in the spirit of the causal view, helps with our putative counterexamples. The person in (SYLVIA) fails to associate the belief that they see Sylvia with the belief that their leg hurts; nor does Fiona in (FIONA) associate her weather belief with her umbrella belief. But the causal theorist is not home and dry yet. A residual worry is that doxastic justification depends on utilizing reasons properly in ways that go beyond (CAUSAL BASING*). Turri (2010a) offers an example:9 (MISS IMPROPER) The proposition that Mansour is guilty is justified for Miss Improper, because she knows that (i) he had a motive to

254

Jesper Kallestrup kill the victim, (ii) his fingerprints are on the murder weapon, (iii) he previously threatened to kill the victim, and (iv) an eyewitness places him at the crime scene. But Miss Improper reasons as follows: tea leaves say that (i)—(iv) make it overwhelmingly likely that Mansour is guilty. (i)—(iv) are true. So, Mansour is guilty.

Here Miss Improper is making inadequate use of the justifying reasons she possesses, because of a mistaken meta-belief about the relation between the target proposition and those reasons. Therefore, her belief is unjustified, and yet (CAUSAL BASING*) seems to be satisfied. The diagnosis is that she associates the proposition believed with the available reasons in the wrong way, thinking erroneously that tea-leaves is what confers justificatory power on those reasons vis-à-vis that proposition. Deploying such dubious belief-forming method constitutes misuse of the impeccable reasons she possesses. Consider instead: (CAUSAL BASING**) S’s believing p is based on justifying reason r =df S’s believing p is directly caused by r and by S having appropriately associated p and r Miss Improper not only possesses justifying reasons to believe the target proposition, those reasons also directly cause her to form that belief together with awareness of their relation of epistemic support. Still, she fails to form a justified belief, because she fails to take proper account of the way in which those reasons provide that support. At any rate, our aim is not to defend an elaborate causal theory against all kinds of objections, but merely to show that any promising account of basing must include causal conditions, in combination with embedded doxastic constraints. Thus, a purely doxastic theory violates a central intuition about basing.10 A crude version has it that: (DOXASTIC BASING) S’s believing p is based on justifying reason r =df S believes that p, and that r, and that r provides epistemic support for p What is distinctive is that S has an appropriate meta-belief to the effect that r is a justificatory reason to hold the belief that p. But it would seem possible to hold a meta-belief that r provides support for p without the belief that p being based on r. Consider: (PRIEST) A trusted priest just testified to James that God created the Universe. James is a devout Christian who believes the priest’s testimony provides support for its content. However, James already believes God created the Universe, for he believes whatever the Bible says; a belief he acquired years ago, and reaffirms every so often, by

From Epistemic Basing to Epistemic Grounding

255

reading the Book of Genesis. Moreover, James fears that taking the word of anyone alive today as evidence about the origin of the universe will lead him to lose faith in the Bible’s authority. James’ creationist belief is based on (his belief about) what the bible says, because that text is the reason why he formed the belief in the first place, and is what sustains his belief to this day. Assuming the text provides a justificatory reason for him to believe as he does, it is the reason for which he holds that belief. But James satisfies the conditions in (DOXASTIC BASING) for his creationist belief being based on the testimony he receives from the priest. While James both trusts the priest, and believes that his say-so supports the target proposition, James’ fear prevents forming a belief in that proposition as a result thereof.11 The only bells and whistles that would help the doxastic theorist accommodate (PRIEST) are ones that pertain to the causal origin (or preservation) of James’ belief.12 If the foregoing is on the right track, basing is an epistemic relation which can do causal explanatory work in epistemology. For an epistemically significant, causal explanation of why S formed a justified belief in p involves citing a reason r which is both the salient cause of S’s belief in conjunction with S adequately associating p with r and the available justifying reason for S to believe p.13 That is, S’s state of belief is explained by r qua justifier together with such association. Naturally, causal explanations of epistemic states outwith epistemology, say in neuroscience, will differ in various respects, but only epistemic explanations essentially appeal to features under epistemic descriptions in their explanans. And on the assumption that doxastic justification is necessary for knowledge (which need not imply any epistemological internalist or externalist commitments about either), basing features principally in causal explanations that make reference to reasons as causes of why knowledge comes about. Obviously, such reasons would need to cause states of knowledge qua undefeated justifiers above any pertinent threshold for knowledge. In sum, by presenting a reason that is both a justifier (modulo such qualifications) and a cause, basing is a salient component in an epistemically significant, causal explanatory account of why S knows p.

III Metaphysical Grounding Section II made a prima facie case for the indispensability of causality in any viable account of epistemic basing. A causal (-cum-doxastic) theory of basing can do explanatory work in epistemology vis-à-vis the generation of doxastic justification and knowledge. Basing is primarily about the causal dependency of justified belief, and hence of knowledge, on their basis, but basing is silent about the metaphysical dependency of such belief and knowledge on their ground. This section outlines the notion of metaphysical grounding, arguing that only a specific, fine-grained,

256

Jesper Kallestrup

dependency relation is fit for purpose. Section IV develops the notion of determination as one such relation, and Section V then adapts that relation to epistemology. Following Block (2015: 113), ontology concerns what types of things exist, whereas metaphysics concerns the ultimate nature of things. Ontology may dictate commitment to the existence of Fs, but only metaphysics will reveal what it is in virtue of which the Fs are F. Adopting that distinction, we take neither to fall within the scope of the other. One way of answering the metaphysical question is to specify what grounds the F. Thus, when we say the derivative is nothing over and above, or obtains in virtue of, the fundamental, we have a grounding claim in mind. The mental is ultimately physical if grounded in the physical. Ground is a guide to nature, and we shall confine metaphysics to the study of ground.14 More precisely, levels of less fundamental goings-on are connected via grounding to levels of more fundamental goings-on, all the way down to the most fundamental level if there is one; if instead the world is a bottomless pit of increasingly fundamental levels, then grounding presumably has no ending. That the world is gungy, containing wholes all parts of which have further proper parts, cannot be ruled out a priori. We say grounding is well-founded just in case a grounding chain terminates in goings-on that themselves lack grounds. Since the (absolutely) fundamental is the ungrounded, grounding is well-founded just in case such chain terminates in fundamental goings-on. Observe that non-well-foundedness need not present a problem, which is fortunate given its epistemic possibility, provided any infinite grounding chain has lower bounds.15 Just as basing is a diachronic notion of causal dependence, grounding is a synchronic notion of metaphysical dependence. Grounding is about the hierarchical structure of the world: how abundant entities at a derivative level metaphysically depend on sparse entities at a lower level; and metaphysics aims to uncover the connecting principles between the grounded and the grounds. Similarly, just as basing backs causal explanation, grounding backs metaphysical explanation. To explain, as Kim noted (1994: 67–69), is to trace patterns of dependence, and dependence comes in such distinct flavours. A causal explanation of your pain traces the way the state causally depends on the damage to your tissue, but a metaphysical explanation traces the way it metaphysically depends on, say, your C-fibres firing. In the first case, the explanans and explanandum are connected through causal mechanisms, but in the second case, they are connected through principles of grounding. For grounding to have such explanatory power requires that it not be a relation of mere modal covariance or entailment, for example, strong supervenience or existential counterfactual dependence. Similar to Horgan’s (1993) insistence that any physicalist claim of ontological supervenience be robustly explainable in a physicalistically acceptable way (what he dubbed ‘superdupervenience’),

From Epistemic Basing to Epistemic Grounding

257

any modal claim of inter-level dependence should be robustly explainable by relations of grounding. Fine (2001), Schaffer (2009) and Jenkins (2011) suggest grounding itself is an explanatory relation. Entailment and supervenience are intensional, but if the grounding locution is regarded as hyper-intensional, then not only can grounding be differentiated from these mere modal relations, perhaps grounding is also inherently explanatory.16 After all, explanation is hyper-intensional. To see that both notions are hyper-intensional consider Fine’s example (1994) that ‘{F} because F’ is true, but intersubstitution yields the falsehood ‘F because {F}’, where {F} is the singleton set of F. That holds irrespective of whether ‘because’ signals explanatory or grounding relations. However, as Schaffer (2016: 73) noted, ‘F’ and ‘{F}’ are intensionally inequivalent, if, as is arguably the case, the set of metaphysically possible worlds differs from the set of set-theoretically possible worlds. Mixing the relata of grounding with abstract and concrete entities may thus fail to illustrate the hyper-intensionality of grounding. Be that as it may, we follow Schaffer (2012: 124) in taking grounding to be a worldly dependency relation that underwrites, or is tracked by, metaphysical explanation:17 One should distinguish the worldly relation of grounding from the metaphysical explanations between facts that it backs, just as one should distinguish the worldly relation of causation from the causal explanations between facts that it backs. Let’s briefly ponder the logic of grounding. The standard view is that grounding is a two-place, factive relation between worldly entities: if [p] is grounded in [Δ], then [p] and [Δ] obtain.18 Note that [Δ] may itself be a plurality of facts, for example, [p & q] is grounded in [p], [q]. But grounding is non-monotonic, just as explanation is. It does not follow from the grounding of [p] in [Δ] that for any [q] compatible with [p] and [Δ], [p] is also grounded in [q & Δ]. A counterexample is easy to construct when [q] is irrelevant to [p] and [Δ].19 Moreover, while grounding cannot be given an exhaustive analysis in modal terms, grounding has modal consequences: if [p] is grounded in [Δ], then [Δ] metaphysically necessitates [p], in that any metaphysically possible [Δ]-world is a [p]-world. But only full grounds carry such necessity: [p & q] is not metaphysically necessitated by its partial ground [Δ].20 Further, grounding is typically taken to be structured by principles of irreflexivity, asymmetry, and transitivity, and so induces a strict partial ordering on the entities in its domain.21 Nothing is grounded in itself; if [Δ] grounds [p], then [p] does not ground [Δ]; and if [Δ] grounds [p], and [p] grounds [q], then [Δ] grounds [q]. All three principles have been challenged in the literature, but for simplicity’s sake we assume they provide structure to the grounding relation.22

258

Jesper Kallestrup

Another important point for our purposes concerns the alleged primitive or irreducible nature of grounding. Here are some representative passages: Grounding is an unanalysable but needed notion—it is the primitive structuring conception of metaphysics. (Schaffer 2009: 364) [T]here is no prospect of a reductive account or definition of the grounding idiom: We do not know how to say in more basic terms what it is for one fact to obtain in virtue of another. So, if we take the notion on board, we will be accepting it as primitive. (Rosen 2010: 113–114) The received view is that grounding is primitive by not being amendable to a reductive analysis,23 which has the advantage of rendering grounding a unified phenomenon. Following Koslicki (2015), one option is to interpret ‘grounded in’ univocally as picking out a single dependency relation across all discourses. Another is to interpret ‘grounded in’ as picking out a genus, that is, a unique generic kind, under which many distinct, specific, dependency relations fall. Both take grounding to be a distinctive relation. To borrow Wilson’s terminology (2014b), call either view big ‘G-grounding’. We agree that ‘grounded in’ cannot be reductively analysed in terms of conditions necessary and sufficient for its application, but that is not tantamount to grounding being metaphysically distinctive. Rather, grounding terminology traffics in a motley of specific dependency relations. Examples of such small ‘g-grounding’ relations include determination, mereological parthood, set membership, functional realisation, type/token identity, and proper subset.24 Following Wilson and Koslicki, we recommend eliminativism about G-grounding, because there is no compelling reason to posit such distinctive relations in addition to the plethora of extant g-grounding relations. Talk of ‘grounded in’ is a schematic placeholder for these more detailed relations, as they resemble each other enough in terms of unifying features to go under that name.25 In particular, G-grounding is not a covering genus having g-groundings as species. By way of analogy, all instances of jade are objectively similar, but jade is not a genus to which the minerals nephrite and jadeite belong as species. One might view G-grounding as a determinable, but only if reducible to a disjunction of its determinate g-groundings.26 The present point is that sufficient unification among g-grounding relations is no reason to posit G-grounding. In fact, there is ample reason for scepticism about G-grounding. Metaphysics is about elucidating the structure of reality, and by telling us how (rather than whether) certain entities exist, metaphysics reveals its nature. The problem is G-grounding is too coarse-grained to furnish such

From Epistemic Basing to Epistemic Grounding

259

metaphysical explanation. Hence, if having metaphysical explanatory power is a test any grounding relation must pass to gain admission to our metaphysical toolkit, then G-grounding should be barred. Suppose, following Schaffer (2017), that G-grounding is what accounts for the metaphysical dependence of the mental on the physical. That provides no answer at all to the question of whether we should endorse (conservative) reduction or non-reduction about the mental. G-grounding is neutral on, or underdetermines, key issues at stake in metaphysical debates over the nature of the mental, for much the same reasons supervenience failed to account for basic metaphysical distinctions. In contrast, g-grounding relations are better suited to do the intended explanatory work, as witnessed by their track record. For example, if the mental is functionally realised by the physical, or if the causal powers of the mental are a proper subset of the causal powers of the physical, then non-reduction is a live option; but if the mental is type- or token-identical to the physical, then corresponding reductions are inevitable. These specific relations have a sufficiently fine-grained structure to illuminate how the mental exists, so as to provide informative answers to pertinent metaphysical questions.27 Against the background of the foregoing, we shall proceed on the assumption that ‘grounded in’ picks out specific g-grounding relations. With this deflationary conception in mind, let’s turn to one such specific relation.

IV The Determination Relation In Section III we took ‘grounded in’ to pick out a relation of metaphysical dependence that can do explanatory work in metaphysics. While that relation is characterised by logical and modal features such as metaphysical necessitation and strict partial ordering, it must itself be robustly metaphysical in order to back metaphysical explanation. Further, we argued that ‘grounded in’ picks out multiple g-grounding relations, rather than a unique G-grounding relation. The latter is too coarse-grained to answer metaphysical questions about how its relata exist. The former, however, possess sufficient structure and fineness of grain to provide such answers. Admittedly, some g-grounding relations lack some of these modal or logical features (just as G-grounding arguably does), but we nonetheless assume they share enough unifying features to warrant our use of the ‘grounded in’ locution, or its cognates, to pick them all out. In any case, we shall focus exclusively on a paradigmatic case of g-grounding, namely the relation of a determinable to its determinates, which offers an attractive framework for theorising about epistemic grounding. Let’s regiment this notion of determination. The guiding thought is that for a determinate to occur is for its determinable, that is, the determinable under which that determinate falls, to

260

Jesper Kallestrup

occur, not simpliciter, but in a specific (or certain) way. More generally, F-ing and a specific way of F-ing are related as determinable to its determinate, where any such relation is relative to a level of specificity, that is, the latter may itself be a determinable relative to a more specific way of F-ing. Colour properties are widely regarded as canonical instances of determination, but diverse types of states, actions, and events may also instantiate the relation.28 Between them, Johnson (1921), Prior (1949), Searle (1959), Yablo (1992), Ehring (1996), Armstrong (1997), Gillett & Rives (2005), and Funkhouser (2006) list approximately ten platitudes with which any set of entities purporting to instantiate the determination relation should accord. The first captures the guiding thought: (i) Specificity. To have a determinate property is to have its determinable properties, not simpliciter, but in a specific way. Being red is a specific way of being coloured. The others are: (ii) Difference. Distinct same-level determinates of a determinable differ in respect of that determinable; where, roughly, two determinates are same-level if they share a common determinable but no common determinates. Red and blue are distinct samelevel determinates of colour, and so are different colours. (iii) Ordering. Determinates under the same determinable systematically resemble each other such that they are comparable in respect of that determinable.29 Red is more similar to orange than it is to yellow. (iv) Strict Partial Order. Determination is transitive, asymmetric and irreflexive. Since scarlet determines red, and red determines colour, scarlet determines colour; scarlet determines red, but red does not determine scarlet; and no colour determines itself. (v) Asymmetric Necessitation. A determinate necessitates all of its determinables, but a determinable does not necessitate any particular determinate. Necessarily, everything that is scarlet is also red, but, possibly, something is red but not scarlet. (vi) No Bare Determinable. Any determinable necessitates the existence of some determinate. Necessarily, anything red is also either scarlet, or burgundy, or crimson, etc. (vii) Non-Conjunctive Determinates. Conjunctions are not determinates of their conjuncts, nor are conjunctions determinates of properties that are determined by only one of their conjuncts. Neither red-and-square nor scarlet-and-square determines red. (viii) Non-Disjunctive Determinables. Disjunctions are not determinables of their disjuncts, nor are disjunctions determinables of properties that determine only one of their disjuncts. Neither red-or-square nor coloured-or-square is determined by red. (ix) Super-Determination. The chains of determinables and determinates are finite at both ends, that is, there is a highest determinable and a lowest determinate. Nothing is determined by colour, and nothing determines, say, Coca-Cola red, or red23. (x) Exclusion. Every determinable has at most one same-level determinate. No coloured object is both (uniformly) red and blue. One may object that some properties are related as determinables to determinates without satisfying all of (i)—(x). Searle (1959: 152) and

From Epistemic Basing to Epistemic Grounding

261

Yablo (op. cit.) offer counterexamples to Exclusion, for example, speeding and reversing are conjoinable determinates of driving.30 But these platitudes should not all be deemed individually necessary for two properties to instantiate the determination relation: ‘determinate’ and determinable’ are philosophical terms of art, and nothing prevents a reading on which two properties stand in that relation just in case they satisfy a weighted majority of these platitudes, or perhaps a Lewis-style cluster, that is, a disjunction of conjunctions of most of them, bearing in mind though that some should be treated as essential. Alternatively, we can follow Wilson (2009: 151–155) in selecting a core subset, from which the others flow except those that are merely typical. With that caveat in mind, determination is a prime candidate for a g-grounding relation. It displays the characteristic logical and modal features, such as a strict partial order and asymmetric necessitation of the grounded by the grounding. The pertinent modality is metaphysical: determinates necessitate their determinables across all metaphysically possible worlds. But just as only full grounds carry metaphysical necessity, so do only full determinates: if [Δ] determines [p] but not [q], then [p] v [q] is metaphysically necessitated, but not determined, by [Δ]. NonDisjunctive Determinables and Non-Conjunctive Determinates show that determination is not mere asymmetric necessitation between combinations of properties, that is, neither disjunctive addition not conjunctive elimination are inference rules that govern determination.31 And determination is not obviously hyper-intensional; at least, F is not a determinate of {F}, and so any necessary equivalence is irrelevant. Moreover, there is no need to postulate primitive fundamentality to fix the direction of priority. For we can say that [Δ] is metaphysically prior to [p] just in case [Δ] necessitates [p] but [p] does not necessitate [Δ].32 It follows from Asymmetric Necessitation that determinates are metaphysically prior to their determinables. In fact, we can use Super-Determination to define at least a relative notion of fundamentality. Say [Δ] is fundamental relative to determinable [p] just in case [Δ] is a determinate of [p], but not a determinable relative to any other determinate of [p]. Coca-Cola red is fundamental relative to colour. Lastly, determination is non-monotonic: it does not follow from the determination of [p] by [Δ] that for any arbitrary [q] compatible with [p] and [Δ], [p] is also determined by [Δ & q]. That follows from Non-Conjunctive Determinates, but is compatible with determination of a single property by a plurality of properties. In contrast with G-grounding, determination is thus a robustly metaphysical relation with specific structure: Specificity, Difference and Ordering are all substantial features beyond the merely logical and modal ones, which can do explanatory work in metaphysics.33 But what is it in virtue of which, say, scarlet falls under red in the specific way it does, cherry and crimson differ in their redness, and merlot is more similar to mahogany than to rose? The determination dimensions are the features by which

262

Jesper Kallestrup

properties are determined, and hence dimensions of variation between determinates of their shared determinable.34 Thus, the determination of any colour requires the definition of three variables: hue, brightness and saturation. These are the dimensions along which colours vary. They are jointly sufficient for distinguishing any colour from any other colour, and so they explain Specificity, Difference and Ordering: differences in hue, brightness, and saturation explain how red is a specific colour, how red and blue are different colours, and how red resembles orange more than green. In general, the range of determination dimension values for some determinates is a proper subset of the range of such values for their determinable. For example, the range of values for rose is a proper subset of the range of values for red.35 Note finally that distinct determinates of a determinable may share additional features distinct from values along their determination dimensions; some are non-determinable necessities, for example, being four-sided is necessarily true of all quadrangles, and some are non-determinable contingencies, for example, being my granny’s favourite shape is true of squares, but neither are dimensions along which quadrangular is determined.

V Epistemic Grounding Grounding claims are ubiquitous in metaphysics, for example, priority monism takes the nature of wholes to be prior to the nature of their constituent mereological atoms, but debates about grounding are also prominent in ethics and philosophy of mind, for example, non-reductive naturalists claim moral properties of our world constitutively depend on its natural properties. Grounding, or its notational variants, have so far not received the attention it deserves in epistemology. Sections III and IV provided the theoretical resources to begin to fill the lacuna. Our contention is that determination captures the epistemic g-grounding of knowledge, where such relation underwrites metaphysical explanation in epistemology.36 In Section II we sketched a causal (-cum-cognitive) view of epistemic basing, but properly based belief is no candidate for a g-ground of knowledge, even if true. What g-grounds knowledge metaphysically necessitates knowledge, which justified (true) belief does not. A better tack is to identify the g-grounds of knowledge with ways of knowing, as there is a striking analogy with colour and ways of being coloured. Here is Williamson (2000: 34): If something is coloured, then it has a more specific colour property; it is red or green or. . . . Similarly, if one knows that A, then there is a specific way in which one knows; one can see or remember or . . . that A.

From Epistemic Basing to Epistemic Grounding

263

Given that the relation between colour and ways of being coloured is relevantly similar to the relation between knowing and ways of knowing, the fact that the former are related as determinable to determinates suggests the latter are also thus related. In fact, our guiding thought, as encoded by Specificity, was that (specific) ways of -ing are determinates of -ing, and so for seeing (that p) to be a way of knowing is for seeing to be a determinate of knowing. Moreover, since determination is a paradigmatic g-grounding relation, seeing and other ways of knowing g-ground knowledge. Our remaining argument is as follows: determinates g-ground their determinables; seeing, etc., and knowing are related as determinates to their determinables, which means seeing, etc., constitute ways of knowing; hence, seeing and other ways of knowing g-ground knowing. Having established the first premise in Sections III and IV, we turn now to the second premise. Dretske’s (1969: 82ff) distinction between objectual and epistemic seeing is instructive.37 Seeingo involves an object o being visually differentiated from its environment by being experienced as looking some way to S, even though S need not believe there is an o. In contrast, Seeinge requires not just that S (truly) believes there is an o, but also that such belief has whatever properties are necessary and sufficient for knowledge. Suppose, as an example, that S seese that there is an o if and only if (i) there is an o which S seeso; (ii) the conditions under which S seeso the o are such that S would have a visual experience as of an o only if there were an o; and (iii) S’s belief that there is an o is true mainly because based on her visual experience as of an o. Such analysis of seeinge in terms of (i)—(iii) departs from Dretske (op. cit.) in three respects. First, in place of the safety condition (ii), Dretske required that the conditions under which S seeso be such that S would not have a visual experience as of an o unless there were an o. Second, Dretske required additional belief that this sensitivity condition be met (which seems unduly intellectual). Third, the virtue-theoretic condition (iii) is absent on Dretske’s account (but is needed to accommodate cases of unsafe knowledge).38 Similar accounts mutatis mutandis can be developed of epistemic hearing, or of epistemic perceiving. Obviously, different theories of knowledge will place different conditions on epistemic seeing, for example, knowledge-first epistemologists would reject the sufficiency of (i)—(iii) for knowledge. Our aim is not to defend a particular theory of epistemic seeing, but merely to insist that whichever epistemic conditions apply to knowledge also apply to such seeing. Equipped with the notion of epistemic seeing, and by extension epistemic hearing, etc., and perceiving more generally, let’s now revisit our platitudes: (i*) Specificity. Seeinge is a specific way of perceivinge, and perceivinge is a specific way of knowing. (ii*) Difference. Seeinge and hearinge are distinct (same-level) ways of perceivinge, and hence of knowing.

264

Jesper Kallestrup

(iii*) Ordering. Seeinge, tastinge and hearinge are orderable in respect of perceivinge, and hence of knowing. (iv*) Strict Partial Order. Since seeinge determines perceivinge, and perceivinge determines knowing, seeinge determines knowing. But since seeinge determines perceivinge, perceivinge does not determine seeinge. Furthermore, seeinge does not determine itself. (v*) Asymmetric Necessitation. Necessarily, if S seese, then S perceivese, and hence knows, but it is possible for S to know without perceivinge and hence without seeinge.39 (vi*) No Bare Determinables. Necessarily, if S knows, then there exists a specific way of knowing such that S knows in that way. (vii*) Non-Conjunctive Determinates. Seeinge-and-hearinge is not a way of seeinge. (viii*) Non-Disjunctive Determinables. Seeinge is not a way of seeinge-or-hearinge. (ix*) Super-Determination. Seeing-througha-microscopee is a maximally determinate way of knowing, and knowing is a maximally determinable. (x*) Exclusion. If S knows, there is only one same-level way of knowing in which S knows. The upshot is that seeinge, etc., and knowing are related as determinates to their determinable, and since for the former to constitute ways of knowing is for them to be related to the latter as determinates to their determinable, these more specific states constitute ways of knowing.40 As we already established that the determination relation is an exemplary instance of g-grounding, it follows that these ways of knowing g-ground knowing. Since such epistemic g-grounding is an instance of the determination relation, it exhibits all the intuitive grounding features which, as detailed in Section IV, characterise that relation. We can now put this notion of epistemic g-grounding to explanatory work in epistemology.41 As we saw in Section II, the question ‘why does S know p?’ is a request for epistemically relevant information about what caused S to be in a state of knowledge. To cite the reason for which S believes p is to answer that question, but only if that reason provides (undefeated, knowledge-level) justification, and is causally responsible, for S’s belief. Epistemic basing backs causal explanations in epistemology. But asking the question ‘how does S know p’ is a request for epistemically relevant information about what it is in virtue of which S knows. To cite the way in which S knows p is to answer that question.42 Ways of knowing, qua epistemic g-grounds, have a sufficiently fine-grained structure to do such explanatory work: Specificity, Difference and Ordering are features, which detail how seeinge, hearinge, etc., are specific ways of knowing which differ from one another and are orderable in respect of knowledge. The determination dimensions for knowledge are the dimensions of variation between ways of knowing with knowledge as their shared determinable, that is, the dimensions along which knowledge varies. As they are jointly sufficient for distinguishing any ways of knowing from any other, they explain Specificity, Difference and Ordering. So, what are the variables that determine knowledge? Again, epistemologists will no doubt disagree, but here’s a tentative proposal: (i) degree of

From Epistemic Basing to Epistemic Grounding

265

epistemic certainty. Following Stanley (2008), epistemic certainty of p is knowing p on the basis of evidence that gives the highest degree of justification for belief in p. But knowing p requires merely some degree above a salient threshold of justification for belief, where strength of justification can be cashed out modally. To believe truly on the basis of strong justification is to track the truth in a range of nearby worlds, and the further away the truth is tracked (out to the first non-p world), the stronger the justification is. Variation in justificational strength above the threshold makes for different ways of knowing. (ii) Epistemic source. Seeinge and hearinge differ with regards to knowing by way of originating from different epistemic sources of visual and auditory perception, but sources could also pertain to vehicles for storage, such as biological memory or an extended memorial process.43 Variation in epistemic source constitutes a difference in ways of knowing. In many cases, variation in epistemic source implies variation in degree of epistemic certainty, and so epistemic determination dimensions are not strongly independent in the way, for some colour spaces, the determination dimensions for colours are. But there are presumably cases where variation in epistemic source entails variation in ways of knowing while holding the degree of epistemic certainty fixed. So, epistemic determination dimensions are weakly independent: some variation along one dimension does not entail variation along another dimension. Without delving into any details, other dimensions for knowledge may include types of epistemic defeat, degree of confidence or propositional content.44 Note also that knowledge and ways of knowing share certain non-determinable necessities, for example, being true and entailing belief. Likewise, there is a multitude of non-determinable contingencies, for example, being the most common way of knowing may be true of seeinge. Before we close a few additional objections deserve brief discussion. The first is that seeinge and knowing cannot instantiate the determination relation, because not all platitudes governing that relation hold of these properties. Thus, Exclusion and Super-Determination seem to have counterexamples: seeinge and hearinge are conjoinable ways of knowing, and seeing-through-a-microscopee is itself determined by seeing-throughan-electron-microscopee, which in turn is determined by seeing-througha-transmission-electron-microscopee, and so on possibly ad infinitum. The reply, to reiterate from Section IV, is that these platitudes need not all be deemed individually necessary; rather, we are at liberty to adopt, say, a Lewis-style cluster approach with the caveat that some platitudes may be considered indispensable, or at least carry more weight than others. Such a view permits the epistemic overdetermination of knowledge, and accommodates the possible absence of a fundamental epistemic level, given that neither Exclusion nor Super-Determination should be treated as essential. Second, by reference to linguistic infelicity, one may claim that seeinge cannot constitute a way of knowing, for while ‘S is seeing that p’ is

266

Jesper Kallestrup

grammatical, ‘S is knowing that p’ is not. The answer is to concede that specific ways of knowing may lack suitable names in language, but that we could introduce them by ostensive definition of a paradigm.45 Suppose S knows at t1 in virtue of seeinge at t1. At t2 S ceases to seeo, and so at t2 S ceases to seee, but typically S would continue to know at t2, thanks to memory.46 We lack natural names to articulate the subtle ways in which memory meshes with seeing, but that is perfectly consistent with the existence of such complex ways of knowing. Third, fake barn county,47 and other cases of environmental luck, show that S can see that there’s a barn without knowing that proposition; hence, seeing fails to count as a g-ground of knowing. One answer is to insist that while S seeso the barn, or, as Williamson (2000: 38) puts it, seeso a situation in which there’s a barn (and may also truly and justifiably believe that there’s a barn), S does not see, just as S does not know, that there’s a barn. Our linguistic intuitions do not reliably track such fine distinctions.48 A different tack is to maintain that S both sees and knows that there’s a barn. Bear in mind at this juncture that if, as on Sosa’s view (2010), knowledge is true belief because of cognitive ability, then S has (“animal”) knowledge in the fake-barn case.49 Such a claim is supported by intuitional evidence from experimental studies; or so Turri (2016, 2017) claims. The key point is that either way Gettier cases involving environmental luck drive no wedge between seeing and knowing.50 Fourth, one may find it objectionable that seeinge, as defined, is inapt to serve as the basis for a belief, because no basis for belief can entail that very belief.51 But, in reply, the source of the worry is unclear. Having a visual experience as of p, or seeingo, or even a perceptual belief, may be a basis for a (distinct) belief. All these may constitute a justifying reason for which S forms a belief, but nothing we said about epistemic basing in section II suggests that epistemic seeing, no matter how that notion is understood, must also be included amongst the reasons for which a belief is based. On our view, such seeing is a ground of knowledge, which is never a basis for a corresponding belief, nor, as mentioned earlier, is a basis for belief ever a ground of corresponding knowledge. As we have been at pains to argue, the way belief relates to its basis is fundamentally different from the way belief relates to its ground. These should not be lumped together as they pertain to importantly distinct, epistemological projects.

VI Concluding Remarks We have argued that epistemic basing is what causally explains knowledge, whereas epistemic grounding is what metaphysically explains knowledge. But, just as with basing, grounding can do the explanatory work only if sufficiently specific and fine-grained. In that regard, G-grounding is a non-starter, but the determination relation holds promise, and we

From Epistemic Basing to Epistemic Grounding

267

demonstrated how epistemic seeing, and by extension other ways of knowing, and knowledge are related as determinates to their determinable. Since such epistemic determination displays all the hallmarks of what any grounding relation intuitively amounts to, epistemic seeing, and other ways of knowing, are g-grounds of knowledge. That means knowledge is connected to its epistemic g-grounds through the principles that govern the determination relation; in particular, knowledge is a derivative state relative to more fundamental ways of knowing on which it metaphysically depends. Whether there are absolutely fundamental ways of knowing such that epistemic grounding is well-founded is not a question we shall settle here; suffice it to say that if such states exist they are non-primitive super-determinates. Knowledge itself, however, is a superdeterminable, as there is no epistemic state for which knowledge is a determinate. Basing is what converts propositional justification into doxastic justification, but grounding is not what turns doxastic justification into knowledge. We offered an illustrative analysis of epistemic seeing, but our claim was merely that whichever epistemic conditions are taken to hold of knowledge will also apply to such seeing. Epistemic seeing is not amendable to a reductive analysis if knowledge is not. Put differently, a true theory of epistemic grounding is no reductive analysis of knowledge, let alone the concept thereof, and so yields no solution to the Gettier problem. Grounding may help with what Lycan (2006) dubbed the Gettier problem problem, that is, the problem of explaining what is distinctively wrong with the Gettier project. For our diagnosis suggests at least that this project has mistaken a possibly unsolvable problem about how to reductively analyse knowledge with a demonstratively solvable problem about how to explicate the epistemic ground of knowledge. So, one distinctive failure of the Gettier project is to ignore the possibility of providing a metaphysical explanation of knowledge without having to specify necessary and sufficient conditions for knowledge. Our aim was primarily to present a framework for theorising about the epistemic ground, and attendant metaphysical explanation, of knowledge, but future work in the metaphysics of knowledge should be devoted to further study of such ground and explanation.52

Notes 1. For dissent see Sylvan and Lord (this volume). Korcz (1997), Neta (2011) and Sylvan (2016) are excellent survey articles on epistemic basing. 2. We shall focus exclusively on the epistemic grounding of knowledge; there is a different but related question about the epistemic grounding of (doxastic) justification. 3. We restrict attention to prima facie reasons, justification and support, and we assume that the support that r provides for p is sufficient to prima facie justify belief in p.

268

Jesper Kallestrup

4. Neta (2011). 5. Alston (1989), Kvanvig (2003), Pollock & Cruz (1999) and Korcz (2000). 6. Lehrer’s Gypsy-Lawyer (1971) is a famous case intended to show that causation is unnecessary for basing. Suppose an epistemically bad reason r initially causes S to believe p. Later S acquires an epistemically good reason r* to believe p (which is true), but even though S recognises r* as a good reason to believe p, S’s belief continues to causally depend only on r. Since intuitively at the later point S’s belief is justified, it is causally independent of the reason on which it is based. Goldman (1979) and Audi (1983) deny the intuition that S’s belief is justified; one problem is that S would believe p even if p were false, another is that S has not properly taken r* into account when forming the belief. Swain (1981) proposed a much-discussed counterfactual analysis of causation, which offers a solution to this problem, while preserving the spirit of the causal theory of the basing relation. 7. McCain (2012). 8. Hall (2000) argued against the transitivity of causation. 9. Turri’s more general argument is that the tables should be turned such that propositional justification is explained in terms of doxastic justification. 10. As in Lehrer (1971) and Swain (1981). 11. Even if James’ belief were taken to be based twice over by distinct reasons, we would still need an explanation of how the priest’s testimony can be a reason why James holds the belief when that testimony plays no causal role in accounting for its formation (or maintenance). 12. Korcz’s (2000) causal-doxastic theory involves a disjunctive analysis of basing, but since both disjuncts include causal constraints, (PRIEST) presents no difficulty. 13. More generally, why-questions are typically requests for information about causes or reasons (or both), as in ‘why did the plane crash?’ or ‘why did Jill and Jim marry? See also fn. 42. 14. Schaffer (2009: 345) distinguishes between Quinean metaphysics, asking ‘what is there?’ and Aristotelian metaphysics, asking ‘how is what there is structured?’. 15. Rosen (2010) and Cameron (2008). 16. ‘Intensional’ is understood in terms of contexts into which co-extensional terms may fail to preserve truth, whereas ‘hyper-intensional’ is understood in terms of contexts into which co-intensional, i.e. necessarily co-extensive, terms may fail to preserve truth. 17. See also Audi (2012: 687–678) and Rodriguez-Pereyra (2005: 28). 18. For grounding to be factive, its relata should be facts or bearers of values. Moreover, since the relata of explanation are facts, grounding should also pertain to facts if it is to back explanation. Still, following Cameron (2008) and Schaffer (2009), we shall be liberal about which ontological entities may enter into the grounding relation. For instance, just as [Socrates exists] grounds [{Socrates} exists], saying that [Socrates] grounds {Socrates}, or that [being Socrates] grounds [being {Socrates}], is permissible. 19. In that respect grounding is similar to causation but different from logical entailment which is preserved under arbitrary expansion of the premise set. 20. See Audi (2012) and Dasgupta (2014), but also Leuenberger (2014) for dissent. For example, one may object that [Xanthippe became a widow] is grounded in, but not necessitated by, [Socrates died]. The reply would be that the former fact is grounded in two facts [Socrates died] and [Socrates and Xanthippe were married]. 21. Cameron (2008), Raven (2013) andRosen (2010). 22. Rodriguez-Pereyra (2005), Fine (2012), Jenkins (2011) and Schaffer (2012).

From Epistemic Basing to Epistemic Grounding

269

23. See also Fine (2001, 2012) and Rodriguez-Pereyra (2005). 24. Hofweber (2009). 25. Most induce a strict partial order, but all have similar modal and explanatory implications. Observe also that some non-grounding relations are strict orderings, e.g. ‘taller-than’. 26. See also Fine (2012). 27. Is G-grounding needed to fix direction of priority, i.e. to characterise the fundamental as the ungrounded and the derivative as the grounded? Wilson’s answer (2014b: 560–563) is that eliminativists about G-grounding should take the fundamental as metaphysically inexplicable, analogous to axioms in a theory. On her view, the fundamental qua primitive fixes direction of priority, and so is hyper-intensional. But as we shall see in Section IV, g-grounding claims do not necessarily need supplementation with basic assumptions about what is fundamental. 28. Johnson (1921: 174) and Yablo (1992: 260–261) mention shutting/slamming, killing/stabbing, speaking/lying and realising/discovering. 29. Johnson (1921: XII, §1; cf. XI, §4) took ‘comparable’ to be synonymous with ‘belonging to the same determinable’. 30. And being true and being false are conjoinable determinates of truth-value, if such gluts are embraced. See also Johnson (1921: XI, §4), Prior (1949, 190) and Armstrong (1997: 48–49). McGrath (1998: 171–172) suggests being round is a way of being round or square, as a counterexample to Non-Disjunctive Determinables. 31. See Audi (2012) and Dasgupta (2014), but also Leuenberger (2014) for dissent. For example, one may object that [Xanthippe became a widow] is grounded in, but not necessitated by, [Socrates died]. The reply would be that the former fact is grounded in two facts [Socrates died] and [Socrates and Xanthippe were married]. 32. See also Armstrong (1997: 50). Wilson (2014b) worries that asymmetric necessitation is insufficient to fix metaphysical priority for the same reason that supervenience is too weak to guarantee nothing-over-and-aboveness. But there are important, relevant differences in that supervenience, unlike determination, is a reflexive and non-asymmetric relation, in fact asymmetry entails irreflexibility, and is entailed by transitivity and irreflexivity. Wilson also suspects asymmetric necessitation may be unnecessary to fix metaphysical priority, because of Fine’s example of F grounding {F}. However, as noted in Section III, there may be good reason to think that ‘F‘ and ‘{F}’ are intensionally inequivalent. 33. There is a separate ontological question about the reality of determinables: anti-realists, e.g. Gillett & Rives (2005), hold that only maximally determinate properties exist; reductionists, e.g. Antony (2003), hold that determinables are real but identical to a disjunction of their determinates; and non-reductionists, e.g. Wilson (2012), hold that determinables exist as distinct from any logical construction of their determinates. Much of the disagreement revolves around the question of whether super-determinates exclude putative causal powers of their determinables, as gaining admission to our ontology, following the Eleatic Principle, involves passing a causal test. Suffice it to say that our claim that determinates are metaphysically prior to their determinables is consistent with these realist ontologies of determinables, indeed even Yablo’s (1992: 259) denial of any causal competition between determinables and determinates is compatible with such metaphysical priority. 34. The following owes much to Funkhouser (2006), but see also Johnson (1921: I, xi, 4).

270

Jesper Kallestrup

35. Funkhouser (2006) proposes that the determination dimensions be represented as the axes of an n-dimensional space. So, the property space for colour is the whole three-dimensional space, and the property space for red is a subregion of that space. Any point in that region corresponds to a specific shade of red. See also Wilson (2009: 161–164) for an argument from metamers to the effect that the determination dimensions for colours are science-relative. But bear in mind that not all properties instantiating the determination relation need share non-trivial determination dimensions. Being 195 cm and being 200 cm are both determinates of being tall, but they seem not to have any such interesting dimensions in common. 36. We need not deny that knowledge is also non-epistemically g-grounded, but any such g-grounding of knowledge will not underwrite explanations in epistemology. 37. Or rather primary epistemic seeing. Dretske (op. cit.) also introduced a notion of secondary epistemic seeing, where S sees that object o is F without seeing o itself. 38. See for example Comesaña (2005). 39. S‘s knowledge may be determined by suitable notions of epistemic introspection, reflection or memory. Note also that Asymmetric Necessitation is different from what Cassam (2007: 339) calls the entailment view: -ing is a way of knowing iff ‘S s that P’ entails ‘S knows that p’. On our view, entailment is insufficient for a state to count as a way of knowing. As Unger observed (1975: 158–162), regretting, admitting and revealing are all counterexamples to the entailment view, but most of our platitudes are false of these states. 40. The relation between knowing and ways of knowing should not be assimilated to the relation between species and genus. Following Searle (1959: 143) and Armstrong (1997: 49), the latter obtains when a conjunctive definition of the form X =def YZ is possible, where the genus Y and the differentia Z are logically independent, e.g. ‘man’ means rational animal. But if seeing were a species, the genus knowledge would entail at least part of the differentia, e.g. belief. 41. Following on from fn. 33, there is also a separate question specifically about the ontological status of knowledge qua determinable, which we shall not settle here. Our agenda, to repeat, is to explore the metaphysical question of the ground of knowledge. 42. Likewise, to use the examples from fn. 13, to cite the way in which the accident happened and the way in which the couple married provide answers to the questions ‘how did the plane crash?’ and ‘how did Jill and Jim marry?’. Having said that, there may be some conversational contexts in which an answer to a why-question is also a satisfactory answer to a how-question. 43. See Clark & Chalmers (1998), and author (b). 44. Menzies (2008) suggests the last two as determination dimensions for belief. 45. Williamson (2000: 34). 46. Cassam (2007: 350–351) objects that seeing that p is rather a way of coming to know p. After all, unlike knowing, seeing occurs at specific times and places, and may be associated with phenomenological characters that knowledge lacks. The reply is that seeinge can be a way of knowing without knowing instantiating every property instantiated by seeinge. Properties of time, place, etc., are irrelevant for seeinge to qualify as a way of knowing, as they are not values along a dimension of determination. Compare: blue is a way of being colored, but not all properties of blue are properties of color. Being the color of the sky is a property of blue, but not a property of color. In any case, we are not denying that ways of coming to know exist, but only that seeinge is one of them.

From Epistemic Basing to Epistemic Grounding

271

47. Goldman (1976). 48. See also the exchange between Turri (2010b) and French (2012, 2013). 49. Obviously, our tentative account of seeinge would need revision so as not to include a safety (or sensitivity, for that matter) condition. 50. Pritchard (2012: 23–27) argues that defeater defeaters show the possibility of seeing that p without knowing p. If S is looking at a barn in normal environmental circumstances, but is told by a reliable informant that S is currently being deceived, then S sees that there’s barn, yet S does not believe, and hence does not know, that there’s a barn. For if S were to subsequently discover that the testimony was false, S would retrospectively treat herself as having seen exactly that. The response is that S both saw and knew that there was a barn. As Rose and Schaffer argued (2013), knowledge requires dispositional belief. Upon seeingo the barn, S forms the dispositional belief that there’s barn, which is temporarily masked by the misleading testimony she receives from a reliable source. After all, S is able to draw non-accidentally on the stored perceptual information that is available to mind. Only when the falsity of the testimony transpires does S consciously endorse the content of her dispositional belief, and thus forms an occurrent belief. 51. Pritchard (2012: 21) calls this the basis problem for the claim that seeing that p entails knowing that p. See also Chudnoff (2011). 52. I am grateful to the participants at workshops in St. Andrews and Copenhagen for helpful discussion. Special thanks to Adam Carter, Pat Bondy and Jessica Wilson for detailed comments on an earlier draft.

Bibliography Alston, W. (1989). ‘Internalism and Externalism in Epistemology’, in W. Alston (ed.), Epistemic Justification: Essays in the Theory of Knowledge, Ithaca, NY: Cornell University Press, 186–226. Antony, L. (2003). ‘Who’s Afraid of Disjunctive Properties?’, Philosophical Issues 13: 1–21. Armstrong, D. (1997). A World of States of Affairs, Cambridge: Cambridge University Press. Audi, P. (2012). ‘Grounding: Toward a Theory of the In-Virtue-Of Relation’, Journal of Philosophy 109: 685–711. Audi, R. (1983). ‘The Causal Structure of Indirect Justification’, Journal of Philosophy 80: 398–415. Block, N. (2015). ‘The Canberra Plan Neglects Ground’, in Horgan, T., Sabates, M. and Sosa, D. (eds.), Qualia and Mental Causation in a Physical World: Themes from the Philosophy of Jaegwon Kim, Cambridge: Cambridge University Press, 105–133. Cameron, R. (2008). ‘Turtles All the Way Down: Regress, Priority, and Fundamentality’, The Philosophical Quarterly 58: 1–14. Cassam, Q. (2007). ‘Ways of Knowing’, Proceedings of the Aristotelian Society 107: 339–358. Chudnoff, E. (2011). ‘The Nature of Intuitive Justification’, Philosophical Studies 153: 313–333. Clark, A. and Chalmers, D. (1998). ‘The Extended Mind’, Analysis 58: 7–19. Comesaña, J. (2005). ‘Unsafe Knowledge’, Synthese 146: 395–404. Dasgupta, S. (2014). ‘On the Plurality of Grounds’, Philosophers’ Imprint 14: 1–28.

272

Jesper Kallestrup

Dretske, F. (1969). Seeing and Knowing, Chicago: University of Chicago Press. Ehring, D. (1996). ‘Mental Causation, Determinables, and Property Instances’, Noûs 30: 461–480. Fine, K. (1994). ‘Essence and Modality’, Philosophical Perspectives 8: 1–16. ——— (2001). ‘The Question of Realism’, Philosophers’ Imprint 1: 1–30. ——— (2012). ‘A Guide to Ground’, in Correia, F. and Schnieder, B. (eds.), Metaphysical Grounding: Understanding the Structure of Reality, Cambridge: Cambridge University Press, 37–80. French, G. (2012). ‘Does Propositional Seeing Entail Propositional Knowledge?’, Theoria 78: 115–127. ——— (2013). ‘Perceptual Experience and Seeing That P’, Synthese 190: 1735–1751. Funkhouser, E. (2006). ‘The Determinable-Determinate Relation’, Noûs 40: 548–569. Gillett, C. and Rives, B. (2005). ‘The Nonexistence of Determinables: Or, a World of Absolute Determinates as Default Hypothesis’, Noûs, 39: 483–504. Goldman, A. (1976). ‘Discrimination and Perceptual Knowledge’, Journal of Philosophy 73: 771–791. ——— (1979) ‘What Is Justified Belief’, in Pappas, G. (ed.), Justification and Knowledge, Dordrecht: D. Reidel, 1–24. ——— (2001). ‘Internalism Exposed’, in Kornblith, H. (ed.), Epistemology: Internalism and Externalism, Malden, MA: Blackwell, 207–230. Hall, N. (2000). ‘Causation and the Price of Transitivity’, Journal of Philosophy 97: 198–222. Hofweber, T. (2009). ‘Ambitious, Yet Modest, Metaphysics’, in Chalmers, D., Manley, D. and Wasserman, R. (eds.), Metametaphysics, Oxford: Oxford University Press, 260–289. Horgan, T. (1993). ‘From Supervenience to Superdupervenience: Meeting the Demands of a Material World’, Mind 102: 555–586. Jenkins, C. (2011). ‘Is Metaphysical Dependence Irreflexive?’, The Monist 94: 267–276. Johnson, W.E. (1921). Logic (Part 1), Cambridge: Cambridge University Press. Kim, J. (1994). ‘Explanatory Knowledge and Metaphysical Dependence’, Philosophical Issues 5: 51–69. Korcz, K.A. (1997). ‘Recent Work on the Basing Relation’, American Philosophical Quarterly 34: 171–191. ——— (2000). ‘The Causal-Doxastic Theory of the Basing Relation’, Canadian Journal of Philosophy 30: 525–550. Koslicki, K. (2015). ‘The Coarse-Grainedness of Grounding’, Oxford Studies in Metaphysics 9: 306–344. Kvanvig, J. (2003). ‘Justification and Proper Basing’, in Olsson, E. (ed.), The Epistemology of Keith Lehrer, Dordrecht: Kluwer, 43–62. Lehrer, K. (1971). ‘How Reasons Give Us Knowledge, or the Case of the Gypsy Lawyer’, The Journal of Philosophy 68: 311–313. Leuenberger, S. (2014). ‘Grounding and Necessity’, Inquiry 57: 151–174. Lycan, W. (2006). ‘On the Gettier Problem Problem’, in Hetherington, S. (ed.), Epistemology Futures, Oxford: Oxford University Press, 148–168. McCain, K. (2012). ‘The Interventionist Account of Causation and the Basing Relation’, Philosophical Studies 159: 357–382. McGrath, M. (1998). ‘Proportionality and Mental Causation: A Fit?’, Noûs 32: 167–176.

From Epistemic Basing to Epistemic Grounding

273

Menzies, P. (2008) ‘The Exclusion Problem, the Determination Relation, and Contrastive Causation’, in Hohwy, J. and Kallestrup, J. (eds.), Being Reduced: New Essays on Reduction, Explanation, and Causation, Oxford: Oxford University Press, 196–217. Moser, P. (1989). Knowledge and Evidence, Cambridge, UK: Cambridge University Press. Neta, R. (2011). ‘The Basing Relation’, The Routledge Companion to Epistemology, New York: Routledge, 109–118. Plantinga, A. (1993). Warrant: The Current Debate, New York: Oxford University Press. Pollock, J. and Cruz, J. (1999). Contemporary Theories of Knowledge, New York: Rowman & Littlefield. Prior, A. (1949). ‘Determinables, Determinates, and Determinants (I, II)’, Mind, 58: 1–20, 58: 178–194. Pritchard, D. (2012). Epistemological Disjunctivism, Oxford: Oxford University Press. Raven, M. (2013). ‘Is Ground a Strict Partial Order?’, American Philosophical Quarterly 50: 191–199. Rodriguez-Pereyra, G. (2005). ‘Why Truthmakers?’, in Beebee, H. and Dodd, J. (eds.), Truthmakers: The Contemporary Debate, Oxford: Oxford University Press, 17–31. Rose, D. and Schaffer, J. (2013). ‘Knowledge Entails Dispositional Belief’, Philosophical Studies 166, 19–50. Rosen, G. (2010). ‘Metaphysical Dependence: Grounding and Reduction’, in Hale, B. and Hoffman, A. (eds.), Modality: Metaphysics, Logic, and Epistemology, Oxford: Oxford University Press, 109–136. Schaffer, J. (2009). ‘On What Grounds What’, in Manley, D., Chalmers, D. and Wasserman, R. (eds.), Metametaphysics: New Essays on the Foundations of Ontology, Oxford: Oxford University Press, 347–383. ——— (2012). ‘Grounding, Transitivity, and Contrastivity’, in Correia, F. and Schnieder, B. (eds.), Metaphysical Grounding: Understanding the Structure of Reality, Cambridge: Cambridge University Press, 122–138. ——— (2016). ‘Grounding in the Imagine of Causation’, Philosophical Studies 173: 49–100. ——— (2017). ‘The Ground between the Gaps’, Philosophers’ Imprint 17, 1–26. Searle, J. (1959). ‘Determinables and the Notion of Resemblance’, The Aristotelian Society, Supplement 33: 141–158. Sosa, E. (2010). Knowing Full Well, Princeton: Princeton University Press. Stanley, J. (2008). ‘Knowledge and Certainty’, Philosophical Issues 18: 35–57. Swain, M. (1981). Reasons and Knowledge, Ithaca: Cornell University Press. Sylvan, K. (2016). ‘Epistemic Reasons II: Basing’, Philosophy Compass 11: 377–389. Turri, J. (2010a). ‘On the Relationship between Propositional and Doxastic Justification’, Philosophy and Phenomenological Research 80: 312–326. ——— (2010b). ‘Does Perceiving Entail Knowing?’, Theoria 76: 197–206. ——— (2016). ‘Knowledge and Assertion in “Gettier” Cases’, Philosophical Psychology 29: 759–775. ——— (2017). ‘Knowledge Attributions in Iterated Fake Barn Cases’, Analysis 77: 104–115. Unger, P. (1975). Ignorance: A Case for Scepticism, New York: Oxford University Press.

274

Jesper Kallestrup

Williamson, T. (2000). Knowledge and Its Limits, Oxford: Oxford University Press. Wilson, J. (2009). ‘Determination, Realization, and Mental Causation’, Philosophical Studies 145: 149–169. ——— (2012). ‘Fundamental Determinables’, Philosophers’ Imprint 12: 1–17. ——— (2014a). ‘No Work for a Theory of Grounding’, Inquiry 57: 1–45. ——— (2014b). Wilson, Jessica, ‘Determinables and Determinates’, Zalta, E. N. (ed.), The Stanford Encyclopedia of Philosophy (Spring 2017 Edition) URL = . Yablo, S. (1992). ‘Mental Causation’, The Philosophical Review 101: 245–280.

15 Well-Founded Belief and the Contingencies of Epistemic Location Guy Axtell

1 Epistemic Location and the Epistemology of Nurtured Beliefs Many of the beliefs that people hold dearest to their sense of personal and social identity are beliefs profoundly shaped by their own upbringing. The impact of a person’s place in time, their familial, communal, cultural, and geographic/demographic setting are some of these contributory causes, but to a varying degree an agent’s controversial view might be highly conditioned by personal temperamental factors or other pragmatic constraints on an agent’s beliefs. We will follow J. Adam Carter (2018) and others who use the term controversial views to refer not just to one’s religious or irreligious views, but also our substantial views in the domains (at least) of morals, politics, and philosophy.2 Nurtured beliefs/opinions are not a ‘domain,’ but the best examples of contrariety among culturally nurtured beliefs fall within these domains. Let a person’s epistemic location refer us to how the individual is located demographically (family; broader culture; class, etc.), in addition to geographically and historically. We will use the term epistemic location problem to highlight etiological challenges to controversial views that bear marks of contingency and of what John K. Davis (2009) terms the impact of trait-dependence upon our cognitive judgments.3 In its most general sense, the epistemic location problem is the problem that our obvious psychographic differences—differences in such things as attitudes, values, and most importantly for this study, what John Rawls terms comprehensive conceptions of the good—are strongly conditioned by contingent matters of the individual’s historic, geographic, and demographic location. Religious identity and whatever beliefs one has stemming from a testimonial faith tradition are a prime example of this familial or cultural inheritance, and the apparent contingency of such nurtured beliefs. Although certainly not the only example, the proximate causes of one’s religious identity and the formation of attendant beliefs are, for most people, a matter of their epistemic location, which in turn appears to be

276

Guy Axtell

an accident of birth.4 Michel de Montaigne gave a version of a ‘contingency’ or epistemic location argument when he wrote, [W]e receive our religion in our own way and by our own hands, and no differently from the way other religions are received. We happen to find ourselves in the country where it has been practiced; or we value its antiquity or the people who have supported it; or we fear the threats it attaches to wrongdoers, or we follow its promises. . . . By the same means another country, other witnesses, similar promises and threats, could in the same way imprint in us a contrary belief.5 Perhaps motivated by a resurgence of social epistemology and epistemology of testimony, and by the increasing importance of understanding group dynamics and the causes of psychographic diversity, philosophers have become increasingly concerned with the epistemology of culturally nurtured beliefs. So there is much agreement that to improve the epistemology of controversial views, philosophers need to focus more carefully on the right target, proper scope, and epistemological force of arguments of the sort that Montaigne and Mill share. Davis’s work takes steps in this direction, as he has plausibly argued (2009) that the concern about the contingency of so many of our testimony-dependent beliefs involves a question about the basing relationship: Does it defeat proper basing if the agent is rationally convinced that she very likely would see her own actual nurtured belief as both false and tainted by unrecognized bias, had she been nurtured in a different culture or epistemic community? George Sher (2001) and Gerald Cohen (2000a, 2000b) are also often credited with igniting a more careful philosophical interest in nurtured belief, and with it, arguments from the contingency of epistemic location. Sher asked us to study “the implications of the fact that even our most deeply held moral beliefs have been profoundly affected by our upbringing and experience—that if any of us had had a sufficiently different upbringing and set of experiences, he almost certainly would now have a very different set of moral beliefs and very different habits of moral judgment.”6 As Sher here indicates, etiological challenges and the contingency anxiety they arouse in those who take them seriously may attach to a far broader group than just religious beliefs.7 Cohen similarly describes ‘paradoxes of conviction,’ paradoxes that he thinks we face in assenting to propositions that we realize, or should realize, are produced and maintained through cultural influences, such that if one had a sufficiently different upbringing they would likely hold different beliefs, attested by different justificatory reasons. Nathan Ballantyne (2015) has also worked on this “pervasive and disconcerting worry about intellectual life: our controversial beliefs regarding morals, politics, religion, and philosophy depend on facts about our personal history.” Yet he is correct to find that in arguments from

Well-Founded Belief

277

contingency both ancient and modern, “pointing is typically all we get— worked-out arguments based on variability are uncommon.”8 To move beyond this, Ballantyne connects serious engagement with the problems posed by Montaigne and Mill not to a generalized skepticism or to dogmatism, but to more concerted attempts to “debunk biased thinkers (including ourselves).” Joshua DiPaolo and Robert Mark Simpson (2016) similarly investigate the epistemic location problem not just with cases described from the armchair, but together with utilization of affects that psychologists are studying, including contingency anxiety and indoctrination anxiety.9 This is an approach that we will pursue, one that marshals not just considerations of epistemic luck/risk, but also the now-vast literature on studies of individual/social biases and heuristics. Before going further, however, I note that Ballantyne draws two general lessons from his study of the impact of the epistemic location problem on the epistemology of controversial domains, including especially nurtured controversial views: “[W]e should hold some of our controversial beliefs with less confidence . . . [and] we need better methods to make judgments about biases.”10 This will be a good starting point for us, since I generally agree with the spirit of both lessons, but will work them out somewhat differently. I begin by locating the historic, geographic, and demographic contingency of people’s nurtured controversial views within a much broader set of recognized sources of cognitive diversity. A person’s epistemic location is one of the most unavoidable of these sources of contrariety, being basic to the human condition. Arguably the contingencies of belief that derive from people’s geographic and demographic diversity are closely related to the evidential ambiguity that affects so many views we hold on morals, politics, philosophy, and religion. Such contingency makes for no sweeping indictment of a belief’s reasonableness: this would be to dichotomize between the rational and the social, which our understanding of the epistemic location problem most expressly should not. Instead I want to develop Ballantyne’s second lesson, the need for more principled and fine-grained application of bias studies to well-motivated etiological challenges. In doing so we will have recourse to work on the differences between benign and malign epistemic luck, to Patrick Bondy and Duncan Pritchard’s discussion of the close connection between malign luck and epistemic risk, and to Ian Kidd’s work on the differences between “robust” and merely “rhetorical” vice charging. This chapter aims to develop several sides of an inductive risk-based account of what it means to motivate serious etiological challenges and to support them empirically through markers of bias. More fully, this chapter sketches an inductive risk-based account of assessments of the doxastic states (belief/alief/credence) and epistemic standings (knowledge, understanding, subjectively/objectively justified belief, rational belief, etc.). Inductive risk is the study of the chance or possibility of getting it

278

Guy Axtell

wrong in an inductive context. The concept of inductive risk is widely used in science, and the possibility of getting it wrong is there acknowledged to often raise moral as well as epistemic concerns. If portable beyond philosophy of science, the concept of inductive risk might serve equally well as common ground for wide-ranging discussions over doxastic responsibility. Counter-inductive thinking entails the highest degrees of inductive risk.11 I will relate motivated etiological challenges to well-founded belief as highly overdetermined by trait-dependent factors. We do indeed need better methods to make judgements about biases and other temperamental factors which negatively affect well-foundedness, in terms both of the agents’ creditworthy exercise of a genuine cognitive ability, and of their personal justification.12 But we do not contend that the normative upshot of the prevalence of trait-dependent beliefs in the domains of controversial views is that they are never well-founded. As seems to be the case with Montaigne and Mill, what normative upshot philosophers should draw is more contextual than that; assessment cannot be uniform because it crucially depends upon the varying degree of blind spot bias that actual agent’s exhibit. Agents may exhibit the bias in various ways, through over-estimation of the epistemic status of their views, through rhetorical and unsupported asymmetries of explanation, and more defensively through rhetorical peer denial. It is the dogmatic ways in which nurtured beliefs are sometimes held that motivates a serious de jure challenge, rather than the amorphous line between properly doxastic and sub-doxastic attitudes. This returns us to Ballantyne’s first lesson, a lesson about guidance. Ballantyne like many others conclude that we should hold “with less confidence” our controversial views. Conformists make this a universal prescription, moving from epistemic assessment to guidance. Even Carter’s qualified version of controversial view agnosticism, which opens up sub-doxastic attitudes such as “suspecting that,” which he rightly points out that the Richard Feldman Triad model of doxastic attitudes neglects, speak in terms of downgrading the degree of confidence in a proposition being the prescriptive upshot of genuine peer disagreement. Under conditions that cover a large portion of our beliefs in controversial subject areas, Carter’s principled controversial view agnostic asserts, “we are rationally obligated to withhold judgment.”13 Carter argues effectively that “a tacit commitment to the Triad View, with its deontological categories of belief, suspension of belief, and disbelief, has the effect of artificially restricting the range of reasonable attitudes we might take up in controversial areas” (15). By opening up other sub-doxastic attitudes besides “suspension,” Carter develops resources for modifying the conformist thesis to make it more ‘liveable,’ while holding on to its guiding principle.14 Carter’s version of conformism we can term principled agnosticism about domains of controversial views. While

Well-Founded Belief

279

I appreciate Carter’s risk-focused account of controversial views, principled agnosticism is still impermissivist. Let me say something about this in order mostly to set it aside. I will return to this in my conclusion, but I hold that the same detail that the bias-studies and inductive risk approaches bring to the table to distinguish motivated from unmotivated etiological challenges, shows how overgeneralized the prescriptions that conformists and steadfasters each ask us to accept as the upshot of genuine peer disagreement are. Epistemic assessment and epistemic guidance need to be more carefully distinguished than has been the case in this debate. These are primarily questions of guidance and thus also of the ethics of belief. On these questions I have written in defense of permissivism.15 I see no easy path, either from moral or epistemic evidentialism, to the kind of universal guidance issued either by the equal-weight view, or by principled agnostics.16 If we should discern a more diverse set of doxastic attitudes than the Triad model allows us to see, I would argue that we should also discern a more diverse set of permissible responses to genuine peer disagreement. More specifically, the importance of the reliable etiology of belief for doxastic justification seems from my pragmatist or inquiry-focused epistemology to cast doubt on why doxastic responsibility and guidance-prescriptions should take a primarily synchronic form.17 So while I won’t try to provide a fuller account of guidance, I just want to state my resistance to any and all of the universalized prescriptions on offer from dogmatists, phenomenological foundationalists, equal-weight conformists, and principled agnostics. Instead I will agree with Ian Church and Justin Barrett (2016) that “psychological dynamics . . . suggest that belief firmness, or a belief’s resilience to revision or relinquishment, are not the only or best relevant metrics for intellectual humility.”18 This allows that there may be different levels at which to exhibit epistemic deference, not just at the level of one’s credences but also on various levels of one’s reasonings (Pittard 2014). Pragmatists like Davis, Susan Haack, and Susanna Rinard insist that we distinguish more carefully between norms for guidance-giving and those for epistemic assessment.19 At the same time, permissivists like virtue theorists hold that “the gap between the ways in which we are meant to normatively assess belief and action may not be as wide as has been thought,”20 and that responsible actions are often not ‘obliged,’ but merely ‘permissible’ actions. An area for merely permissible belief seems missing on the impermissivist view. So before moving on I just want to leave it in the mind of readers that a pragmatist and permissivist ethic of belief—in so far as it is a permissivism ‘with teeth’—may prove itself more effective than the principled agnostic/evidentialist demand for synchronic ‘downgrading,’ in challenging a dogmatic thinker’s faulted attempts to epistemically privilege their own or their ingroup’s nurtured beliefs.21 For if epistemologists, as I believe, have useful guidance to give

280

Guy Axtell

agents in regard to the limits of reasonable disagreement, that guidance should not assume an ideal or atemporal agent without pragmatic interests or constraints. It should be guidance consistent with Montaigne’s two points: 1) that “we are all of the common herd,” a thought commensurable with psychological studies of biases and heuristics, and 2) that due to our directional thinking especially in matters we care a great deal about, we are constantly guilty of confusing affectively conditioned commitments with possessing truth, warrant, proper basing, and right epistemic motives.22

2 Environmental Luck-Based Etiological Challenges: A Tess Case This section aims to elaborate the epistemic significance of the distinction between contexts of inquiry dependent only on benign evidential luck, and contexts of inquiry impacted by malign environmental epistemic luck. Pritchard’s post-2005 splitting of veritic luck into environmental and intervening types has strong implications for the epistemology of nurtured beliefs. So might Bondy and Pritchard’s recent identification of propositional luck as a malign form of epistemic luck in addition to veritic luck, but given our limited space I must pass over discussion of propositional luck.23 What these authors do that more directly concerns us is to translate questions about epistemic luck into questions about the modal riskiness of a belief-forming cognitive strategy.24 With this in mind, what I want to do is to discuss the importance for well-founded belief of the distinction between evidential (as benign) luck and environmental (as malign) luck. Can we always correctly distinguish them when presented with a case, and if so, how? Note first that intervening and environmental luck, while subtly different, are both forms of what epistemologists refer to as veritic luck. In cases of veritic luck, it’s a matter of luck if the belief one holds is true— that is, one very easily could have believed incorrectly. The intervening form is the form of luck that we find in standard Gettier cases such as the famous sheep-in-the-field case. Environmental luck by contrast is the kind that we find in barn facade cases. Environmental epistemic luck, understood as veritic luck and hence as distinct from simple evidential luck, is not compatible with knowledge, most epistemologists hold. If they are correct in this, it is because Barney’s belief, considered modally, appears to be unsafe. By contrast, Pritchard’s taxonomy of forms of epistemic luck recognized several benign kinds. I will treat only evidential luck, since it is the complicated relationship between malign environmental luck and benign evidential luck that I want to get at. Simple evidential luck does not violate the safety principle; it is the luck of being situated in a way that others might not be to have supporting evidence for a true belief.

Well-Founded Belief

281

Ernest Sosa uses the simple paradigm example of coming to hold the true belief that there is a crow in the yard, but only because one happened to glance out the window at that particular moment it flew by. By contrast, environmental luck does violate the safety principle. It is the luck that one’s belief is true, given a set of modal or other epistemic circumstances that are inhospitable to the reliability of the utilized doxastic strategy (mode of belief-uptake). What is importantly different between intervening (Gettier) and environmental luck cases is that in the former it is no matter of ability or competence or achievement that a true belief is acquired, whereas in environmental luck cases the agent’s beliefs are the product of the exercise of a cognitive ability that in more cooperative epistemic circumstances might provide more positive epistemic status to their beliefs. The concepts of luck and risk helps us analyze how agents achieve or fall short of more valuable epistemic states or standings— rationality, personal justification, knowledge, understanding, etc. But they may not apply in quite the same way across domains of controversial views.25 With that much said, I now want to argue that it is not difficult to construct testimonial environmental luck cases, cases in which our intuitions about epistemic status basically parallel those that people report about Barney cases, where the agent’s visual perception is the primary source of the target belief. The predominance of visual perception cases in epistemology is partly due to their relative simplicity, but partly also to many decades where methodological individualism was assumed. If so, testimonial cases allow the philosophy of luck/risk to better engage contemporary social epistemology. Here is such a case. The Basic Tess Case Imagine Tess, a good friend of Barney, travelling to visit relatives in Land of Lakes County. In the base case, this is Tess’s first visit, and she does not know that many others refer to this county as ‘Fake News County.’ Scattered about on corners of the town and the whole county are brightly colored metal or plastic, free publication newsstands, each advertising its wares in its small front window. Sometimes there were several such boxes at the same corner, but most often just one. Tess, who knew none of this, is met at the train station by her uncle Sal, and before they get to his ride they pass a corner outside the station with a blue metal newsstand. Tess had just asked her uncle a question about the history of the county, and Sal goes to the box and gets them each a copy. “Blue-box publications. Yes, this one you can trust!” says Sal, and to emphasize his point he flips the paper over and taps the large printed warning on its back page: “Remember, trust only the news from this box! All of the other boxes contain fake news.”

282

Guy Axtell Tess finds this a bit quizzical, in part because she has not encountered other boxes; but they have much to talk about, and the conversation quickly takes another direction. But that night when she retires to the guest bedroom, she finds the paper on her dresser, and reads it in bed. It contains many tales about the county and its founding citizens that Tess finds quite moving and even profound. Although there were seemingly fantastical elements to these stories, and some of them drew strong moral lessons that clearly went beyond factual information, Tess remembers her uncle’s assurance of the paper’s trustworthiness, and she accepts the content of the paper pretty much at face value. While being driven back to the station after her pleasant weekend visit, Tess notices for the first time a different colored newsstand, then another, then another. Indeed walking into the station she comes upon a veritable array of such boxes in a row. Having been so enamored of the first, Tess starts walking up to a yellow plastic one to get an issue of it for some reading on the way home. Picking it up, she is surprised to see the same strong warning against trusting other papers that Sal had called attention to on her blue-box paper. But immediately Sal stops her, saying, “All these other boxes are from different publishers, and they give only fake news. Return them. They are worthless—only trust the papers in a blue-box. They tell you all you need to know.” It saddens Tess a bit that she won’t get more such stories, but out of respect for her uncle she puts it back and refrains from gathering more papers. On the train, though, she pulls out her blue box paper and reads it again. It is growing on her, and when she tells her sister about her trip, what she relates as factual about the history of Land of Lakes County and its founding citizens is what she says any blueboxer like herself holds as true.

Now we can imagine multiple variations on this base Tess Case. Perhaps Tess learns that had she listened to Sal’s neighbor, she would have been introduced only to a red-box paper, and been told that that was the reliable one. Perhaps all the people in Sal’s family trust the blue-box paper, but most people in the county trust the yellow, or vice versa. Perhaps Tess knows that she is in Fake News County (Enlightened Tess) or perhaps she does not (as in the base case). In each such case, although it is testimonial transmission rather than visual perception that is the source of belief in Tess cases, it must be acknowledged as an environmental veritic luck-impacted context of inquiry if Tess was veritically lucky (that is, lucky that she came to acquire a true rather than false belief) given the doxastic method she employed in her specific epistemic environment. I hold that these conditions are fulfilled in the Tess case, and that her belief fails to be knowledge even if there was one wholly true newsstand

Well-Founded Belief

283

and it was the one she vested authority in. True, this argument requires modal closeness, and relevant similarity of basis, but these conditions seem to be fulfilled and I do not see other ways to pry the Barney and Tess cases apart. This does not imply that multiple pieces of independent evidence might not mitigate the risks that in Tess’ case constitute an environment of malign veritic luck. Not all testimonial transfer is unsafe, and not all testimonially based beliefs are insensitive in the way that Tess’ are. I will discuss the compounding of benign “evidential” by malign “environmental” luck later; but my claim is about Tess, as described, and not about all persons who have invested authority in a testimonial source under conditions of contestation. So neither do I think that recognizing the impact of malign luck on Tess’s beliefs about the history of the county must inevitably lead us some much broader skepticism about testimonial knowledge generally. The way that safety and sensitivity are here construed does not invite, but will indeed I think provide grounds for rejecting the broad “parity” response popular in religious apologetics: the response that to be skeptical about the epistemic status of Tess’ testimonial beliefs will result in excessive skepticism about a much wider range of ordinary testimony cases.26 Another reason why Tess’ and other agents’ beliefs in the narratives of one or another Fake News County newsstand must be seen as the product of a highly risky doxastic strategy is the relationship between the testimonies that the papers provide about the history of the county: contrariety of content itself. Part of the intuition that there are propositional defeaters to Tess’ personal justification for her testimonial beliefs is that the base case describes significant contrariety of content. Further, it describes what we will term symmetrical contrariety, in that each publisher claims all other publisher’s publications are untrustworthy. Now had there not been such actual or reported contrariety to their contents, would Tess’ beliefs, if we assume them true, be less impacted by malign environmental luck? If we answer to this question ‘Yes,’ as I want to argue that we should, then why do epistemologists of testimony seem so often to ignore such factors, and think only in terms of the reliability of the single testimonial chain an agent “trusts”?27 The inductive risk account shows as epistemically significant not just the diversity of beliefs in a domain, but contrariety of those beliefs. Not all testimonies or testifiers are as polemical as the described news publishers, demanding counter-inductive inference to the unique or complete truth of just one. These are things that compound Tess’ situation with malign luck. As an agent acquires more information about the contents of the papers in the different colored boxes, the degree of contrariety and mutual vice-charging further impacts the well-foundedness of a belief acquired on the basis of acquaintance with just one of the numerous publishers. Epistemologists refer to an epistemic environment as “hostile” if it is one that is unsafe for the doxastic method employed, and this is a key

284

Guy Axtell

characteristic that distinguishes mere evidential luck from a condition where it is compounded by environmental luck. Barney’s method of coming to believe ‘That is a red barn’ is unsafe, because easily could he have gotten it wrong in his driving environment, trusting only to his eyesight from the roadway. If Barney looks out the window a minute earlier or later, he acquires a belief with relevantly similar content, yet false. Barney’s belief is also insensitive since what Barney affirms as a barn he would have affirmed, even if he was not lucky enough to have come across one of the few real barns. Analogously, I argue that Tess’ method of coming to believe that the blue box described true history of Land of Lakes County is unsafe, because easily could she have gotten it wrong in her news reporting environment, trusting only to the testimony of a friend or family member, and to the vivid phenomenal seemings she has in reading blue box stories. Tess’ belief is also insensitive because we have to surmise that if the publisher was not reliable, Tess would have still believed that it was. If she would trust the news of just the first box she came to, when it might be a small and/or unrepresentative sample, or because it is uniquely recommended by one among many disagreeing residents, or by someone she considers reliable because a kinsman, then Tess would believe the same thing even were it false. Insensitive beliefs are not typically a byproduct of a hostile epistemic environment, as Pritchard describes unsafe belief, but of what I propose to term a beguiling environment. I will develop connections between a beguiling evidential situation and epistemic responsibility in the final section. But as a general point, when trait-dependent factors are salient in one’s belief, it is commonplace to find the agent reasoning backwards from the assumption that their belief is true, to the benign nature of any luck they may have had in coming to that belief. It is commonplace to find them confusing (presumed) truth with objective justification, which in turn beguiles them into thinking that their (obviously good) luck of having grounds for a true belief also guarantees the benign nature of luck’s impact on their epistemic situation. But the assumption is naïve: the malign/benign epistemic luck distinction is instead a modal one, and one that is inextricable, for agents performing inquiry, from recognition of the centrality of inductive norms to epistemic assessment. But perhaps the most interesting conclusion that might be drawn from the preceding is that the positive epistemic status of beliefs based on testimonial transmissions is not guaranteed, even if it is maintained that the particular testimonial chain that sources the beliefs to be assessed is a trustworthy testimonial chain.28 It is not guaranteed any more than that Barney’s true belief that he sees a barn has positive epistemic status.29 That depends not just on the object, and his process, but upon a third factor that situates his inference in an inductive context. What in a simple, non-fake barn country scenario would certainly seem to have positive

Well-Founded Belief

285

epistemic status, is far more problematic in Barn County. Whether recognized or not, Tess like Barney is in an epistemic environment in which there are defeaters to personal justification. An inductive context implies inductive epistemic risk. High epistemic risk derives from epistemic situations inhospitable to the epistemic strategy one is employing. Modal riskiness marks epistemic luck as veritic and malign. Environmental luck is veritic luck, and the epistemic standing of Barney’s luckily true belief is doubtful because of the inductive norms Barney violated in forming his belief.30 Epistemic success arguably requires an agent’s doxastic strategy being modally safe, for modal riskiness marks epistemic luck as veritic and malign. This may be an externalist perspective, but if we think in terms of argument structures we can translate these concerns into ones of defeaters and defeat. Tess like Barney has a propositional defeater of the undercutting variety, which is a most serious matter. Propositional defeaters are conditions external to the perspective of the cognizer that prevent even a personally justified true belief from counting as knowledge. The pertinent external fact is that our two agents Tess and Barney are in such conditions that there is no level of generality that a reliable belief-forming process plausibly explains the truth of their beliefs. Our account can certainly be flexible enough to allow partial defeaters. But these bare facts of Tess and Barney’s inductive contexts of inquiry are arguably each as much a propositional defeater of the undercutting variety as that the wall in front of me is being irradiated with a red light is an undercutting defeater for my belief that the wall in front of me is painted red because it visually appears red to me. It could still be true that it is painted red and not white or some other color; but my trusting my eyes as the rational basis for that belief is undercut by this further fact of which I was unaware when I formed my belief. Summarizing, Tess like Barney does an inductive ‘fail,’ although an agent’s culpability for being ignorant of their inductive context of course depends upon details of the case described.31 My point is that environmental luck is a serious worry about the well-foundedness of belief whether one is aware or ignorant of their inductive context. That environmental luck threatens to impact the well-foundedness of an agent’s belief, and that the agent’s context of inquiry is properly describable as an inductive context, are nearly synonymous. We should say that environmental luck, when it affects an agent’s epistemic situation, compounds evidential luck. It is not as if evidential luck went away and a malign kind just ‘replaced’ it. An epistemic context can change by degrees, much as the assessment of the strength of an inductive argument can change by degrees. Evidential luck, as the only way we are in a position to know anything beyond the analytic and a priori, is ever-present to the human condition, but its benign status is upset when malign conditions change its demeanor. This compounding thesis

286

Guy Axtell

challenges epistemologists who treat the benign/malign distinction more as separate buckets than as dialectically negotiated borders.32 This negotiation is quite apparent when we recognize how closely the epistemology of nurtured controversial views depends on testimony and testimonial transmission. As Rachel E. Fraser points out, “Recent epistemological history has inclined towards ‘testimonial optimism,’ keen to stress the division of epistemic labour and the ubiquity of our dependence upon the words of others.”33 Not incidentally, testimonial optimism is associated with Christian evidentialist apologetics, phenomenological conservativism, and the unmovably steadfast position its proponents justly describe as “dogmatism.” These views seem at opposite extremes from the broad skepticism about knowledge in domains of controversial views mentioned earlier (Cohen and Sher). I would like to think of our inductive risk-based account as a third option in what Fraser seems right to see as an important emerging debate between testimonial optimists and testimonial pessimists.34 But as a third option it is not completely neutral between an account that makes for ‘easy knowledge’ even of religionspecific claims (so long as one thinks their purported special revelation is more special than other purported special revelations), and a view that is skeptical of that. We should all be skeptical of that, and the specialness of the home religion’s special revelation is an article of faith, not a premise in an argument that those not already predisposed to should accept on the basis of its epistemic merits. Every religious testimonial tradition’s revelation is reliable to its adherents, just as every theology or sect is orthodox unto itself.35 The self-reassurance of religious knowledge here becomes an article of faith. But there are good philosophical criteria for when a testimonial environment is impacted by malign environmental luck. Even setting aside apologetic motivations for testimonial optimism, phenomenological conservativism, etc., in testimonial cases generally we have to look not just at the source, but with agents engaged in inquiry, at how to naturalistically describe their belief-forming cognitive strategies, and at their objectively described inductive context. Only in this way can we assess whether the kind of epistemic luck operating in particular real or imagined cases is benign, or instead malign, that is, undercutting of positive epistemic status.

3 Trait-Dependent Overdetermination, Risk, and Well-Founded Etiological Challenges: The Psychology and Epistemology of Our Importunate Presumptions Our eyes see nothing behind us. A hundred times a day we make fun in the person of our neighbour, and detest in others, defects which are more clearly present in ourselves, and we marvel at them with prodigious impudence and heedlessness. Oh, importunate presumption! —Montaigne36

Well-Founded Belief

287

Montaigne’s passage describing our ‘importunate presumptions’ captures quite well the contemporary recognition of our common bias blind spot. Our obvious psychographic diversity, and the polemical dynamics involved in our ‘culture wars’ are compounded on the agential side by the invisibility of our biases to ourselves. The judgments we make in ignorance of our own biases Montaigne calls our importunate presumptions, and he suggests a host of practical factors that make them appealing. Along with its denial in favor of exceptionalism, Montaigne points out that the cost of these ego, ethnic, and anthropocentric presumptions is that, sadly, “it comes to pass that nothing is more firmly believed than things least well-known.” This is an ironic caricature of dogmatism and bias to be sure, but Montaigne is noticing persons and groups, and that while the impact of directional thinking on nurtured controversial views is very significant, it’s bearing on well-foundedness is not all-or-nothing. Which of our beliefs can claim to be free from underdetermination/overdetermination? So I agree with Davis (2009) and other permissivists that recognized trait-dependence in the aetiology of belief does not undermine the basing relationship in any sweeping sense: it undermines that relationship only if and when that traitdependence takes the form of personal or social bias. We will return to specific, scalar markers of this shortly. But when it does not undermine the basing relationship then agent reasons competently, and the influence of personal traits need only be regarded as one of the many sources of the faultless cognitive diversity that John Rawls explained as grounds for reasonable pluralism. Davis makes the connection between trait-dependence and Rawlsian reasonable pluralism explicit by quoting the deservedly famous “burdens of judgment” section of Political Liberalism: “To some extent (how great we cannot tell) the way we assess evidence and weigh moral and political values is shaped by our total experience, our whole course of life up to now; and our total experiences must always differ” (25). Although many nurtured beliefs may be biased, we cannot assume that all are without begging the interesting philosophical questions.37 The ‘trait-basing question,’ which asks whether and when trait dependence defeats the basing relationship, requires investigation. In order to investigate it, Davis thinks we should first adequately distinguish simple trait-dependence from bias, in order to compare them. He defines ‘trait’ broadly to include “not only personal traits such as gender or features of one’s personality, but also such properties as socioeconomic background, rigorous training, exposure to certain individuals or groups, subscribing to a certain ideology or religion, or having a certain personal history”(24). Differences in people’s experiences, background beliefs, and available testimonial evidences can make robust the evidential ambiguity that affects so many views we hold on morals, politics, philosophy, and religion. The robustness of evidential ambiguity and the resultant need for holistic judgments on the part of agents, in turn impacts both epistemic

288

Guy Axtell

assessment. We accordingly describe epistemic location not straight away either as bias or as the intrusion of epistemically-irrelevant influences, but rather as a source of (sometimes but not always) faultless disagreement.38 Consistent with Davis’ divergentism (2015), I take faultless disagreement and responsibility in doxastic as well as sub-doxastic ventures as the charitable default assumption about controversial views. This kind of faultlessness does not imply relativized truth, or the idea of both parties being right.39 So censure on the basis of an agent’s doxastic irresponsibility is the exception, and has the burden of evidence upon it.40 But there are many exceptions where censure is well-founded because the agent’s beliefs are not well-founded, and these exceptions may readily occur in any of the four domains of controversial views. An etiological challenge has to be mounted domain-by-domain, and case-by-case. Those cases where wellfoundedness is especially challengeable are ones where belief-formation or maintenance flow from a risky doxastic strategy, and/or where the agent exhibits psychological marks of undue influence by subjective factors. The agents may “mirror” known biases, engage in rhetorical vicecharging, and/or exhibit certain psychological affects like contingency or indoctrination anxiety, or confabulation. Philosophically, some thought experiments that heighten these effects can be helpful for agents to gain perspective on their nurtured controversial views. What if the agent is rationally convinced that she very likely would see her own actual nurtured belief as false had she been nurtured in a different culture or epistemic community? What if the agent concedes she would likely see it as a product of unrecognized bias? Would these outcomes of the thought experiment be defeaters to proper basing? Sensitivity is often criticized as a strong demand, and I am not assuming that it is a condition of knowing. But it seems to track some relevant aspects of reasonableness, despite the fact that the belief of the victim of a malin genie that he has a physical body is insensitive also. My entitlement to hold fast to these metaphysical beliefs in a physical universe, other minds, etc. where all sources of empirical evidence support my causal story may not extend to an entitlement to hold steadfast in the case of more controversial views.41 For a culturally nurtured belief to be insensitive, we do not have to imagine radical deception scenarios.42 We only have to make some quite modally close changes, such as: growing up in a politically or religiously conservative family instead of a liberal one, growing up in our same society but in a different religious tradition; growing up in a different society that has a different majority religious tradition, etc. So it is plausible that sensitivity tracks reasonableness when the closest error-possibilities are nearby, and it doesn’t track reasonableness when the closest errorpossibilities are distant.43 Whatever we can say about the truth-aptness of beliefs in domains of controversial view, and about trusting putative

Well-Founded Belief

289

moral or religious experts, it is clear that beliefs in these domains are exceptionally insensitive.44 But their insensitivity and their causally overdetermined aetiology are almost indistinguishable. Now even if the mentioned thought experiments regarding the safe and sensitive founding of our beliefs are indeed epistemologically significant, does the bias blind spot allow agents to see the implications? Does it render them able to see when the attributions of bias or intellectual vice which they readily apply to others, apply as well or better to themselves? The approach taken here is far from defeatist, because I think there is much to be said that can redress the bias blind spot. The numerous indications of bias are a resource for epistemologists, just as they are for psychologists. True, epistemologists are always going to be censuring those who are least likely to acknowledge their importunate presumptions, or to be motivated to re-evaluate their beliefs. Indeed that is why we censure what we perceive as bias and circular reasoning: for without a mirror to hold oneself up against, it is almost impossible to see that the ‘inductive finger’ points not just outwards at holders of contrary views, but frequently back at them.45 It would be impossible to understand that people with contrary views to ours may still be made in our same image; instead they become trapped in seeing their deviance from our opinions as confirmation of their bias. So the problem of motivation to de-bias oneself is one in which philosophers can seek the aid of psychology. The horses I am familiar with do drink when led to water, and if they don’t then I suggest thereafter riding them much harder. But our project here is much concerned with what philosophers can contribute to the assessment of bias and other defeaters to well-founded belief. Here I see a lot of untapped resources. Epistemology can show the enemy in the mirror to those who need most to see it, although the act of recognition, since it requires proper motivation, has to come from within. We have started to sketch an account of well and ill-founded belief based upon low and high inductive risk. Our account says that agents mitigate epistemic risk by acknowledging an inductive context and abiding by inductive norms. It says that agents exacerbate moral and epistemic risk by asymmetrically positing themselves or their sources of belief as exemptions to a recognized pattern. In one sense this is really just the philosophical analysis of what psychologists call my-side, or belief-bias. Complementary to discussion of degrees of trait-dependence, I want to introduce overdetermination theory. The problematic sort of overdetermination stems from finding multiple trait-dependent factors each sufficient to produce the target belief. In such cases we have trouble isolating which of these processes actually caused the belief. If the belief somehow is true, we have lost the connection with creditworthiness on the part of the agent.

290

Guy Axtell

To develop the inductive risk account further, let’s very briefly take a closer look at two further inductive risk-indicators: confabulation, and merely rhetorical or self-deceived bias-charging. Confabulation is counterpoint to contingency anxiety, though they could be seen as two different ways to deal with the cognitive or moral dissonance. As Andreas Mogensen (2017a) explains, “Etiological Challenges encourage us to pay attention to notable facts about our belief-forming processes that would otherwise be ignored.” Mogensen usefully gives a name—contingency anxiety—to the anxiety that a person might have who rationally concedes to counterfactual statements indicating that they would in other circumstances have come to hold beliefs that are by their own lights wrong, or more to the point to reject as false beliefs that are by their own lights true. DiPaolo and Simpson focus on a close cousin: Indoctrination Anxiety, on our usage, is something narrower than Genealogical Anxiety, in which an individual is caused to ‘worry that the origins of her beliefs will turn out to be a source of discredit not vindication,’ and also narrower than a more general feeling of Contingency Anxiety, in which an individual is led into ‘a feeling of unease due to discovering that she holds certain beliefs because of arbitrary factors in her background.’ Indoctrination Anxiety, rather, is the distinctive sense of unease a person experiences when she’s led to suspect that her beliefs resulted from a systematic program of doctrinal inculcation.46 Where one or another form of anxiety and attendant epistemic humility is appropriate yet lacking in an agent, we can hypothesize that she will be quick to engage in confabulatory explanation. Confabulation is counter-point to contingency anxiety, though they could be seen as two different ways to deal with the cognitive or moral dissonance. William Hirstein writes, “Confabulation involves absence of doubt about something one should doubt: one’s memory, one’s ability to move one’s arm, one’s ability to see, etc. It is a sort of pathological certainty about ill-grounded thoughts and evidences.” More than simple rationalization, “Confabulators don’t know that they don’t know what they claim.”47 Hirstein gives these conditions: Jan confabulates if and only if: 1) 2) 3) 4) 5) 6)

Jan claims that p (e.g., Jan claims that her left arm is fine). Jan believes that p. Jan’s thought that p is ill-grounded. Jan does not know that her thought is ill-grounded. Jan should know that her thought is ill-grounded. Jan is confident that p.48

Well-Founded Belief

291

Sharp and apparently unprincipled explanatory asymmetries are another key marker of bias. In “You Don’t Know me, but I Know You: The Illusion of Asymmetric Insight,” Pronin, Kruger, Savitsky, and Ross discuss psychological studies confirming that people often exhibit “an asymmetry in assessing their own interpersonal and intrapersonal knowledge relative to that of their peers.”49 We tend to judge others as biased, especially when they disagree with our controversial views. Yet we are more likely to take our own views (and our own attributions of bias) to be bias-free. So perhaps ironically, one place to look for such asymmetries is in “bias-charging” behavior itself. Peer denial through ill-founded bias-charging is a very common but highly dogmatic way to insulate particular beliefs from rational criticism. Emily Pronin and Lee Ross in particular have also suggested broader application of their findings by describing the psychology of “naïve realism” that biased trait-attributions often presupposes: “although this blind spot regarding one’s own biases may serve familiar self-enhancement motives, it is also a product of the phenomenological stance of naive realism.”50 Naïve realism as psychologists discuss it is connected with what philosophers such as Lisa Bortolotti and Matthew Broome (2009) refer to as failures of belief ownership and authorship.51 So psychologists, and philosophers who utilize psychological research are both interested in “the relevance of these phenomena to naïve realism and to conflict, misunderstanding, and dispute resolution.”52 Ian Kidd (2016) relatedly points out that vice-charging can either be an encouragement for needed self-awareness and doxastic responsibility in the agent who is criticized, or it can be a strategy of protecting oneself or one’s beliefs from criticism. Kidd distinguishes rhetorical complaints and robust charges, where only the latter qualify as legitimate modes of criticism: “A rhetorical vice charge involves an agent expressing a negative attitude, opinion, or evaluation of some other agent . . . but not the presentation of any reasons, evidence, or feelings in support of them, so they do not do any real critical work”53 Robust vice-charges require a clear concept of epistemic responsibility, and so “should be sensitive to the aetiology of vice and the ecological conditions of epistemic socialisation.” Each of these psychological effects might motivate a strong etiological challenge to the well-foundedness of belief. Each suggests the salient causes of belief to be temperamental factors that, if they do not exhibit, at least must be acknowledged to mirror known personal or social biases. To summarize this section, the study of trait-dependence is vital to the epistemology of controversial views. Trait-dependence appears to be indicative of the overdetermination of belief by temperamental factors.54 The beguiling nature of evidence in environmental veritic luck cases is exacerbated by the trait-dependent overdetermination of belief. This appears to be especially so with culturally nurtured controversial views. In cases of causally overdetermined belief, an agent’s belief might have

292

Guy Axtell

been adopted on any one of several different trait-dependent bases, although on an ex ante basis the belief is underdetermined by the agent’s actual evidence, evidence which would not rationally convince persons not already disposed toward the belief. Although we rarely treat them this way, evidential underdetermination and causal overdetermination are paired theses. The first is a philosophical concept and the second is a scientific one, but the two are conceptually linked. Overdetermination theory is still a largely unexplored approach in debates over the basing relationship. But it is motivated by the holistic nature of people’s reasoning about worldview beliefs, under conditions of uncertainty and other pragmatic constraints, as Rawls alerted us to. It is motivated also, we have now seen, by some specific psychological studies, research that illuminates how trait-dependent judgment contributes to psychographic contrariety.

4 Conclusion This chapter has argued for a fairly common-sense view: That our nurtured beliefs being exposed to the epistemic location problem need not undercut their reasonableness or our right to hold them. The bearing of epistemic location on nurtured controversial views is very significant, but need not fall evenly across domains. Nor does it fall on all agents the same, as we need to know more specifics about the sensitivity of an agent’s reasons for her belief in order to assess how seriously to take an etiological challenge. People are not necessarily intellectually vicious for accepting nurtured beliefs and holding them without a great deal of reflection. But neither does such a permissivist account rationalize dogmatism or imply the reasonability, tout court, of holding to what we are taught. Permissionism should sharpen reasoned criticism rather than lead to its abandonment, and the turn to risk and inductive risk in particular, I have been arguing, shows us how. There is faultless or reasonable disagreement aplenty on the present view, but it occurs primarily in minimally truth-apt discourses where the parties to the disagreement recognize their discourse as minimally truth-apt. The disputants then take commitments in the domain of their disagreement as requiring a greater degree of epistemic and moral humility than disagreements over straightforwardly empirical claims or questions. Faultless disagreement does not occur under conditions of self-deception or of bias mirroring. Ways of acquiring or maintaining a belief dependent upon apparent violations of inductive norms are far from faultless, and the beliefs of agents who rely on such methods are especially exposed to serious etiological challenge. To conclude, nurtured beliefs cannot all be evaluated in a uniform way as conformists and dogmatists have assumed.55 Our commitments in domains of controversial views, do not deserve to be assessed as illfounded simply because they are conditioned by temperament and epistemic

Well-Founded Belief

293

location. But our approach suggests that neither do they deserve the free pass (as personally justified and as enjoying positive epistemic status) that epistemic conservatives and epistemic dogmatists issue them on the basis of an agent’s phenomenal seemings.56

Notes 1. Consistent with work on epistemic injustice, I hold that attitudes and beliefs about others can wrong others. But this claim is not uncontroversial. For recent work on this question of doxastic responsibility and its limits, see the journal special edition edited by Rima Basu and Mark Schroeder (2018a), and their paper “Epistemic Wronging” (2018b) which (like Axtell 2013) appears to defend Susan Haack’s (1997) moral-epistemic “overlap” account. 2. According to J. Adam Carter (2018), recognition of peer disagreement implies that “we are rationally obligated to withhold judgment about a large portion of our beliefs in controversial subject areas, such as philosophy, religion, morality and politics.” He recognizes that a thorough-going agnostic suspension of the kind recommended by some epistemologists is open to objectionable consequences of ‘spinelessness,’ and impracticability—the un-livability objection. So he distances his version of controversial view agnosticism from these worries, qualifying it such that it allows for ‘suspecting that’ but not ‘believing that.’ 3. Trait-dependent belief is widely acknowledged in domains where informal logic, interpretation, weighing factors or norms, and the exercise of judgment are normal aspects of epistemic assessment. In the acquisition of trait-dependent beliefs, “the subject reasons competently from justifying considerations to a belief, but would not do so if he or she lacked some trait that appears to be epistemically irrelevant, even if the subject and the situation were the same in all other respects.” The agent “would not take those considerations to justify that belief if she had a different socioeconomic back-ground, religious affiliation, temperament, political ideology, or the like, even if she were otherwise in the same epistemic circumstances” (23). 4. See Dan M. Kahan (2013) for a discussion of empirical work on motivated reasoning, and (2017) on how identity-protective cognition can generate inaccurate perceptions. 5. Montaigne, Apology for Raimond Sebond, 6. 6. Sher (2001). See also Mogensen (2017a), Ballantyne (2013), Vavova (2018), and DiPaolo and Simpson (2016). Vavova (2018) also argues that evidence of irrelevant belief influence is sometimes, but not always, undermining. It is surprising how little this quite plausible thesis has been systematically explored, but epistemologists of disagreement have been enamored of universalist answers. 7. The truth-aptness of philosophical judgments have been challenged in similar ways. See Amia Srinivasan (2015) for a treatment of “genealogical skepticism” about philosophical judgments. Srinivasan’s account overlaps mine where she provides discussion of different arguments for genealogical skepticism and in her response to it; more especially, in her noting that, “Epistemologists differ over the extent to which luck plays a role in the acquisition of knowledge. All epistemologists will agree that luck has some role to play. . . . Where epistemologists disagree is on just how much knowledge we can acquire through good luck” (347). The Tess Case (later) will argue that this is a matter of the kinds of luck in play, and not merely the domain. I thank

294

8.

9. 10. 11.

12.

13.

14.

15.

Guy Axtell Adam Carter for bringing similarities with Srinivasan’s learned paper to my attention. It is hard to say, for instance, just what the conclusion of Mill’s argument about the “London Churchman” in On Liberty is. Bogardus (2013) very plausibly suggests that the alleged wrong—and the basic reason for contingency anxiety—is a violation of safety. “For Mill, there are nearby possibilities in which one forms religious beliefs via the same method she actually used, and yet in which she would believe something which is, by her own lights, false.” But it also seems to involve failure of sensitivity. See also Ballantyne (2013) and Baker-Hytch (2014). DiPaolo and Simpson (2016). They bid philosophers to more carefully investigate the question, “How does recognition of the contingent cultural etiology of one’s beliefs affect their epistemic standing?” See also Simpson (2017). Ballantyne (2015, 159). Counter-induction is defined in dictionaries as a strategy that whether selfconsciously or not reverses the normal logic of induction. In its most formal sense, counter-inductive thinking is something much more specific than weak inductive reasoning. It is not just weak analogy, weak causal inference, or faulty generalization, to refer to the three forms of inductive reasoning, analogical, causal, and generalization. Rather, counter-inductive inference is the logically illicit move of reasoning oppositely to what induction suggests. For our purposes it is more simply a logical failing to apply to one’s self (or to one’s own epistemic situation) an explanation that one recognizes as applying to others (and others’ epistemic situations). I prefer to interpret well-foundedness broadly enough to indicate plural kinds of epistemic assessment. Where it might indicate merely subjective justification, or a sense of rationality or personal justification related deriving from ‘synchronic evidential fit,’ objective factors are obscured from view and I can no longer find it a very interesting concept. Well-foundedness cannot mean such a subjective status as personal or subjective justification, cut off from causal etiology and objective justification, without losing its philosophical interest. Thanks to Ru Ye for pointing out the need for me to say more about well-foundedness. See Axtell (2011) for a fuller critique of the evidentialist way of construing grounding, which fixates on ‘synchronic evidential fit’ as answering questions both about agent rationality and about epistemic assessment. Actually that concept can answer neither of these concerns. I portray it as an answer in search of a question. It may be that the traditional project of analysis shows that people “know” far less than they imagine. I suspect this is true, but it is a question in the project of analysis, and I do not see the connections which others would make that one therefore should not believe. The language of being rationally obligated to withhold judgment but to adopt only a lesser doxastic attitude than belief, is a language that I suspect is objectionably voluntaristic. To accommodate the unlivability objection to principled agnosticism, Carter expands the connotation of “agnosticism” to include the sub-doxastic attitude of ‘suspecting that,’ when conditions are right. But in this prescription, much like Feldman, there is still assumed a single right response to revealed peer disagreement among controversial views: agnosticism. Like Feldman it appears that Carter’s categories of doxastic attitudes are still essentially treated deontologically, since they line up with epistemic duties or entitlements. These are things denied by permissivists like myself. See especially the work of Thomas Kelly, and Matthew Kopec and Michael Titelbaum. In defense of permissivism, see especially Kelly (2013), Booth and Peels (2014), and Kopec and Titelbaum (2016). Virtue theory in contrast to

Well-Founded Belief

16.

17.

18.

19.

20. 21.

22.

295

internalist evidentialism I take to be champion of diachronic norms, viz., the axiological-etiological or forward/backwards spectrum. For my own virtuetheoretic account of permissivism, which I term doxastic responsibilism, see Axtell (2013, 2018). Principled agnosticism has seemed to its critics to prohibit actions where an act is forced. I do not want to stray very far into questions of guidance-giving, since I take the norms that should inform it to not be highly commensurate with norms that inform epistemic assessment, or the project of analysis of propositional knowledge. I do not see these projects as very commensurate and wonder what concept of rationality can bear the burden of guidance that one should always ‘split the difference’ with our disagreeing peers, or again that one should ‘not traffic’ in belief at all in the domains of controversial views. I have elsewhere (2011) argued that the norms that inform an ethic of belief are typically more diachronic than synchronic, and that guidance-giving takes place in the context of ecological rationality, not ideal agency where the order of acquired evidence should make no rational difference as all. Note that the objections I present to Feldman and Conee’s explicitly epistemic evidentialism are meant to be complemented by my direct response (Axtell, 2018) to the over-weaning moral evidentialism of Scott Aiken and Rob Talisse (2018). Both parties I think mis-apply the rational uniqueness thesis to the epistemology of controversial views. It is admittedly difficult to say what appropriate deference in one’s reasoning is, and this difficulty is made worse by belief-focused epistemology where synchronic measures of evidential fit are privileged due to the ex ante approach taken to personal justification. The highly influential Protestant conceptions of religious faith that are prescriptively anti-evidentialist yet also identify faith with assent to belief I take as sufficient to show why this ‘degree lowering’ kind of guidance is of little value in the debate over the rationality of religious belief. Church and Barrett’s (2016) alternative is welcome. They are unhappy with extent accounts of intellectual humility, and so propose a doxastic account on which intellectual humility “is the virtue of accurately tracking what one could non-culpably take to be the positive epistemic status of one’s own beliefs.” I do not want to say too much about the nature of our control over belief, but I agree with Rinard that practical considerations can serve as motivating reasons for belief. See her critique of Feldman and the rational uniqueness thesis. Contrasting that evidential principle with the more pragmatism-friendly Equal Treatment, Rinard correctly argues that “Insofar as we have control over these beliefs (be it direct or indirect), Equal Treatment acknowledges the moral dimension as highly relevant to the question of what we should believe” (“Equal Treatment for Belief,” 2018c). See also Rinard (2018a, 2018b) and Miriam Schleifer McCormick's chapter (this volume). Booth and Peels (2014). As children of time, we deserve respect for background beliefs and for many other effects of culture. Guidance must be consistent with psychological acknowledgment of pragmatism about reasons and of the ecological rationality of human agents. I would not presume to say that belief may never be permissibly responsive to non-epistemic reasons. We must not forget that we rightly reason holistically, and that as creatures of time as well as of place, we inevitably ‘live forward.’ Looking backwards, as Montaigne correctly says, is much more difficult for us, and this is where philosophy and the sciences help the most. These points are arguably recognized in Rawls’ guidance of expecting much faultless disagreement as an implication of conditions of free inquiry in a

296

23.

24. 25.

26.

27.

28.

29.

Guy Axtell democracy. The burdens of judgment are coordinate with sources of cognitive diversity over which we have little control. But they appear to be denied by those whose self-ascriptions of knowledge distain the burdens of judgment, if not also by impermissivists as a consequence of allowing no place for doxastic permissions/invitations not reducible to epistemic duties to believe, suspend belief, or disbelieve. Propositional luck is not a form of veritic luck because it is not a type that comes ‘betwixt the agent and the world,’ as is the case in Gettier-type (intervening veritic luck) and barn-façade-type (environmental veritic luck) cases. In Bondy and Pritchard’s explication of propositional epistemic luck (PEL), they write, “S’s belief B is propositionally epistemically lucky iff S has a good reason R (and therefore, propositional justification) for B, but it is only a matter of luck that she does. . . . All cases of propositional epistemic luck are cases where a subject has a belief which is propositionally but not doxastically justified (though, as we will shortly see, not all cases of beliefs which enjoy propositional but not doxastic justification will involve propositional epistemic luck). There are two ways in which a belief that is propositionally justified can fail to be doxastically justified: it can be held on the basis of a bad reason, or it can be held on the basis of a good reason but in a bad way.” For other important recent work on the turn to epistemic risk, see Carter (2017) Freedman (2015), and Riggs (2008). This differs somewhat from Carter’s treatment; Carter sets up helpful conditions of “centrality” and “symmetry” of disagreement in a domain as a way to distinguish less and more well-foundedness. But he then seems to generalize to these conditions being met in all four domains of controversial views, without recourse to more specifics that might distinguish them. It is unclear to me why a person’s knowing that they are not in the market for knowledge, is thereby rationally from being in the market for belief. Belief that p entails believing that p is true in some sense, but does not without absurdity entail believing that one has everything else required of knowledge over and above true belief. Baker-Hytch (2018, 189) uses a parity approach to argue that contingency arguments like Mill’s will “result in excessive skepticism concerning a range of ordinary testimony cases.” But Mill already mentioned political ideologies in his argument, and we have already conceded that the epistemic location problems affect domains of controversial views. For a deep and provocative treatment of broad parity responses to etiological challenges to knowledge in a testimonial religious tradition, see Quinn (1991, 2007). Contrariety is something largely lacking is perceptual cases, and this is part of why Barney cases do not aid philosophers in analyzing environmental luck in ways that adequately distinguish it from benign evidential luck. Why is ‘actual trust plus posited truth’ supposed to be a defeater for the importance of testimonial diversity, and not ‘testimony diversity with symmetrical contrariety’ a defeater for rational trust and the right to claim truth? I agree also with John Bishop (and others including Jennifer Lackey 2018) who argue “against the proposal that faith’s similarities to interpersonal trust merit its being considered reasonable or virtuous.” Bishop argues that “there is an important analogy between faith and trust that is crucial to understanding the content of faith. But the disanalogies between the two sever the attempt to justify faith along the same lines as trust” (Bishop 2014). For recent work on trust, see Faulkner and Simpson (2017). As Rachel E. Fraser points out, “Recent epistemological history has inclined towards ‘testimonial optimism,’ keen to stress the division of epistemic labour and the ubiquity of our dependence upon the words of others” (2018,

Well-Founded Belief

30.

31.

32.

33.

297

204). Several authors have explored testimonial pessimism, which takes a more dour view of testimonial transmission in domains of controversial views. See also Howell (2014) and Mogensen (2017b). As both Fraser and Mogensen argue, our dealings with testimonial reception bring in tow an ideal of authenticity that places special demands upon us. I understand the key demand as primarily one to symmetrically apply inductive norms, rather than to think counter-inductively. Charlie Pelling’s (2013) “Assertion and Safety” offers an account that connects a safety condition on knowing with “a safety account of assertion, according to which one asserts p properly only if one asserts p safely. The central idea is that an assertion’s propriety depends on whether one could easily have asserted falsely in a similar case.” This kind of translation between 3rd and 2nd personal (roughly, externalist and internalist) perspectives on epistemic assessment is always welcomed. On Pritchard’s ‘turn to risk,’ see his 2017 and recent papers on epistemic dependence and ALVE papers. As a proponent of ALVE, Pritchard does not make sensitivity a generally necessary condition on knowing. Indeed a strong sensitivity condition for everyday beliefs may invite radical skepticism. But I will be interested to try to explain why it does mark out belief formation that, viewed in terms of high epistemic risk, sometimes motivates a serious etiological challenge. It also motivates an inference that contradicts the agent’s own account of reasons for belief: the inference that the belief is insensitive because biased. The agent’s bias blind spot prevented them from seeing that they have no good grounds for denying that the inductive finger points back at them, and that their own belief was about as strongly overdetermined by cultural and temperamental factors as was the contrary belief of their peers. Questions of agent culpability I take to go with a theory of doxastic responsibility or epistemic rationality. That theory may well connect with aims of guidance and perhaps censure, not with the project of analysis. These I see as separate projects. Indeed I think there are three not two sorts of normativity epistemologists are interested in: “personal” justification (synchronic and diachronic), “epistemic” justification (epistemic ability and any further anti-luck conditions), and “guidance” which may censure, but is purely ideal unless it respects human ecological rationality and pragmatic constraints on doxastic attitudes. So evidentialists like Feldman are mistaken in the first place to treat synchronic rationality as basic to analysis of knowledge, and mistaken again when taking epistemic evidentialism as grounds for his evidentialist ethics of belief. See Booth (2011) in support of the separate projects idea, the ‘divorce’ between the theory of rationality and the analysis of knowledge earlier proposed by Richard Foley. What I present is certainly more contextual an account than one where religious apologists and philosophical theologians treat the benign/malign and the evidential/environmental as lining up, but just determinable by whether one thinks that good religious epistemic luck has given them a uniquely reliable testimonial source. This is circular reasoning that privileges the ‘home’ religion without acknowledging the extent that adherents of contrary testimonial faith traditions do just the same. Arguably its social consequences, far from respecting the Rawlsian burdens of judgments, promotes rhetorical vice-charging. I argue in Problems of Religious Luck (Axtell 2019) that it appears to recommend embracing us/them group polemics rather than facing down our bias blind spot or allowing any possibility, consistent with faith, that as far as religious epistemology goes, the ‘inductive finger’ points back at them. Fraser (2018, 204). Charlie Pelling (2014, 2013) relatedly argues that there are significant differences between the norms of “assertion” and of “telling,”

298

34.

35.

36. 37.

38.

39.

40.

41.

42.

Guy Axtell differences that are often overlooked among testimonial optimists. They need to be treated separately because the norms of telling are not reducible to those of assertion. Several authors besides Fraser have explored testimonial pessimism, which takes a more dour view of testimonial transmission. See also Howell (2014) and Mogensen (2017b). My own Problems of Religious Luck (2019) also engages this dispute. Post-liberals maintain that all discussion should proceed “intra-textually.” Plantinga holds that if Christianity is true then there are no propositional defeaters to warranted Christian belief. But these arguments are both examples of the epistemically circular attempt to justify one’ framework with reference to that very framework. Montaigne, “Of the Art of Discussion,” in Frame (ed.), 709. Davis does not think it is correct to say that traits consist of holding particular beliefs, though the trait may involve holding some beliefs (24). A bias in cognition “is a tendency towards a certain kind of distortion in one’s process of reasoning and belief formation.” But in trait-dependence cases as Davis wants to understand them, ex hypothesi “there are no such distortions—just the dependence relation. . . . Trait-dependence is not a cognitive distortion unless trait-dependence defeats the basing relationship—and that is the question before us; we cannot assume an answer to it” (24). On the one hand, Davis reasons, it is implausible to present all controversial views as merely biased. Agents can be quite sincere and competent in offering justificatory reasons for their nurtured beliefs, and we often cannot detect overt bias or cognitive distortions in the reasons they offer. Davis (2009) thinks that judgment can be based on the considerations the agent claims as her reasons or justification even when those reasons depend on personal traits. On the other hand, Davis points out, there remains a deeply worrisome relation between a person’s actually held belief and the counter-factual consideration that the agent would not take those considerations to justify that belief if his or her ‘epistemic location’ were different than it actually is. I agree with Davis that it is implausible that either on a moral evidentialist or epistemic evidentialist basis, guidance on doxastic responsibility given to agents should demand strict suspension of nurtured beliefs. The treatment both of epistemic assessment and of guidance-giving needs to be more contextual than this, and as I will add, these two forms of normativity need to be more carefully distinguished. Thanks to Adam Carter for bring Max Kolbel’s work (2004) and the relativist/ contextualist view to my attention. It may be that all domains of controversial views are only minimally truth apt, but even if this is so, it does not imply relativism. I am using ‘faultless’ in a sense of the bounds of reasonableness. I could concede that there must be some error when anyone comes to a false belief, since we should not reduce error to culpable fault. See especially Bishop 2007a and b on how moderate fideism can constrain doxastic ventures. Some degree of failure to apply inductive norms is common place in our thinking. See also Erik Baldwin and Michael Thune (2008) on the relationship between experiential and testimonial transmission account of warranted religious belief. Kevin Wallbridge (2018) argues that while inductive knowledge may not be strongly sensitive, it is weakly sensitive. The point is supportive of the importance of the insensitivity of counter-inductive thinking to epistemic rationality or reasonableness, even though I do not take sensitivity as a general necessary condition on knowing. Insensitive belief that is not based upon shared facts but rather on private intuitions or other subjective factors cannot claim the same reasonableness

Well-Founded Belief

43. 44. 45.

46.

47.

48.

49. 50. 51.

52.

299

as the belief that I am not radically deceived. My entitlement for the latter does not imply entitlement to the former, potentially much more temperamental choice. I thank Patrick Bondy for suggesting this formulation of the difference. James Fritz (2018) argues that there no easy or compelling root from pessimism about moral deference to steadfastness about moral disagreement. Some will reply that abiding by inductive norms is the council of caution, but that a council of courage is just as or more legitimate in some domains like that of religion. But our account already acknowledges pragmatic reasons for beliefs in CV domains; it targets only self-deception and overt bias mirroring, while allowing some influence of personal temperament, as quite reasonable and permissible. To the principled agnostic it says, ‘Seek your ataraxia your own way’; to the political ideologue, the moralist, and the religious enthusiast it says, ‘Seek your personal perfection, but don’t let your faith venture risk me or others unjustly.’ Interestingly, Carter utilizes the dual-aims point, but neglects responding to Kelly (2013) who had earlier used it as a central reason to support permissivism over impermissivism. DiPaolo and Simpson continue, “Indoctrination consists in the use of educational practices that serve to impart something like absolute and inflexible acceptance. To indoctrinate is to ‘teach someone to fully accept the ideas, opinions, and beliefs of a particular group and to not consider other ideas, opinions, and beliefs.’” Confabulation is arguably of special philosophic concern when it manifests in connection with the holding of controversial views for which there are strong aetiological challenges. If we fill in that that “believes that p” in the above is a belief about a trait asymmetry between Jan and another person or persons, then we can see that asymmetry and confabulation are often found combined. Rationalizing an asymmetric ascription or explanation on weak rational grounds invites overt confabulation on the part of the agent, and perhaps the more so as it incites psychological contingency anxiety or another form of cognitive dissonance. William Hirstein (2005, 209 and 187). Compare Lisa Bortolotti: “When people confabulate they ignore some of the psychological processes responsible for the formation of their attitudes or the making of their choices, and produce an ill-grounded causal claim when asked for an explanation” (2018, 235). Pronin, Kruger, Savitsky, and Ross (2001, 639). See also Pronin, Lin, and Ross (2002) andPronin, Gilovich, and Ross (2004). Pronin, Gilovich, and Ross (2004, 781). The strong connections between failure of belief ownership and appeals to luck should be obvious. Bortolotti and Broom argue that “by appealing to a failure of ownership and authorship we can describe more accurately the phenomenology of thought insertion, and distinguish it from that of nondelusional beliefs that have not been deliberated about, and of other delusions of passivity.” Breyer and Greco (2008), in their account of the epistemological importance of cognitive integration and the ownership of belief, hold that a belief is well-integrated in the way that brings abillity and epistemic credit to the agent, not only if the subject owns the belief, but also only if the (real or putative) process or ability is not subject to any defeaters to which the agent has access. But contrary to these authors, I argue that counter-inductive methods of belief-formation have defeaters of which the agent has access. Disowning grounds for the truth or justification of belief of one’s belief, against inductive pattern, is itself a most serious violation of the second condition of the absence of accessible defeaters. Pronin, Lin, and Ross (2002, 369).

300

Guy Axtell

53. Kidd (2016, 183). 54. If we were to determine that this is the case with all extent beliefs in the domain, then even though evidence always plays some significant role, it might suggest to neutral observers that we are dealing with a domain of discourse that, unlike straightforward empirical domains, is only minimally truth-apt. So trait-dependent overdetermination and the grounds for antirealism about discourse in a domain, are also interestingly linked. Axtell (2019) also looks at this in terms of what conditions might lead one to legitimately move from a de jure to a stronger de facto objection. For example, when might one move from a Humean criticism of the rationality of testimonially grounded belief in one purported miracle event, to a stronger metaphysical or de facto (‘debunking’) claim that the best explanation of the epistemic shortcomings of many/all miracle beliefs is that there are no miracles. But there as here I leave the legitimacy of these shifts as open questions, since my point is that there is no direct inference to be made, but only an argument based on inference to the best explanation. So my aim is merely to articulate conditions under which they gain in their plausibility. 55. See Haidt (2012). Davis distinguishes broad and narrow traits, and argues that one’s belief “is properly based when such traits operate just like broader traits whose basing relationship is not disputed” (2009, 36). Broad traits help one form many uncontroversially true beliefs, and when they do so reliably “are good candidates for faculties, capacities, or epistemic virtues” (30). Getting credit for true belief through well-integrated cognitive abilities is indeed earned credit. It is not credit on the cheap as we find in cases so insensitive and unsafe that “warrant” (positive epistemic status) for the agent would disappear if the belief were not presumed true. Plantinga for example concedes that what he terms “warranted” Christian belief would not be warranted, by his own account of the term, were the content of the claim not true. So warrant simply follows the purport of truth in metaphysical claims, and that truth itself is a felix culpa. In my work (2019) I develop Plantinga’s concession as indicating a self-described externalist apologetic that ‘leans on luck,’ and fails to accept the full implications of externalism for epistemic assessment. 56. Examples of philosophers taking these positions are McCain (2008) for phenomenological conservativism, and Fantl (2018) and Tucker (2011) for dogmatism.

Bibliography Aiken, Scott and Robert B. Talisse. “The Will-to-Believe Is Immoral.” In William James, Moral Philosophy, and the Ethical Life: The Cries of the Wounded, edited by Jacob Goodson, 143–160. Lanham, MD: Lexington Books, 2018. Axtell, Guy. Problems of Religious Luck: Assessing the Limits of Reasonable Religious Disagreement. Lanham, MD: Lexington Books, 2019. ———. “William James on Pragmatism and Religion.” In William James, Moral Philosophy, and the Ethical Life: The Cries of the Wounded, edited by Jacob Goodson, 317–336. Lanham, MD: Lexington Books, 2018. ———. “Possibility and Permission? Intellectual Character, Inquiry, and the Ethics of Belief.” In William James on Religion, edited by S. Pihlstrom and H. Rydenfelt, 165–198. London, UK: Palgrave MacMillan UK, 2013. ———. “From Internalist Evidentialism to Virtue Responsibilism: Reasonable Disagreement and the Ethics of Belief.” In Evidentialism and its Discontents, edited by Trent Dougherty, 71–87. Oxford, UK: Oxford University Press, 2011.

Well-Founded Belief

301

Baker-Hytch, Max. “Testimony Amidst Diversity.” In Knowledge, Belief, and God: New Insights in Religious Epistemology, edited by Matthew A. Benton, John Hawthorne, and Dani Rabinowitz. Oxford, UK: Oxford University Press, 2018. doi:10.1093/oso/9780198798705.003.0010. ———. “Religious Diversity and Epistemic Luck.” International Journal for Philosophy of Religion 76, no. 2 (2014): 171–191. Baldwin, Erik and Michael Thune. “The Epistemological Limits of ExperienceBased Exclusive Religious Belief.” Religious Studies 44, no. 4 (2008): 445–455. doi: 10.1017/S0034412508009530. Ballantyne, Nathan. “De-Biasing Biased Thinkers (Including Ourselves).” Journal of the American Philosophical Association 1, no. 1 (2015): 141–162. ———. “The Problem of Historical Variability.” In Disagreement and Skepticism, edited by D. Machuca, 239–259. New York, NY: Routledge Press, 2013. Basu, Rima and Schroeder, Mark (eds.). “Can Beliefs Wrong?” special edition, Philosophical Topics 46, no. 1 (2018a). ———. “Epistemic Wronging.” In Pragmatic Encroachment in Epistemology, edited by Brian Kim and Matthew McGrath. London: Routledge, 2018b. Bishop, John. “Trusting Others, Trusting in God, Trusting the World.” In Religious Faith and Intellectual Virtue, edited by Lauren F. Callahan and Timothy O’Connor, 159–173. New York, NY: Oxford University Press, 2014. ———. “How a Modest Fideism May Constrain Theistic Commitments: Exploring an Alternative to Classical Theism.” Philosophia 35, no. 3–4 (2007a): 387–402. ———. Believing by Faith: An Essay in the Epistemology and Ethics of Religious Belief. New York, NY: Clarendon Press, 2007b. Bogardus, Tomas. “The Problem of Contingency for Religious Belief.” Faith and Philosophy 30, no. 4 (2013): 371–392. Bondy, Patrick and Duncan Pritchard. “Propositional Epistemic Luck, Epistemic Risk, and Epistemic Justification.” Synthese (2016): 1–10. https://doi. org/10.1007/s11229-016-1262-2. Booth, Anthony R. “The Theory of Epistemic Justification and the Theory of Knowledge: A Divorce.” Erkenntnis 75, no. 1 (2011): 37–43. Booth, Anthony R. and Rik Peels. “Why Responsible Belief Is Permissible Belief.” Analytic Philosophy 55, no. 1 (2014): 75–88. Bortolotti, Lisa. “Stranger than Fiction: Costs and Benefits of Everyday Confabulation.” Review of Philosophy and Psychology 9 (2018): 227–249. https://doi. org/10.1007/s13164-017-0367-y. Bortolotti, Lisa and Matthew Broome. “A Role for Ownership and Authorship in the Analysis of Thought Insertion.” Phenomenology and the Cognitive Sciences 8, no. 2 (2009): 205–224. Breyer, Daniel and John Greco. “Cognitive Integration and the Ownership of Belief: Response to Bernecker.” Philosophy and Phenomenological Research 76, no. 1 (2008): 173–184. Carter, J. Adam. “On Behalf of Controversial View Agnosticism.” European Journal of Philosophy (2018): 1–13. https://doi.org/10.1111/ejop.12333. ———. “A Modal Account of Luck Revisited.” Synthese 194, no. 6 (2017): 2175–2184. Church, Ian M. and Justin Barrett. “Intellectual Humility.” In Everett L. Worthington Jr, Don E. Davis & Joshua N. Hook (eds.), Routledge Handbook of Humility, 62-75. Springer, 2016.

302

Guy Axtell

Church, Ian M. and Robert J. Hartman. The Routledge Handbook of the Philosophy and Psychology of Luck. London, UK: Routledge Press, 2019. Cohen, Gerald A. “Paradoxes of Conviction.” In If You’re an Egalitarian, How Come You’re So Rich?, edited by G. A. Choen, 7–19. Cambridge, MA: Harvard University Press, 2000a. ———. If You’re an Egalitarian, How Come You’re So Rich? Cambridge, MA: Harvard University Press, 2000b. Davis, John K. “Faultless Disagreement, Cognitive Command, and Epistemic Peers.” Synthese 192, no. 1 (2015): 1–24. ———. “Subjectivity, Judgment, and the Basing Relationship.” Pacific Philosophical Quarterly 90, no. 1 (2009): 21–40. DiPaolo, Joshua and Robert M. Simpson. “Indoctrination Anxiety and the Etiology of Belief.” Synthese 193, no. 10 (2016): 3079–3098. doi: 10.1007/ s11229-015-0919-6. Engel, Mylan, Jr. “Epistemic Luck.” Internet Encyclopedia of Philosophy. www. iep.utm.edu/epi-luck/. ———. “Is Epistemic Luck Compatible with Knowledge?” Southern Journal of Philosophy 30, no. 2 (1992): 59–75. Fantl, Jeremy. Limitations of the Open Mind. Oxford, UK: Oxford University Press, 2018. Faulkner, Paul and Thomas Simpson. (eds.). The Philosophy of Trust. Oxford: Oxford University Press, 2017. Fraser, Rachel E. “Testimonial Pessimism.” In Knowledge, Belief, and God: New Insights in Religious Epistemology, edited by Matthew A. Benton, John Hawthorne, and Dani Rabinowitz, 203–227. Oxford, UK: Oxford University Press, 2018. Freedman, Karyn L. “Testimony and Epistemic Risk: The Dependence Account.” Social Epistemology 29, no. 3 (2015): 251–269. doi: 10.1080/02691728.2014. 884183. Fritz, James. “Conciliationism and Moral Spinelessness.” Episteme 15(1) (2018): 101–118. Haack, Susan. “‘The Ethics of Belief’ Reconsidered.” In The Philosophy of Roderick M. Chisholm, edited by Lewis E. Hahn, 129–144. Chicago, IL: Open Court, 1997. Haidt, Jonathan. The Righteous Mind: Why Good People Are Divided by Politics and Religion. New York, NY: Vintage Books, 2012. Harman, Gilbert. “Knowledge, Reasons, and Causes.” Journal of Philosophy 67, no. 21 (1970): 841–855. Hirstein, William. Brain Fiction: Self-Deception and the Riddle of Confabulation. Denver, CO: A Bradford Book, 2005. Howell, Robert J. “Google Morals, Virtue, and the Asymmetry of Deference.” Noûs 48, no. 3 (2014): 389–415. Kahan, Dan M. “The Expressive Rationality of Inaccurate Perceptions.” Behavioral and Brain Sciences 40 (2017). ———. “Ideology, Motivated Reasoning, and Cognitive Reflection.” Judgment and Decision Making 8 (2013): 407–424. Kelly, Thomas. “Evidence Can Be Permissive.” In Contemporary Debates in Epistemology, 2nd ed., edited by Matthias Steup, John Turri, and Ernest Sosa, 298– 311. Hoboken, NJ: Wiley-Blackwell, 2013.

Well-Founded Belief

303

Kidd, Ian J. “Charging Others with Epistemic Vice.” The Monist 99, no. 3 (2016): 181–197. ———. “A Phenomenological Challenge to ‘Enlightened Secularism’.” Religious Studies 49, no. 3 (2013): 377–398. Kölbel, Max. “Faultless Disagreement.” Proceedings of the Aristotelian Society 104 (2004): 53–73. Kopec, Matthew and Michael Titelbaum. “The Uniqueness Thesis.” Philosophy Compass 11, no. 4 (2016):189–200. Lackey, Jennifer. “Experts and Peer Disagreement.” In Knowledge, Belief, and God: New Insights in Religious Epistemology, edited by Matthew A. Benton, John Hawthorne, and Dani Rabinowitz, 228–245. Oxford, UK: Oxford University Press, 2018. Mawson, Tim J. “Mill’s Argument against Religious Knowledge.” Religious Studies 45, no. 4 (2009): 417–434. McCain, Kevin. “The Virtues of Epistemic Conservatism.” Synthese 164, no. 2 (2008): 185–200. Mill, John S. On Liberty. BLTC Research. www.utilitarianism.com/ol/three.html (accessed October 30, 2017). Mogensen, Andreas L. “Contingency Anxiety and the Epistemology of Disagreement.” Pacific Philosophical Quarterly 98, no. 1 (2017a): 590–611. ———. “Moral Testimony Pessimism and the Uncertain Value of Authenticity.” Pacific Philosophical Quarterly 95, no. 2 (2017b): 261–284. Montaigne, Michel de. The Essays of Montaigne, translated by Charles Cotton, edited by William C. Hazlitt. e-books@Adelaide. https://ebooks.adelaide.edu. au/m/montaigne/michel/essays/contents.html. ———. Apology for Raymond Sebond, translated by Roger Ariew and Marjorie Grene. Indianapolis, IN: Hackett Publishing, 2003. ———. The Complete Essays of Montaigne, translated by Donald Frame. Redwood City, CA: Stanford University Press, 1958. Pelling, Charlie. “Assertion, Telling, and Epistemic Norms.” Australasian Journal of Philosophy 92, no. 2 (2014): 335–348. ———. “Assertion and Safety.” Synthese 190, no. 17 (2013): 3777–3796. Pittard, John. “Conciliationism in Religious Disagreement.” In Michael Bergmann and Patrick Kain (eds.) Challenges to Moral and Religious Belief: Disagreement and Evolution, 80–99. Oxford: Oxford University Press, 2014. Plantinga, Alvin. “Pluralism: A Defense of Religious Exclusivism.” In The Rationality of Belief and the Plurality of Faith, edited by T.D. Senor. Ithaca, NY: Cornell University Press, 1995. Pritchard, Duncan. “Epistemic Risk.” The Journal of Philosophy 113, no. 11 (2016a): 550–571. ———. “Anti-Luck Virtue Epistemology and Epistemic Defeat.” Synthese 195, no. 7 (2016b): 3065–3077. https://doi.org/10.1007/s11229-016-1074-4 ———. Epistemic Luck. Oxford, UK: Oxford University Press, 2005. Pronin, Emily, Thomas Gilovich, and Lee Ross. “Objectivity in the Eye of the Beholder: Divergent Perceptions of Bias in Self versus Others.” Psychological Review 111, no. 3 (2004): 781–799. Pronin, Emily, Justin Kruger, Kenneth Savitsky, and Lee Ross. “You Don’t Know Me, But I Know You: The Illusion of Asymmetric Insight.” Journal of Personality and Social Psychology 81, no. 4 (2001): 639–656.

304

Guy Axtell

Pronin, Emily, Daniel Y. Lin, and Lee Ross. “The Bias Blind Spot: Perceptions of Bias in Self versus Others.” Personality and Social Psychology Bulletin 28, no. 3 (2002): 369–381. Quinn, Philip. “Fideism.” In Oxford Companion to Philosophy, 2nd ed., edited by Ted Honderich. Oxford, UK: Oxford University Press, 2007. ———. “Epistemic Parity and Religious Argument.” Philosophical Perspectives 5 (1991): 317–341. Rawls, John. Political Liberalism. New York: Columbia University Press, 1993. Riggs, Wayne D. “Epistemic Risk and Relativism.” Acta Analytica 23 (2008): 1–8. Rinard, Susanna. “Believing for Practical Reasons.” Noûs (2018a): 1–22. ———. “Pragmatic Skepticism.” Philosophy and Phenomenological Research forthcoming (2018b). ———. “Equal Treatment for Belief.” Philosophical Studies (2018c): 1–28. Sher, George. “But I Could Be Wrong.” Social Philosophy and Policy 18, no. 2 (2001): 64–78. Silva, Paul. “Knowing How to Put Knowledge First in the Theory of Justification.” Episteme 14, no. 4 (2017): 393–412. Simpson, Robert M. “Permissivism and the Arbitrariness Objection.” Episteme 14, no. 4 (2017): 519–538. Srinivasan, Amia. “The Archimedean Urge.” Philosophical Perspectives 29 (2015): 325–362. Tucker, Chris (ed.). Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism. Oxford, UK: Oxford University Press, 2013. ———. “Phenomenal Conservatism and Evidentialism in Religious Epistemology.” In Evidence and Religious Belief, edited by K. J. Clark and R. J. VanArragon. Oxford: Oxford University Press, 2011, 52–73. Vavova, Katia. “Irrelevant Influences.” Philosophy and Phenomenological Research 96, no. 1 (2018): 134–152. Wallbridge, Kevin. “Sensitivity, Induction, and Miracles.” Australasian Journal of Philosophy 96, no. 1 (2018): 18–126. White, Roger “Evidence Cannot Be Permissive.” In Contemporary Debates in Epistemology, edited by Matthias Steup and John Turri. Oxford, UK: Blackwell Publishers, 2013, 312–323.

16 The Epistemic Basing Relation, and Knowledge-That as Knowledge-How Stephen Hetherington

1 Introduction One cannot dwell for long among contemporary epistemologists without talking about the nature of epistemic justification. It can be an object of epistemological interest in its own right. Or it might be paired with the nature of knowledge, so that we are trying to understand both knowledge and justification at once. The latter pairing is both conceptually and historically grounded, and I will begin by discussing that historical dimension, before moving into the conceptual realm. This order of inquiry is not random: the aim is to enrich the conceptual realm by attending to that historical dimension. We will focus on a specific aspect of justification and knowledge—namely, the basing relation, as this contributes to epistemic justification and thereby to knowing. When epistemologists reflect on where it all began for them—including the entry into philosophy of an idea of epistemic justification—most generally they hearken back to Plato’s Meno.1 There we find a striking suggestion by Socrates about the nature of knowing. His suggestion—that knowing includes not only a true belief but also a sort of grounding of that belief—has become part of epistemological lore. Even so, has that Socratic lesson been fully absorbed? I suspect not, and so this chapter will try to do fuller justice to what Socrates was saying about knowledge and (what we refer to as) epistemic justification. The result will be what I call a practicalist conception of both Socrates’ lesson and the epistemic basing relation as it functions within knowing. I will argue for an account of epistemic basing that locates it naturally within my practicalist conception of knowing.

2 The Meno and Basing The textbook parsing of Socrates’ epistemological lesson in the Meno is this. When knowledge is needed, part of what is needed is a logos. This term is generally translated as ‘account’. A logos is a story, however full, about

306

Stephen Hetherington

how the accompanying opinion is true. In Dominic Scott’s (2006: 179) translation, a logos is ‘explanatory reasoning’—explanatory of how the opinion in question is true. In the absence of a logos, argued Socrates, one has at most a correct opinion—what we now call a true belief. Once that absence is rectified, knowledge is the outcome. Knowing is thus one’s possessing both a true belief and an associated logos. Fast forward, through several centuries of philosophy: at least some of that Socratic conception of knowledge is still present. At the core of epistemological orthodoxy we find the thesis that knowledge is not only a true belief—for it also includes the presence of good (epistemic) justification. This thesis is often called a justificationism about knowledge’s nature. For most epistemologists, it is undeniably true.2 Let us now revisit Socrates’ famous reasoning for his version of that thesis (Meno 97e–98a):3 To acquire an untied work of Daedalus is not worth much . . . for it does not remain, but it is worth much if tied down, for his works are very beautiful. What am I thinking of when I say this? True opinions. For true opinions, as long as they remain, are a fine thing and all they do is good, but they are not willing to remain long, and they escape from a man’s mind, so that they are not worth much until one ties them down by (giving) an account of the reason why. And that, Meno my friend, is recollection, as we previously agreed. After they are tied down, in the first place they become knowledge, and then they remain in place. That is why knowledge is prized higher than correct opinion, and knowledge differs from correct opinion in being tied down. Socrates’ reasoning is centred upon the idea of Daedalus’ statues. These were distinctive. If not ‘tied down’, they would ‘not remain’: possessing the power of self-movement, they would run away!4 It is little wonder, then, that there would be notable value in possessing such a statue—so long as one could tie it in place, thereby keeping it under one’s control. In effect, only a tethered statue5 by Daedalus has significant value—being ‘worth much’. And such a statue, claimed Socrates, was analogous to knowledge: an untethered Daedalus-statue is like a true belief; a tethered one is like knowledge. Moreover, just as a tethered—not an untethered— Daedalus-statue is what we would want, it is generally knowledge—not merely a true belief—that we want. Although that is a dramatic analogy, it does not tell us why we want knowledge. Even so, perhaps it can help us. In order to understand Socrates’ answer to the question of why we want knowledge, we should stay with his analogy, asking why a tethered status by Daedalus would be so valuable (as Socrates takes it to be).6 Once we do, we should find Socrates’ tale increasingly instructive: we will see that justificationism

The Epistemic Basing Relation

307

is not the only significant epistemological moral to take from Socrates’ analogy. In particular, it will matter that Socrates’ tale portrays our wanting knowledge, rather than merely true belief, because knowledge will help us to perform our intended action skilfully. Imagine what Socrates implicitly asks us to imagine. One embarks on the travelling, following the correct path—always aware, of course, of the danger that one’s belief, even if true (about the direction to follow), will leave, perhaps mid-journey. Once one is stranded, bereft of that true belief, one might curse regretfully—but in vain—that one had set out on the journey with only a true belief, not knowledge, as to how to complete the journey. So, of course we need knowledge, not mere true belief. Taking our cue from Socrates, then, the point of requiring a logos is not simply to mark the difference between a true belief and knowledge. The deeper point is to characterise that difference. And surely we may take from Socrates’ story the deeper point that a logos allows knowledge definitely to help us as we wish to be helped—and as a mere true belief need not help us—in performing an action. As Socrates sees it, a true belief unaccompanied by a logos is apt to run away, like an untethered Daedalus-statue. In preventing this, the logos is a tether. Only with its presence can there be knowledge; and Socrates offers no reason to wish for the presence of the logos, as necessary for the knowledge’s being present, other than the claim that with knowledge there will definitely be a successful journey—a predictably completed action, of travelling to Larissa. But note the use just now of ‘only’—instead of ‘if’ or ‘if and only if’. Again, the moral standardly taken from Socrates’ reasoning is justificationist: a true belief accompanied by a logos—which contemporary epistemologists parse as epistemic justification accompanying the true belief—is necessary for knowing. Yet Socrates also talks as if a true belief’s being accompanied by a logos suffices for its being knowledge— hence, for one’s having whatever knowledge is necessary for accomplishing the action of travelling to Larissa. And in fact most contemporary epistemologists would not quite concur with Socrates here. We have been professionally trained to deny that sufficiency claim: we will say ‘That sufficiency claim fails, because even a true belief with a logos could be Gettiered, thereby failing to be knowledge.’ That is a large epistemological topic, and this chapter will set it aside. We will bypass the challenge that was, and possibly still is, posed by Edmund Gettier’s (1963) questioning of how to define what it is to know a fact.7 Even apart from Gettier’s challenge, though, there is another reason why epistemologists might deny that sufficiency claim. Most will say that even a true belief accompanied by a logos need not be properly based on that logos. In which case, the fact of a true belief’s being accompanied by a logos is insufficient for its being knowledge: the true belief is not being justified in a way that already (even with all else being equal) makes it knowledge.8 If we are to understand knowing, therefore, as Socrates was

308

Stephen Hetherington

aiming to do, we need to understand what it is for a belief to be properly based on a logos—not merely what it is for a logos to be present. Not even a logos in the believer’s mind, for instance, suffices. It has to have played a proper role. The belief needs to be properly based on the logos. It is elegant to formulate the issue in that Socratic way. We may also formulate it in more contemporary terms, via the distinction between epistemic internalism and epistemic externalism. Hence we might formulate this explicative challenge disjunctively. We need to understand what it is for a belief to be properly based (i) on some good, and consciously present-to-mind, evidence or reason, or (ii) in some justificatory circumstance (such as by being reliably produced). Alternative (i) would be internalist proper basing. Alternative (ii) would be externalist proper basing. In what follows, I will generally talk in terms of (i), but the picture to be developed applies to both.9

3 Basing as an Action We may begin our explicative journey by asking about what kind of relation the proper basing relation is. This question need not concern the relation’s precise details or its applicability. The question is a more general conceptual one. It is metaphysical, about an epistemic phenomenon. And a first step towards answering it should be our recognising that, in Socrates’ terms, what is at stake is not a tether as such, in the sense of a logos being present alongside the pertinent true belief. What is also vital, for knowledgeably heading to Larissa, is that there be a tethering. The action of tethering, not merely the result of that action, is what matters. This is the difference between a logos being present with the potential to have been used in tethering, and that same logos being present because it has been used in tethering. Presumably, Socrates has in mind the latter. He uses relevantly active language. One needs to tie down the true opinion. One needs to be ‘(giving) an account of the reason why’ (my emphasis). Only then—only once it has been actively tied down—is a true belief knowledge. This is seemingly what Socrates would say. Interpretive charity tells us that he would not have regarded it as enough, if knowledge is to be present (with the correlative benefits for successful action), for the person simply to have in mind a logos, even one that is now sitting alongside an apparently appropriate true belief. It seems as if Socrates would have endorsed the applicability to the situation of a conception of what we would call proper basing. For what also clearly matters to him is the prior process—how the true belief’s presence has been achieved. It needs to have arisen due to a use of the logos. What matters here is something active. In contemporary terms, the Socratic point is this: it is not enough, if knowledge is to be present, that a true belief is accompanied by good evidence, say, or that a true belief is present within an epistemically apt

The Epistemic Basing Relation

309

circumstance. There needs to have been proper basing; which means at least that the true belief has come into existence (and is in place) as a proper effect of that evidence or that circumstance. And we may interpret ‘proper’ here as ‘actively generated’: in some sense of ‘appropriate’, some appropriate action has generated the true belief, for a start. We may now link that with the idea of justification: The true belief has been properly based on that evidence, or properly based in that circumstance. This constitutes the justification’s contributing to the knowledge, in that there has been justifying of the true belief. This justifying can be thought of as an activation of what is, through the mere presence of the evidence, for instance, a potential for justifying the true belief. We thus have a distinction between (i) the presence of a logos, or more generally of evidence, with a consequent potential for a justified belief to be generated, and (ii) that potential having been fulfilled, with an actually justified belief having arisen. And this distinction brings us to the contemporary distinction between doxastic justification and propositional justification. We were bound to reach this point sooner rather than later; for the proper basing relation—this chapter’s explicandum—marks that distinction. Although overtly a contemporary distinction (witness those terms ‘doxastic’ and ‘propositional’), it can also be formulated, well enough, in the terms used by Socrates’ thinking in the Meno. Propositional justification stands to doxastic justification, much as a logos being present to one’s mind, say, stands to a tethering of that logos having occurred for one’s mind: once the tethering has occurred, one has used the logos properly, turning it into a tether (as against its staying merely as internal content or external circumstance). In short, when there is doxastic justification for a belief, propositional justification for the belief has been used properly. Let us now parse this distinction along the following (nonSocratic) lines: When a belief is doxastically justified, some activity—perhaps more reflectively and consciously undertaken;10 perhaps more automatically and unconsciously performed11—has taken the epistemic agent from (1) a state where she has good evidence, or is in a justificatory circumstance (thereby having propositional justification), to (2) a state where she has the belief—somehow properly—on the basis of (1). Relevant journeys can now begin: a physical journey, such as to Larissa, can be guided by what has already been an epistemic journey. A logos would have properly generated a true belief (so that the belief has

310

Stephen Hetherington

been properly based on that logos). The true belief, now that it is knowledge (provided that all else is equal), can be used aptly, properly generating accurate movements towards one’s physical destination. Hopefully, all of this can occur for as long as is required, until—finally—one reaches Larissa.

4 Basing as Skilful Action I have been using the word ‘properly’, because it is the word used by contemporary epistemologists in this setting. But what is a proper basing activity, if justification is to be present, especially as part of knowledge’s being present? In this setting, we should understand ‘properly’ as ‘skilfully’. This was not Socrates’ term, but it fits well with his thinking here. Reflect again on that journey to Larissa. How is it to be accomplished? For it is indeed to be an accomplishment. So, it is indeed to be a result of skill—to have been performed skilfully.12 What are the ingredients in this? Again we look to what Socrates says about the need for the logos. The point of the logos (as featured in his argument) is to contribute to the journey’s skilful performance. The point of the logos is not simply to be present, even throughout the travelling. There is no envisaging of the journey as merely a sequence of correct movements (that is, movements that happen not to be leading one astray)—each occurring as what might be a purely fortuitous response to a persisting logos, with the correlative possibility of their sum amounting just to a fortuitously completed task. On the contrary: the Socratic picture is of a deliberate, guided, and accomplished action, albeit a complex one. The journey is to flow smoothly and predictably, with the agent holding the logos in mind and using it properly so that it contributes to that flow, that predictability, that outcome. Accordingly, Socrates’ picture is one whereby knowledge’s value is its link to action—and whereby the value in the logos contributing to the knowing is also this link to that action.13 This realisation allows us to introduce, into our Socratically inspired discussion of knowledge, an idea that has received much attention in recent epistemological writing—the idea of knowledge-how. When we say that Socrates’ goal in this part of the Meno is to understand knowledge’s nature, we mean that he seeks to understand what we call knowledge-that (propositional knowledge). How does talk of knowledgehow enter the conversation? Can the idea of knowledge-how be blended into an explication of knowledge-that? Knowledge-that has long been distinguished by epistemologists from knowledge-how: the former is said to be contemplative in form (being knowledge that something is the case), while the latter is practical in form (being knowledge how to do some sort of action). Will this potentially categorial difference between two kinds of knowledge be a problem for my line of thought?

The Epistemic Basing Relation

311

Not at all. In order to see why, let us focus on the initial way in which the idea of knowledge-how seems to have entered the discussion. The journey to Larissa is an action (a temporally extended and mechanically complex one). So, it can be performed more, or less, skilfully. Presumably, this would be partly a function of the agent’s use of the knowledge (the knowledge-that, thanks to the logos, the tethering) that is deemed by Socrates to be distinctively valuable: for him, knowledge’s value is at least partly its power to guide to completion that action of travelling. But what we should now acknowledge is that this guiding-to-completion would not reflect purely one’s knowledge, or even one’s having the knowledge. What matters is how one uses the knowledge-that. For example, even having the same map, two travellers could use it in ways that would, more or less skilfully, accomplish their shared goal of arriving safely at Larissa. Other things being equal, the one who uses the knowledge-that (the map) more intelligently will have travelled more skilfully. How could that happen? We might say that the journey would be performed more or less sagely, by the respective travellers, as the complex action that it is. One early stage in this skilful—this more, or this less, sagely performed—action is one’s knowing how to convert or translate the map—the specific logos—into a true opinion as to how best to proceed, before proceeding to proceed. Even once one has evidence, or is in an appropriate circumstance, one needs to know how to use that evidence or that circumstance sagely; and having this knowledge-how, I suggest, is literally part of having the knowledge-that. For instance, how well does the evidence support one belief rather than competing ones, or rather than belief-suspension? We can describe logical relations between the contents of evidence and belief. Nonetheless, this leaves us only with a description of some propositional justification. What is needed for describing some doxastic justification, and thereby some knowledge-that, is our citing the use of a skill—knowledge-how—in believing in accord with one’s evidence or one’s circumstance insofar as true belief (or Larissa, as one’s desired destination) is sought. One would, for example, know to believe more or less strongly that ‘here’ is where one should turn left rather than right, reflecting one’s evidence. This knowing-to might be part of the knowing-how. One would be manifesting or expressing one’s pertinent skill—some knowing-how— even in knowing that ‘here’ is where to turn left, en route to successfully performing one’s intended action.14 On that programmatic picture, knowing-that includes knowing-how. The rest of the chapter examines some, but not all, aspects of the picture.15 Here is what we have so far. •

Properly basing a belief on evidence, or in a circumstance, is one’s skilfully gaining that belief in light of that evidence, or in response to that circumstance.

312 •

•

• • •

Stephen Hetherington We are following Socrates’ lead, both explicit and implicit, in thinking of that combination (other things being equal) as knowledge. (What is implicitly Socratic within it is the mention of skill.) Yet this makes knowledge—in the sense of knowing-that—partly an exercise of knowledge-how. This is so, in virtue of the knowledgethat’s including one’s ‘skilfully gaining that belief’. For such an action is a manifestation or expression of pertinent knowledge-how. Somehow, therefore, having some knowledge-how is part of having knowledge-that. How is that so? Before we answer that question, we should note a fundamental implication of it: knowing is the preserve of an epistemic agent (rather than an epistemic subject).16

5 Knowledge-Practicalism Section 4’s guiding suggestion is that proper basing is a kind of knowledgehow—a suggestion that soon leads, we saw, to a picture of knowing-that as including some knowing-how. But Section 4 also mentioned that epistemologists have long treated knowledge-that and knowledge-how as in categorial conflict. We should therefore ask whether our picture so far is coherent, given that possible metaphysical datum about the warring natures of knowledge-that and knowledge-how. That need is increased by the fact that Section 4’s suggestion seems at odds with some views of the metaphysical relationship between knowledge-that and knowledge-how. Intellectualism about knowledgehow—particularly in the hands of Stanley and Williamson (2001) and Stanley (2011)—seems to impress many epistemologists. It regards all knowledge-how—its ‘knowledge aspect’—as ultimately knowledgethat.17 So, consider my hypothesis that proper basing (which we are supposing, in an epistemologically standard spirit, to be needed within knowledge-that) is a form of knowledge-how. An intellectualist about knowledge’s nature would treat that hypothesis as reducing just to a picture of any knowledge-that’s needing to include further knowledgethat. This would conduce to what might seem like welcome metaphysical homogeneity within any instance of knowing: the need for proper basing within knowing would amount to some further knowledge-that being nested within knowledge-that. An intellectualist might claim that she can adopt this picture while retaining the fact of knowledge-how being action-orientated. I will not critically evaluate that sort of claim here, though; I will take the direction in a contrary constructive direction. I will explain my preferred way of conceiving of the metaphysical relationship between knowledgehow and knowledge-that—and how this conception can accommodate Section 4’s suggestion for conceiving of proper basing as a component within knowing-that. I call this larger picture knowledge-practicalism. It

The Epistemic Basing Relation

313

conceives of knowing-that as entirely a matter of knowing-how: knowledge-that is a kind of knowledge-how—in this sense, of practical knowledge. I have developed this picture elsewhere,18 but without discussing the basing relation in detail. Before seeing how to conceive of the basing relation in knowledge-practicalist terms, we need a shared sense (as follows) of the general knowledge-practicalist picture. Again, its basic idea is that any instance of knowledge-that is an instance of knowledge-how. Here are some forms that this can take. First, a given instance of knowledge-that can be a complex case of knowledge-how. Not only can a given kind of knowledge-how be complex in itself; knowledge-practicalism allows that a given instance of knowledge-that could be a complex of different kinds of knowledge-how. (I will provide an example in a moment.) Second, two instances of knowledge that p, even for the same p, can differ in the respective kinds of knowledge-how by which each is constituted. The kinds of knowledge-how constituting your knowing that there is an ibis in front of you might differ from those constituting my knowing that there is an ibis in front of me. This is so, even leaving aside differences in the indexically determined contents of our instances of ibis-here-now-knowledge. Thus, think of what directly expressive skills we might expect a person to have, in knowing that p: she can answer various questions bearing upon p, and/or she can pose such questions, pointedly and aptly, and/or she can reason in ways that reflect and build upon accepting that p, and/or she can describe to herself more or fewer aspects of what p involves, and/or she can guide herself and others around the world in p-related ways (such as by travelling to Larissa!), and/or . . . etc.19 Then we need to make the following metaphysical choice. •

•

Causally-Produced. We might say that some or all such skills are present simply because one has the knowledge that p—as a result of having the knowledge. Constitutively-Inherent. Or we might say that some or all such skills are present in one’s having the knowledge that p—as constitutive elements within the knowing.

Which is it to be? Most contemporary epistemologists would regard the first alternative as the traditional choice. The second choice is the knowledgepracticalist option. On Causally-Produced, knowing can have its constitutive nature independently of these (and other) skills; and then, hopefully because one has some such skills, one can use the knowledge in appropriate ways, manifesting or expressing those skills—those kinds of knowledge-how.

314

Stephen Hetherington

According to Constitutively-Inherent, knowing’s nature is these (and other) skills; and then, hopefully, one can use the knowledge in these ways, manifesting or expressing those skills—those kinds of knowledge-how. (For example, we might say that one can answer questions bearing upon p only because one already knows that p.) But Constitutively-Inherent (the knowledge-practicalist alternative) also implies that, in performing any such actions, one is ipso facto manifesting or expressing the knowledgethat (since the knowledge-that is literally those skills, those instances of knowledge-how). This already makes knowledge-practicalism a prima facie preferable way to conceive of knowledge (knowledge-that), insofar as—like Socrates in the Meno—we see the value and thereby the nature of knowing as its link with action. Knowledge-practicalism makes that link essential to the nature of knowing. As mentioned, most epistemologists would point to Causally-Produced as the traditional alternative. Standardly, we are taught as students to think of knowledge as having an intrinsic nature that can obtain, regardless of whether the knowledge, with that nature, is ever used in action. (For example, we might be told that knowing is a justified true belief satisfying an anti-Gettier condition. This description will probably make no essential mention of the knowing’s potential to be used in action.) Perhaps ironically, however, there is a sense in which Constitutively-Inherent—the practicalist alternative—should be seen as traditional. For (I have been arguing) it allows us to do better justice to the Socratic picture in the Meno—as venerable an epistemological picture as there is! After all, the practicalist alternative makes knowing a matter—intrinsically so—of being able to perform various actions. And, again, this was clearly the ground on which Socrates accorded such respect to knowing rather than to having a mere true belief. On Causally-Produced (the non-practicalist alternative), those associated skills—those kinds of knowledge-how— amount to metaphysically extrinsic extras: they are merely associated with the knowing-that, in that one could know-that without having some such range of knowledge-how. For knowledge-practicalism, in contrast, some such array of knowledge-how is metaphysically intrinsic to the knowledge-that. So the choice is clear. Insofar as we seek a conception of knowing that embraces an essential or intrinsic link between knowing and action, we should not settle for a non-practicalist picture—Causally-Produced—on which the best that we can say is that, by knowing that p, one is causally enabled to have some such skills. Instead, we should embrace a (practicalist) picture—Constitutively-Inherent—on which, in knowing that p, one has some such skills.

6 Proper Basing and Knowledge-Practicalism With Section 5 having conveyed a basic sense of what knowledge-practicalism is (and a prima facie reason why we should consider adopting

The Epistemic Basing Relation

315

it), let us investigate further the chapter’s proposal that the proper basing relation is a kind of knowledge-how. I took my methodological cue from Socrates, conceiving of the relation initially as a constitutive element within knowing. I then argued that we can think of knowing in practicalist terms, so that knowing (even a single instance of knowingthat) is ‘knowledge-how all the way down’. Knowledge-practicalism thus provides a natural conceptual home for the idea that proper basing is a kind of knowledge-how, nestled within knowledge-that. We saw how, on knowledge-practicalism, an instance of knowing-that can be a complex of kinds of knowledge-how. What we may now include in that complex is some form(s) of the proper basing relation, conceived of as knowledge-how. For example, one might have formed a belief that p, as part of knowing that p, by basing the belief properly on one’s evidence. One would thereby have manifested or expressed a pertinent skill—apt knowledge-how.20 One knew how to use one’s evidence. Or one knew how to think within one’s circumstance.21 One then acted in a way that manifested or expressed this knowledge-how. Talking in that way of manifesting or expressing knowledge-how is particularly important in this setting, given what I am trying to respect as the Socratic emphasis on linking knowing with action. Any such action is what Gilbert Ryle (1946, 1949: ch. 2) called an intelligent action. Such actions are ‘activities which directly display qualities of mind’ (1949: 26). He was not restricting himself to discussing what, in everyday settings, we deem intelligent actions, such as answering exam questions well. His aim was to understand how any action can be intelligent in the more general sense of manifesting or expressing some form of knowledge-how. How can knowledge-how ever be manifested or expressed in action? Ryle argued that what is not always needed is the independent presence within an agent of related knowledge-that, distinct from the knowledge-how but nevertheless guiding and overseeing the manifestation or expression (in what is thereby intelligent action) of the knowledge-how. Knowledgehow is its own metaphysical creature, categorially distinct from knowledge-that. What Ryle did not consider, however, was the possibility of a knowledge-practicalism, on which knowledge-that is itself knowledgehow, so that there is no need to posit two categorially distinct kinds of knowledge—because all of it is knowledge-how. This perspective may then prompt us to regard proper basing within knowing-that as likewise just more knowledge-how. We will be doing so as part of an independently motivated conception of knowing-that, in its metaphysical entirety. When knowing a fact has involved properly basing a belief on good evidence, say, one has acted skilfully in a way that manifests or expresses a skill—knowledge-how—that one has in using this (and sufficiently kindred) evidence to form, or not to form, a belief in accord with the content and epistemic standing of one’s evidence. This continues to sound like a sensible way to conceive of the proper basing relation.

316

Stephen Hetherington

7 Proper Basing and Justification-Skills Discussions of the proper basing relation usually focus on it directly as a component in epistemic justification. Specifically, proper basing is held to be a distinguishing feature of doxastic justification. On the standard tale, a belief could be propositionally justified without being doxastically justified, and what transforms the former into the latter is the addition of proper basing. Basing a belief properly on propositionally good evidence, say, produces a doxastically justified belief; previously, one had only propositional justification for that belief. This distinction is of epistemological interest in itself, if we wish to understand epistemic justification apart from its potential or actual role within knowing. However, this chapter’s approach has been to focus directly on knowledge, not justification. Is this a methodologically sound approach? I believe so. First, the Meno’s Socratic reasoning may be viewed as treating (what we call) justification’s nature as flowing from its role within knowing. Socrates talks of a logos only in that knowledge-instrumental way: it is a tethering tool, hence (as Socrates sees it) a tool for constituting knowledge from what is otherwise merely a true belief. And proper basing is proper tethering. So, it enters our Socratically inspired epistemological discussion only as helping to constitute knowing. Just as Socrates does not treat tethering as having independent epistemological significance, nor need we regard basing any differently, insofar as we are working with what is of independent and foundational epistemological significance— ‘foundational’ in a methodological sense, as we reflect on the genesis and nature of our epistemological past and options. In that respect, I take Socrates’ story to be an ur-story as we enter this epistemological valley. Without holding in mind the aim of understanding the nature and (action-oriented) point of knowing, we would lack a simple way at the outset to recognise, and to describe the value in, a tether. Putting that point in contemporary language: it is natural to aim first to understand justification in terms of its actual or potential contribution to knowing.22 We have seen how this is a Socratic moral—a Meno moral. It is also what current epistemologists call a knowledge-first moral.23 It says that the primary conceptual home for talk of proper basing is in attempts to understand knowledge. We describe knowing; if need be, we then talk of justification, describing its role within knowing; if need also be, we then talk of proper basing, describing its role within justification’s playing its role within knowing. Still, having followed that path, we are free to retrace our steps, so as to isolate the nature of justificatory proper basing—proper basing insofar as it helps to constitute the presence of justification, considered independently of any metaphysically constitutive contribution to knowing. Towards this end, my roughly formulated and programmatic proposal for understanding purely justificatory proper basing has these two steps.

The Epistemic Basing Relation

317

(1) One is propositionally justified in believing that p, if and only if (i) one has strongly p-favouring evidence, and one knows how to believe in accord with it, or (ii) one is in a strongly p-favouring circumstance, and one knows how to believe in accord with it. In short, for one, the presence of the evidence or the circumstance enables one (with all else being equal) to form skilfully a belief that p, and to do so as an expression of that ability. (2) One’s belief that p is doxastically justified, if and only if (i) one is propositionally justified in believing that p, and (ii) one has acted on that knowledge-how (the knowledge-how within that propositional justification), so that one has gained the belief that p by way of a manifestation or expression of that knowledge-how. In short, one has put into effect—expressed—the ability that was a vital part of one’s being propositionally justified in believing that p. That brief exploratory account is practicalist in its underlying thinking: note the presence of some of the key terms that peppered our earlier discussion of knowledge-practicalism. We are portraying the possession of propositional justification as a potential—a potential constituted by knowledge-how—for having doxastic justification. The knowledge-how within that propositional justification is partly (like any knowledge-how) a potential for the skilful bringing-about of manifestations or expressions of it. Doxastic justification is thus present only when some such manifestations or expressions are at least part of how the person has gained the pertinent belief. The belief has been properly based on the evidence, or in the circumstance.

8 Unactivated Knowledge, Activated Knowledge, and Active Knowledge We can extend Section 7’s programmatic proposal about justification, focusing now on knowledge alone. That proposal’s way of distinguishing propositional from doxastic justification suggests a correlative distinction between two kinds of knowledge. I call this the difference between unactivated knowledge that p and activated knowledge that p. Here is the basic idea:24 With all else being equal, an agent has unactivated knowledge that p when she knows that p but her justification for believing that p is propositional. (Her justification is thereby a kind of knowledge-how, as characterised in Section 7. But it is unactivated justification—not yet having resulted in actions that would skilfully manifest or express it, en route to her believing that p. She already believes that p.) With all else being equal, an agent has activated knowledge that p when she knows that p and her justification for believing that p is

318

Stephen Hetherington doxastic. (So, her belief that p has resulted from a process that included skilfully manifesting or expressing her propositional justification— thereby activating her previously unactivated knowledge that p—as characterised in Section 7. She has done this by basing her belief properly on strongly p-favouring evidence, or in a strongly p-favouring circumstance.)

Why might this difference matter, between knowing unactivatedly and knowing activatedly? I will offer a slightly metaphorical answer, returning us to where this chapter’s discussion began. Like Western epistemology in general, we started with Socrates and his metaphor of Daedalus’ statues. Western epistemology has said much since it started in that way; sometimes, though, we should revisit a tradition’s beginning-point, if only to see whether we have overlooked something significant within it, something that might have encouraged subsequent epistemology, for example, to adopt a different shape. I suggest that the difference between unactivated knowledge and activated knowledge can build upon—and redirect—our shared Socratic ur-story in that way. First, we may say that to have knowledge that has recently been activated is to have knowledge still with momentum within it. The belief would be a product of a recent active process, recent proper basing: justificatory movement would recently have occurred, with the activating of what was until then propositional justification containing just the potential for such movement to transpire, to result either directly or indirectly in the belief.25 I say ‘recently’ to capture the possibility of that justificatory movement— the activating of the propositional justification—not yet having entirely ended. The resulting knowledge could include some ‘remnant’ movement within itself, which might make it simpler for the knowledge to be used, in accompanying or guiding further actions.26 One would be harnessing that remnant movement—redirecting it. Think here of using the knowledge as Socrates might have wanted it to be used, such as by travelling to Larissa. By activating the propositional justification, one derives a properly based belief that can be activated knowledge; and one can thereby build upon the skilful action of proper basing, by using the activated knowledge in further actions. So, a conception of recently activated knowledge fits especially well with what Socrates in the Meno was describing—the grounding of successful action in knowing (rather than merely in true belief), along with this grounding (this tethering) therefore being regarded as a constitutive feature of the knowing (a feature that distinguishes knowledge from mere true belief). The point applies to all activated knowledge, though. We might begin with unactivated knowledge that p. If we start using it properly—so that proper grounding occurs—then activated knowledge that p can eventuate. This would be an activated knowledge-state. It would not yet be knowledge being used. No matter: it would be available

The Epistemic Basing Relation

319

to be used. And if we do use that activated knowledge (as Socrates was, in effect, envisaging), we have what could be thought of as active knowledge that p, in that we are using it as part of performing an action. In this way, it would be knowledge in action. For example, we can be using activated knowledge in order to travel well to Larissa. Proper grounding turns unactivated knowledge into activated knowledge; whereupon we might proceed to have active knowledge, by using that activated knowledge aptly. This is a natural progression, with proper grounding at its core, once we conceive of the situation in practicalist terms. We can continue illustrating these issues through Socrates’ memorable thought experiment. He seemed to regard any statue by Daedalus as being at its most remarkable when untethered—when it can act as only a Daedalus-statue, among all statues, could act. Socrates seemed to think that anyone contemplating ownership of a Daedalus-statue has two choices: • •

leave it untethered—and thus risk losing it to its own whims or aims, when inevitably it runs away; tether it—tightly, so that it has no more capacity for moving and hence leaving than an ordinary statue has.

There was always a third option, though: •

tether the statue—but loosely, allowing it to move freely yet without being able to run away.

An untethered Daedalus-statue would be remarkable, Socrates acknowledges. No less so, however, would be a loosely and sympathetically tethered one. Less dramatically but more realistically, we might bring to mind a lively and intelligent dog, tethered sympathetically. Other things being equal, this sort of tethering allows the dog to lead us where it wants to go—a freedom that could be surprisingly helpful to us. Nonetheless, this tethering allows us to remain in control: the dog can be restrained from taking the lead when our aim is not well-served by its having that specific freedom. And this is likewise how a Daedalus-statue could be tethered. Such an arrangement would remain remarkable. It would allow one to use—to utilise, to build upon—the Daedalus-statue’s power of movement. One would turn the Daedalus-statue’s power into something useful for one, rather than simply a threat to one’s continued ownership and control of the statue. One might even be able to allow the loosely tethered Daedalus-statue to lead one all of the way to Larissa! This would be impossible for a too-tightly tethered Daedalus-statue. Yet that is how Socrates seems to envisage tethering any Daedalusstatue; which would turn an otherwise special statue into .  .  . an inert

320

Stephen Hetherington

lump—in effect, an ordinary statue. The Daedalus-statue would become a mere statue. Socrates talked as though one would retain the value in the Daedalus-statue by tethering it. But he was mistaken: no longer would one have a Daedalus-statue worthy of the name. Insofar as tethering was ever needed at all, this was due to the statue’s power of self-movement. Hence, at least some of this power needs to remain even after the tethering, if something remarkable is still to be present. Socrates tried to capture knowledge’s remarkability (as we could call it), by analogy with the remarkability of the Daedalus-statue, with his talk of knowledge—rather than mere true belief—being present helpfully during the journey to Larissa. Again, though, his story turned the statue into nothing more, in effect, than an inert lump; and where is the guidance, the activity, in that? In fact, where is the remarkability in that? We are not being shown a remarkable state of affairs. Consequently, insofar as (in Socrates’ hands) this combination—the Daedalus-statue plus its tether—is supposed to be modelling knowing, it undermines the idea of there being extra value in knowing. In order to do justice to Socrates’ main insight, then, we should let knowing remain (usefully) active—by analogy with the idea of letting the Daedalus-statue remain usefully active. I have conceptually prepared the way for this with the idea of the tethering—the proper grounding— within the knowing being an activity, a skilful one. This fits smoothly into the chapter’s larger model: once proper grounding is present, activated knowledge is present; so, knowledge is present in a state whereby it is available to be used actively. And if it was only recently activated, for example, it still includes a justificatory momentum within it that can be redirected into further action.27 This knowledge is then not only activated; it is active knowledge, knowing being used in action (such as by being used to guide one’s journey to Larissa). So, Socrates was even more insightful than epistemologists generally notice, in reaching for the idea of a Daedalus-statue when attempting to understand knowledge. But he used the idea less fully than he might have done. Contemporary epistemologists, too, have not noticed the further insight that was latent within Socrates’ picture. I hope that our discussion of proper basing—the tethering relation—can contribute to rectifying that long-standing oversight within epistemology.28

Notes 1. Sometimes, too, they mention the final few pages (200d–210d) in Plato’s Theaetetus, where a more specific, but less influential, version of the Meno’s suggestion is investigated. This chapter discusses only the Meno’s proposal. 2. Still, its truth has been questioned: see Sartwell (1991, 1992), Foley (1996, 2012), and Hetherington (2001: ch. 4, 2011a: ch. 4, 2018a, 2018b). 3. This translation is Grube’s (1981: 86). 4. Obviously, such statues were not real. It is not clear whether Daedalus himself was real (Bluck 1961: 409–410).

The Epistemic Basing Relation

321

5. Bluck (1961: 411) and Scott (2006: 178) talk of binding the statues. Nothing of epistemological substance in this chapter depends on that choice of term. 6. In fact, this Socratic view is odd. As Socrates conceives of the situation, to tether a statue by Daedalus is to tame it. But a tethered Daedalus-statue would be functionally like other statues—and hence not so remarkable. We will return to this point at the close of the chapter. See also Hetherington (2018a, 2020). 7. For an extensive critical discussion of post-Gettier epistemology, see Hetherington (2016). 8. For a statement of this problem, see Kornblith (1980). 9. That is appropriate, since a given epistemologist’s view could combine (i) with (ii): having some evidence in mind and using it wittingly is internalist; but also using that evidence unwittingly, ‘automatically’, or unconsciously (either reliably or unreliably) adds an externalist element to one’s view. So, I am streamlining and simplifying my discussion slightly. For a simple version of the distinction between epistemic externalism and epistemic internalism, see Hetherington (1996: chs. 14, 15). 10. Paradigmatically, this might include, for example, some self-consciously reasoned belief-formation by an epistemic agent. 11. Paradigmatically, this might include, for example, some un-self-consciously perceptually based belief-formation by an epistemic agent. 12. I thereby take my first step towards what I hope will be an epistemologically useful corralling of some inter-related ideas, by parsing ‘properly’ in this setting as ‘skilfully’. I will not then be conceptually analyzing skilfulness. (Not all epistemological progress takes that form. For an instance of such progress, I claim, that takes an alternative form, see Hetherington [2016: ch. 7].) 13. This does not entail there is no skill in travelling successfully by using merely a true belief, with no logos in mind. But that is a different form of skill, arguably a weaker one. At any rate, presumably Socrates would insist that it will remain less firmly entrenched within the traveller than will the skill of travelling successfully with a logos in mind. 14. For more on the nature of such knowing-to, see Hetherington and Lai (2015) and Hetherington (2019). 15. On the nature of knowledge-how, see, for example, Stanley and Williamson (2001), Stanley (2011), and some of the papers in Bengson and Moffett (2011). Here, I will not delve into that further debate. 16. This choice of term—‘epistemic agent’ versus ‘epistemic subject’—is epistemologically substantial. See Hetherington (2011a: 70–73). See also Leite’s (2004) argument against what he calls the Spectatorial Conception of a belief’s being justified (including its being so as part of being knowledge, I assume). On that conception (ibid.: 222), ‘positive justificatory status is something which one finds out about, not something which one brings about.’ Leite argues for a view that could be seen as a special case of part of this chapter’s: he likewise stresses the justificatorily constitutive role of an individual’s ability and willingness to undertake a continuing process of actively justifying a belief. 17. Stanley and Williamson (2017) have also extended their intellectualism so as to describe the nature of skill. They regard any skill as a disposition to have knowledge—in particular, situation-specific knowledge-wh. I do not interpret the term ‘skill’ in that intellectualist way. I am presenting a view of skills as constitutive of knowledge. And I assume in this chapter that skills need not always be constituted by (a relation to) knowledge. For more on situationspecific knowledge’s role in knowing-how, see Hetherington and Lai (2015) and Hetherington (2019) on what we call knowing-to.

322

Stephen Hetherington

18. See, for example, Hetherington (2011a: ch. 2, 2011b, 2013, 2017a, 2017b). Here, I will gesture at just a few of its main aspects. 19. This list is highly schematic. But I want to highlight one extra dimension of its flexibility. I have mentioned that the person could have many or fewer of these skills, on a given occasion. Well, she could also have any one of those skills more or less fully, being more or less skilful in the respects in which she is at all skilful on that occasion. A child might have fewer, and less developed, such skills in knowing that p, when the ‘p’ in question is more conceptually sophisticated. Even an adult might know that p quite automatically, such as when having some perceptual knowledge that p, lacking any associated skill for discussing, let alone analysing, the state of affairs p. 20. This would be an instance of knowing, in part by believing—a seemingly standard conception of knowing. But knowledge-practicalism also allows the possibility of knowing without believing—a quite unusual conception of what knowing can ever be like. On this latter possibility, see Hetherington (2018a: sec. 9). 21. We might also say (as I did earlier) that one knew to act as one did, by forming the belief that one did. For a conception of knowing-to, within a knowledgepracticalist framework, see Hetherington and Lai (2015) and Hetherington (2019). 22. This contribution, incidentally, is independent of whether the justification is always true. Perhaps there is enough—even if not only—truth involved in the justification. Precisifying this cautious optimism about justification’s not needing to be entirely falsity-free is a long-standing epistemological challenge, of course, especially for fallibilists. It is not this chapter’s aim, though. 23. Williamson (2000) is the leading contemporary advocate of knowledge-first epistemology. For critical evaluation, see McGlynn (2014). 24. For more on it, see Hetherington (2017a). 25. In saying ‘either directly or indirectly’, I am meaning to take no stand on the vexed but not immediately vital question of whether doxastic voluntarism could be true. 26. This condition—of recent activation—is not necessary for the knowledge’s being used. I am describing just one illustrative way for that to occur, before focusing on activated knowledge more generally (not only recently activated knowledge). 27. Recently activated knowledge would still be quivering, like a husky who can barely contain its excitement at the day’s sledding that is about to begin. This makes it a touch easier for that knowledge to be used as part of further actions. 28. I am grateful to Pat Bondy, Adam Carter, and Hamid Vahid for very helpful comments on earlier versions of this chapter.

References Bengson, J. and Moffett, M.A. (eds.) 2011. Knowing How: Essays on Knowledge, Mind, and Action. New York: Oxford University Press. Bluck, R.S. (ed. and trans.) 1961. Plato’s Meno. Cambridge: Cambridge University Press. Foley, R. 1996. ‘Knowledge is accurate and comprehensive enough true belief.’ In Warrant in Contemporary Epistemology: Essays in Honor of Plantinga’s Theory of Knowledge, (ed.) J.L. Kvanvig. Lanham, MD: Rowman & Littlefield, pp. 87–95.

The Epistemic Basing Relation

323

Foley, R. 2012. When Is True Belief Knowledge? Princeton: Princeton University Press. Gettier, E.L. 1963. ‘Is justified true belief knowledge?’ Analysis 23: 121–123. Grube, G.M.A. (trans.) 1981. Plato: Five Dialogues. Indianapolis: Hackett Publishing. Hetherington, S. 1996. Knowledge Puzzles: An Introduction to Epistemology. Boulder, CO: Westview Press. Hetherington, S. 2001. Good Knowledge, Bad Knowledge: On Two Dogmas of Epistemology. Oxford: Clarendon Press. Hetherington, S. 2011a. How To Know: A Practicalist Conception of Knowledge. Malden, MA: Wiley-Blackwell. Hetherington, S. 2011b. ‘Knowledge and knowing: Ability and manifestation.’ In Conceptions of Knowledge, (ed.) S. Tolksdorf. Berlin: De Gruyter, pp. 73–100. Hetherington, S. 2013. ‘Skeptical challenges and knowing actions.’ Philosophical Issues 23: 18–39. Hetherington, S. 2016. Knowledge and the Gettier Problem. Cambridge: Cambridge University Press. Hetherington, S. 2017a. ‘Knowledge and knowledge-claims: Austin and beyond.’ In Interpreting Austin: Critical Essays, (ed.) S.L. Tsohatzidis. Cambridge: Cambridge University Press, pp. 206–222. Hetherington, S. 2017b. ‘Knowledge as potential for action.’ European Journal of Pragmatism and American Philosophy 9. http://journals.openedition.org/ ejpap/1070. Hetherington, S. 2018a. ‘Knowing as simply being correct.’ In A Dialogue between Law and Philosophy: Proceedings of the International Conference on Facts and Evidence, (eds.) B. Zhang and S. Tong. Beijing: Chinese University of Political Science and Law Press, pp. 68–82. Hetherington, S. 2018b. ‘The redundancy problem: From knowledge-infallibilism to knowledge-minimalism.’ Synthese 195: 4683–4702. Hetherington, S. 2019. ‘Knowing-to.’ Manuscript. Hetherington, S. 2020. ‘Knowledge-minimalism: Reinterpreting Plato’s Meno on knowledge and true belief.’ In What the Ancients Offer to Contemporary Epistemology, (eds.) S. Hetherington and N.D. Smith. New York: Routledge, pp. 25–40. Hetherington, S. and Lai, K. 2015. ‘Knowing-how and knowing-to.’ In The Philosophical Challenge from China, (ed.) B. Bruya. Cambridge, MA: The MIT Press, pp. 279–301. Kornblith, H. 1980. ‘Beyond foundationalism and the coherence theory.’ The Journal of Philosophy 77: 597–612. Leite, A. 2004. ‘On justifying and being justified.’ Philosophical Issues 14: 219–253. McGlynn, A. 2014. Knowledge First? Basingstoke: Palgrave Macmillan. Ryle, G. 1949. The Concept of Mind. London: Hutchinson. Ryle, G. 1971 (1946). ‘Knowing how and knowing that.’ In Collected Papers, Vol. 2. London: Hutchinson, pp. 212–225. Sartwell, C. 1991. ‘Knowledge is merely true belief.’ American Philosophical Quarterly 28: 157–165. Sartwell, C. 1992. ‘Why knowledge is merely true belief.’ The Journal of Philosophy 89: 167–180.

324

Stephen Hetherington

Scott, D. 2006. Plato’s Meno. Cambridge: Cambridge University Press. Stanley, J. 2011. Know How. Oxford: Oxford University Press. Stanley, J. and Williamson, T. 2001. ‘Knowing how.’ The Journal of Philosophy 98: 411–444. Stanley, J. and Williamson, T. 2017. ‘Skill.’ Noûs 51: 713–726. Williamson, T. 2000. Knowledge and Its Limits. Oxford: Clarendon Press.

Contributors

Guy Axtell is Professor of Philosophy at Radford University. He is author of Objectivity (Polity Press, 2015), and Problems of Religious Luck: Assessing the Limits of Reasonable Religious Disagreement (Lexington, 2019). His published papers engage with philosophy of science, intellectual virtues, dual-process theory, cognitive science of religion, comparative fundamentalism, and the ethics of belief. Patrick Bondy is Assistant Professor in the Department of Philosophy at Wichita State University. His research focus is primarily in epistemology. His publications include Epistemic Rationality and Epistemic Normativity (Routledge, 2018), as well as articles on topics related to knowledge, justification, inference and argument, and the basing relation. J. Adam Carter is a lecturer in philosophy at the University of Glasgow and director of Glasgow’s COGITO Epistemology group. He works on a range of topics in epistemology, including social epistemology, extended and collective epistemology, virtue epistemology, know-how, relativism, and disagreement. Stephen Hetherington is Professor of Philosophy at the University of New South Wales, in Sydney, Australia. He has written several books on epistemology, including Knowledge and the Gettier Problem (Cambridge University Press, 2016/2018) and What Is Epistemology? (Polity, 2019). He has edited several books, including Epistemology Futures (Oxford University Press, 2006) and The Gettier Problem (Cambridge University Press, 2019), has general-edited the four-volume The Philosophy of Knowledge: A History (Bloomsbury, 2019), and is currently the Editor-in-Chief of the Australasian Journal of Philosophy. Jesper Kallestrup is Professor of Philosophy at the University of Copenhagen. He works on a number of different topics at the intersection of epistemology, metaphysics and philosophy of mind, especially revolving around the metaphysics of knowledge.

326

Contributors

Keith Allen Korcz is Assistant Professor of Philosophy at the University of Louisiana at Lafayette. His publications include articles on the epistemic basing relation, applied ethics, and teaching philosophy. Errol Lord is Associate Professor in the philosophy department at the University of Pennsylvania. His main research interests are in ethical theory, epistemology, aesthetics, and the philosophy of action. Kevin McCain is Associate Professor of Philosophy at the University of Alabama at Birmingham. His research is focused on issues in epistemology and philosophy of science—particularly areas where the two intersect. He is the author of Evidentialism and Epistemic Justification (Routledge, 2014), The Nature of Scientific Knowledge: An Explanatory Approach (Springer, 2016), and Uncertainty: How It Makes Science Advance (OUP, 2019—with Kostas Kampourakis). Miriam Schleifer McCormick is Associate Professor in the philosophy department at the University of Richmond. Her primary research interests focus on the nature and norms of belief. Andrew Moon is Assistant Professor of Philosophy at Virginia Commonwealth University. His area of specialty is epistemology, and he has interests in philosophy of mind and philosophy of religion. He has recently written on the natures of belief and confidence, religious epistemology, memory, the internalism/externalism debate, and evolutionary debunking arguments. Luca Moretti is Reader of Philosophy at the University of Aberdeen and Visiting Professor at the Munich Center of Mathematics. His research is mainly in epistemology (epistemology of perception, formal epistemology, social epistemology and epistemology of education). Ram Neta is Professor of Philosophy at the University of North Carolina, Chapel Hill. His research is aimed at understanding the roles of evidence, knowledge, justification, and reasoning in the life of an agent. Tommaso Piazza is Associate Professor at the Philosophy Section of the Department of Humanities of the University of Pavia, Italy. His research is mainly in philosophy of language (truth and meaning) and epistemology (evidentialism, epistemic justification, epistemic transmission, and epistemic defeaters). Duncan Pritchard is UC Distinguished Professor of Philosophy at the University of California, Irvine, and a Professor of Philosophy at the University of Edinburgh. His monographs include Epistemic Luck (Oxford University Press, 2005), The Nature and Value of Knowledge (co-authored, Oxford University Press, 2010), Epistemological Disjunctivism (Oxford University Press, 2012), and Epistemic Angst (Princeton University Press, 2015).

Contributors

327

Mona Simion is Lecturer in Philosophy at the University of Glasgow, the COGITO Epistemology Group. Her research is in epistemology (epistemic norms, social epistemology, knowledge first epistemology), philosophy of language (assertion, conceptual engineering, contextualism), ethics (blame, trust, distributive justice, media ethics), and feminist philosophy (epistemic injustice, gender concepts). Kurt Sylvan is a lecturer in philosophy at the University of Southampton. His research interests are in epistemology as well as ethics and practical reason. John Turri is Canada Research Chair in Philosophy and Cognitive Science. He directs the Philosophical Science Lab and is author of Knowledge and the Norm of Assertion: An Essay in Philosophical Science (Open Book Publishers, 2016). Hamid Vahid is Professor of Philosophy and the Head of the Analytic Philosophy Faculty at the Institute for Research in Fundamental Sciences in Iran. His research interests are in epistemology, philosophy of mind, and philosophical logic. Ru Ye is a research associate in Philosophy at Wuhan University. Her research is mainly in epistemology (disagreement and relativism, permissivism, higher-order evidence, pragmatic encroachment, formal epistemology).

Index

ability 3, 61, 166, 181, 266, 281, 317 agency 7, 190–193, 198–199 Alston, W. 39, 227 Armstrong, D. 142, 260 Arpaly, N. 8, 229–230 Audi, R. 125, 142 Axtell, G. 9 Ballantyne, N. 276–278 Barrett, J. 279 basing 4–9, 15–16, 19, 22–30, 34–5, 39–44, 47, 53–59, 65–68, 94, 96–98, 103, 107, 116, 119–120, 122, 125–130, 134–137, 143, 153, 162, 164–165, 182–187, 192, 197–198, 201–202, 207–209, 216–222, 228, 230, 238–241, 244, 251–256, 264, 266–267, 276, 287, 305, 308–316; see also cause; justification, propositional and doxastic; well-foundedness Bergmann, M. 37 Block, N. 256 Blouw, P. 107 Boghossian, P. 28, 83–84, 130–131, 196, 219–221, 232–233 Bondy, P. 6, 277, 280 BonJour, L. 38 Bortolotti, L. 291 Boyle, M. 41 Bradley, R. 113–114 Broome, J. 163 Broome, M. 291 Buckwalter, W. 93, 105 Carnap, R. 62–63 Carroll, L. 21, 115, 133–134, 150 Carter, J. A. 6, 275, 278–279 Cassam, Q. 270n46

cause 3–8, 15–21, 23, 27–28, 30, 39, 41–42, 44–45, 53–57, 63–67, 80–85, 87–88, 94, 97–98, 103, 106, 115, 119–120, 122, 125–129, 131, 134–137, 141, 144–145, 147, 155–158, 161, 165, 185–187, 190–194, 197, 201–202, 207–209, 215, 217, 224, 238–240, 251–257, 264, 289, 291–292, 313–314; deviant 4–6, 16–21, 30, 53–58, 65–66, 131, 145, 147–149, 150–157, 165, 217, 239–240, 252–253 Chisholm, R. 203 Christensen, D. 24–25 Church, I. 279 Clark, H. 96 Clifford, W. K. 222 Cohen, G. 276, 286 Comesaña, J. 30n1, 216, 218 Conee, E. 38, 75 conservatism, epistemic 8–9, 77, 201–209 Cottrell, J. 95 Cruz, J. 16, 75 Davidson, D. 54–55 Davis, J. 275, 279, 287–288 defeat 24–26, 58, 114, 202, 204, 242–245, 252, 276, 283, 285 DiPaolo, J. 277, 290 disjunctivism 8, 167, 235–246 disposition 5–6, 17–18, 22–27, 55–68, 80–82, 85–87, 155–163, 198–199, 219 Dretske, F. 20, 263 Ehring, D. 260 Elgin, C. 203 Evans, I. 170n20

Index evidence 5, 24–26, 34–35, 38–40, 43, 45, 47, 53, 75, 92, 115–116, 118, 135–136, 191–192, 194, 201–209, 221, 224, 280, 292, 309, 317 evidentialism 25, 36–37 explanation 2, 8, 16, 26, 151–154, 157, 191, 194–195, 197–198, 219, 237–238, 251, 255–257, 259, 262, 267 Feldman, R. 75, 204–205 Fine, K. 257 Fraser, R. 286 Fricker, M. 7, 177–179, 182 Friedman, O. 105 Frise, M. 207 Fumerton, R. 201 Funkhouser, E. 260 Gettier, E. 307 Gigerenzer, G. 93 Gillett, C. 260 Goldman, A. 9, 53, 115, 136, 252 Grice, P. 144, 167 grounding 8, 74, 251, 255–264, 266–267, 318–320 Haack, S. 279 Harman, G. 125 Heider, F. 95, 104 Hetherington, S. 9 Hirstein, W. 290 Horgan, T. 256 Hume, D. 226 inference 6, 15, 37–38, 40, 56, 74, 77–89, 113–120, 126, 129–136, 150–151, 196, 219–221, 252 injustice, epistemic 177–187 James, W. 222 Jenkins, C. 257 Johnson, W. E. 260 justification: internalism and externalism 40, 44–47, 74–75, 81, 83, 115, 150, 285, 308; practical 8, 216, 220, 222, 225–230; propositional and doxastic 1–2, 15, 22–23, 25, 36–37, 40, 45, 53, 59–68, 75–79, 86–89, 116–120, 125, 127, 134–137, 153, 164–166, 182–185, 217, 230, 252–253, 255, 279, 309, 311, 316–318; see also reasons; well-foundedness

329

Kallestrup, J. 8 Kelly, T. 8, 56, 216, 221, 225 Keynes, J. M. 157 Kidd, I. 277, 291 Kim, J. 256 knowledge 8–9, 36, 106, 122–123, 126, 129, 133, 136, 142, 186, 191, 195, 230, 238–239, 241, 246, 251, 255, 262–267, 280, 282–286, 305–320 Korcz, K. 3, 6, 17, 217, 239 Kosko, B. 93 Koslicki, K. 258 Kripke, S. 62 Kvanvig, J. 75, 134–135 Lackey, J. 178 Leary, S. 217, 221, 224 Lehrer, K. 2–3, 6, 125–126, 129, 240 Lewis, D. 63–64 Lord, E. 6 luck 151, 266–267, 280–286 Lycan, W. 203, 267 Lyons, J. 43 McCain, K. 8, 10, 43, 81 McCormick, M. 8, 221 McDowell, J. 104 McGrath, M. 43, 45 meaning 28, 72, 90, 122, 145, 178–179, 227–228 Medina, J. 188n1 metaphysics 72, 171–172, 198, 251, 256, 258–259, 261–262, 267–268, 272–273 Mill, J.S. 277–278 Millar, A. 153 Montaigne, M. 280, 286–287 Moon, A. 5, 39, 188 Moretti, L. 6, 78 Moser, P. 29, 185, 252–253 Nakagawa, S. 97 Neta, R. 7, 131–132, 218–219, 221, 225–227 normativity 25, 156, 172, 189, 225, 232, 233, 279, 284–285, 289, 292, 295, 297–299, 303 norms see normativity Pelling, C. 153, 297n29 perception 53, 58, 80, 90, 95, 104, 108–109, 116, 113, 143, 145, 146–147, 166–168, 172, 183, 213, 235, 242, 249, 265, 281–282, 293

330

Index

Piller, D. 216 Plantinga, A. 4, 16, 42, 239, 252 Plato 9, 51, 91, 114, 123, 140, 213, 248, 274, 305, 320 platonic see Plato Pollock, J.L. 16, 54, 56–57, 75 Prior, A. 260 Pritchard, D. 8, 277, 280, 284 Pronin, E. 291 Pryor, J. 68, 78 rationality 225, 281, 294, 297, 300, 302 Rawls, J. 275, 287, 292 realism 272, 291, 300 reasons: epistemic 6, 9, 32, 52, 71, 74, 75, 94, 109, 117, 125, 150, 166, 168, 172, 209, 224, 231–234, 236–238, 244–246, 273, 295; explanatory 2, 141, 167, 216; motivating 137, 141–142, 144, 151–152, 154, 158–159, 166–168, 216, 221–222, 224; normative 7, 131, 133, 142, 144, 145, 146–165, 168–172, 222; practical 8, 30, 32, 215, 217, 210, 221–227, 229–234 regress 21–22, 27–28, 30, 115, 117, 113, 206, 217 Reisner, A. 221 reliabilism 29, 30, 32, 71–72, 139, 172–213, 181, 281 reliability see reliabilism Rinard, S. 221–223, 279 Rosen, G. 258 Ryle, G. 82, 133–134, 315 safe see safety safety 139, 263, 271, 280, 281, 283, 285, 289, 294, 297, 303 Sartwell, C. 320 Schaffer, J. 257–259 Schroeder, M. 224 Schtulman, A. 93 Searle, J. 260 Sellars, W. 92–93 Senor, T. 43 sensitivity 149, 152–154, 156–157, 159, 165, 170, 263, 271, 283, 288, 292, 294, 297–298, 304 Setiya, K. 7, 190–195, 198–199 Sher, G. 276, 286 Siegel, S. 219–220, 229 Silins, N. 77 Singh, K. 221 Simion, M. 7 Simpson, R. 277, 290

Smithies, D. 25 Socrates 305–306, 307–312, 314–316, 318–321 Socratic see Socrates Sosa, E. 162, 266, 281 Srinivasan, A. 293n7 Stanley, J. 266, 312 Starmans, C. 105 Sullivan-Bissett, E. 221 Swain, M. 143 Sylvan, K. 6, 47, 218 Titelbaum, M. 24 Toulmin, S. 2 truth 24, 31, 37–38, 50–51, 66, 71, 84, 105, 109, 113, 117, 122, 147, 150, 155, 158, 165, 170, 171, 183–184, 195, 202, 204, 206, 215, 224–225, 226, 227–229, 232, 238, 248, 265, 268–269, 273, 280, 283–285, 288, 292–293, 296, 298–300, 320 truthmaker 66, 71, 165 Turri, J. 6, 16–8, 22–23, 47, 57–58, 60–61, 64–65, 67, 93–95, 99, 107, 116–124, 153, 164, 253, 266 Vahid, H. 5–6 value 17, 57–58, 102, 105, 112, 132, 215, 225, 227, 229–230, 323, 234, 246, 249, 269, 276, 282, 295, 303, 306, 310, 311, 314, 316, 320 van Wietmarschen, H. 25 virtue epistemology 71, 162, 146, 249, 303, 325 vision see visual experience visual experience 18, 38, 51, 78, 88, 145, 165, 167, 251, 263, 266 voluntarism, doxastic 167, 322 Wedgwood, R. 55, 57, 67 Weisberg, J. 75 well-founded belief see wellfoundedness well-foundedness 6, 9–10, 128, 136, 256, 267, 275, 277–281, 283, 285, 286, 288 White, R. 77 Williams, B. 199, 225 Williamson, T. 28, 266, 312 Wilson, J. 258, 261 Winer, G. 95 Wright, C. 77 Yablo, S. 260–271 Ye, R. 5