The Ethics of Belief [Hardcover ed.] 0199686521, 9780199686520

Citation preview

The Ethics of Belief

The Ethics of Belief Individual and Social

edited by

Jonathan Matheson and Rico Vitz

1

1 Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © The several contributors 2014 The moral rights of the authors‌have been asserted First Edition published in 2014 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2014932718 ISBN 978–0–19–968652–0 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

For our families, in gratitude for their love and support: Lesley, Karis, Thomas, Lincoln, and Asher Rose, Dominic, and Francis

Acknowledgements The initial impetus for this book was the A. David Kline Public Philosophy Symposium on the ethics of belief, held at the University of North Florida in 2011. We would like to thank the Florida Blue Center for Ethics at the University of North Florida for providing generous funding for the event, David Kline for his work as the Founding Director of the Center, and Mitch Haney and Alissa Swota for their continued leadership of the Center, without which the Symposium would not have been possible. We would also like to thank the speakers and panelists for the event—Heather Battaly, E. J. Coffman, Mark Dechesne, Kate Elgin, Rich Feldman, Sandy Goldberg, Alvin Goldman, Heidi Grasswick, Tom Kelly, Arie Kruglanski, Ted Poston, Ernie Sosa, and Sarah Wright— as well as our colleagues in the Department of Philosophy at the University of North Florida—Andy Buchwalter, Paul Carelli, Brandi Denison, David Fenner, Erinn Gilson, Julie Ingersoll, Bert Koegler, and Sarah Mattice—for all their help in making the Symposium a success.

Contents Notes on Contributors Introduction

xi 1

Part I.  The Ethics of Belief: Individual 1. The Powers that Bind: Doxastic Voluntarism and Epistemic Obligation Neil Levy and Eric Mandelbaum

15

2. Deciding to Believe Redux  Andrei A. Buckareff

33

3. Varieties of Epistemic Vice Heather Battaly

51

4. Knowledge and Time: Kripke’s Dogmatism Paradox and the Ethics of Belief  Ernest Sosa

77

5. Can there be a Knowledge-First Ethics of Belief? Dennis Whitcomb

89

6. Truth as the Fundamental Epistemic Good Duncan Pritchard

112

7. Wide-Scope Requirements and the Ethics of Belief  Berit Brogaard

130

8. The ‘Ethics of Belief ’ is Ethics (Period): Reassigning Responsibilism Trent Dougherty

146

Part II.  The Ethics of Belief: Social 9. The Psychology of Knowledge Formation: Its Impetus, Mechanism, and Social Context  Arie W. Kruglanski and Lauren M. Boyatzi

169

10. Perspectives on Social Knowledge Mark Dechesne and Charlotte de Roon

184

11. Contagion, Community, and Virtue in Hume’s Epistemology  Rico Vitz

198

12. Understanding Epistemic Normativity in Feminist Epistemology Heidi Grasswick

216

x Contents 13. The Commonwealth of Epistemic Ends  Catherine Z. Elgin

244

14. Assertion and the Ethics of Belief Sanford Goldberg

261

15. Evidence of Evidence is Evidence Richard Feldman

284

16. Believers as Thermometers  Thomas Kelly

301

17. Disagreement: Idealized and Everyday Jonathan Matheson

315

Index

331

Notes on Contributors Heather Battaly is Professor in the Department of Philosophy at California State University—Fullerton. She is the editor of Virtue and Vice, Moral and Epistemic (Wiley-Blackwell), and Perspectives on the Philosophy of William P. Alston. (Rowman and Littlefield), as well as the author of a number of articles, including “Thin Concepts to the Rescue:  Thinning the Concepts of Epistemic Justification and Intellectual Virtue” (in Virtue Epistemology), “Epistemic Self-indulgence” (in Metaphilosophy), and “Intellectual Virtue and Knowing One’s Sexual Orientation” (in Sex and Ethics). Lauren M. Boyatzi received her PhD in Social Psychology from the University of Maryland, College Park in 2013. Currently, she is Operations Research Analyst for the Sexual Assault Prevention and Response Office, Headquarters U.S. Air Force, Washington, D.C. She conducts analysis and reporting of sexual assault statistics that meet congressionally mandated requirements; reports weekly data on sexual assault reports from the Defense Sexual Assault Incident Database; and has worked with a team to develop the Department of Defense Victim Experiences Survey. Dr. Boyatzi has authored and delivered reports and briefings for senior USAF leadership. She served as Psychological Consultant on the Influencing Violent Extremist Organizations: Planning Influence Activities While Accounting for Unintended Side Effects (I-VEO) Strategic Multilayer Assessment project in collaboration with USSTRATCOM and the National Consortium for the Study of Terrorism and Responses to Terrorism (START) Center at the University of Maryland in 2011. She also served as Statistical Consultant to the University of Maryland Psychology Department from 2010 to 2013. Dr. Boyatzi is a recipient of the Department of Defense Science, Mathematics, and Research for Transformation (SMART) Scholarship-for-Service award. Berit Brogaard is Professor of Philosophy in the Department of Philosophy at the University of Missouri—Saint Louis. She is the editor of Does Perception have Content? (Oxford UP) and the author of the monograph Transient Truths (Oxford UP) as well as numerous articles, which have appeared in various journals, including the Journal of Philosophy, Noûs, and Philosophy and Phenomenological Research. Andrei A. Buckareff is Associate Professor in the Department of Philosophy at Marist College. He is the co-editor of New Waves in the Philosophy of Action (Palgrave-Macmillan), Causing Human Action: New Perspectives on the Causal Theory of Action (MIT Press), and Philosophy of Action: 5 Questions (Automatic Press), as well as the author or co-author of nearly two dozen articles on various topics in philosophy of mind, philosophy of action, and philosophy of religion that have appeared in

xii  notes on Contributors Canadian Journal of Philosophy, International Journal for Philosophy of Religion, Journal of Philosophical Research, Philosophical Studies, Religious Studies, and elsewhere. Mark Dechesne is an associate professor at the Centre for Regional Knowledge Development of the University of Leiden—Campus The Hague. He completed his dissertation on the effects of fear on social behavior in 2001. He received a Dutch innovational research incentive grant (VENI) in 2003. From 2006 to 2008, he was a visiting fellow at the National Consortium for the Study of Terrorism and Responses to Terrorism (NC-START), the DHS Center of Excellence located at the University of Maryland, College Park. His current research focuses on knowledge formation and development, human motivation, and security. Trent Dougherty is Assistant Professor of Philosophy at Baylor University.  He is editor of Evidentialism and Its Discontents (Oxford UP, 2011).  He is a member of the Board of Editorial Consultants for American Philosophical Quarterly.  He is the author of numerous articles in epistemology, philosophy of language, and philosophy of religion appearing in such journals as Analysis, The European Journal of Philosophy, Faith and Philosophy, Philosophy and Phenomenological Research, Religious Studies, Synthese.  He has written numerous reference articles including “Fallibilsm” for the Routledge Companion to Epistemology and “Evidence” for Oxford Bibliographies.  He has recently discovered he has an interest in applied ethics, the ethics of belief. Catherine Z. Elgin is Professor of Philosophy in the Graduate School of Education and an associate of the Edmond J. Safra Foundation Center of Ethics and the Professions at Harvard University. She is the author of Considered Judgment (Princeton, 1996), Between the Absolute and the Arbitrary (Cornell, 1997), With Reference to Reference (Hackett, 1983), and the co-author (with Nelson Goodman) of Reconceptions in Philosophy and Other Arts and Sciences (Hackett, 1988). She is also the editor of The Philosophy of Nelson Goodman (Garland, 1997), and co-editor (with Jonathan E. Adler) of Philosophical Inquiry (Hackett, 2009). She is the recipient of fellowships from the National Endowment of the Humanities, the American Council of Learned Societies, the John Dewey Foundation, the Spencer Foundation, the Andrew Mellon Foundation, and the Bunting Institute. Richard Feldman is Professor of Philosophy and Dean of the College at the University of Rochester. He is the author of Reason and Argument (Prentice-Hall) and Epistemology (Prentice Hall), the co-author (with Earl Conee) of Evidentialism (Oxford UP), the co-editor (with Ted A. Warfield) of Disagreement (Oxford UP), as well as the author of dozens of articles, which have appeared in journals such as The Philosophical Review, Philosophy and Phenomenological Research, Philosophical Studies, Australasian Journal of Philosophy, The Monist, Analysis, American Philosophical Quarterly, Philosophical Perspectives, Philosophical Issues, and Faith and Philosophy.

NOTES ON CONTRIBUTORS  xiii Sanford Goldberg is Professor in and Chair of the Department of Philosophy at Northwestern University. He is the author of Anti-Individualism: Mind and Language, Knowledge and Justification (Cambridge UP) and Relying on Others:  An Essay in Epistemology (Oxford UP), the co-author (with Andrew Pessin) of Gray Matters: An Introduction to the Philosophy of Mind (M.E. Sharpe), the editor of Internalism and Externalism in Semantics and Epistemology (Oxford UP), the co-editor (with Andrew Pessin) of The Twin Earth Chronicles (M.E. Sharpe), and the author of numerous articles, which have appeared in journals such as Noûs, Philosophy and Phenomenological Research, Philosophical Quarterly, Analysis, Synthese, Pacific Philosophical Quarterly, Philosophical Studies, Grazer Philosophische Studien, and American Philosophical Quarterly. Heidi Grasswick is the George Nye and Anne Walker Boardman Professor of Mental and Moral Science in the Department of Philosophy at Middlebury College. She is the editor of Feminist Epistemology and Philosophy of Science: Power in Knowledge (Springer), the co-editor (with Mark Owen Webb) of Feminist Epistemology as Social Epistemology (a special issue of Social Epistemology), and the author of a number of articles, including “Individuals-in-Communities: The Search for a Feminist Model of Epistemic Subjects” (Hypatia), “The Normative Failure of Fuller’s Social Epistemology” (Social Epistemology), and “Feminist Social Epistemology” (The Stanford Encyclopedia of Philosophy). Thomas Kelly is Professor in the Department of Philosophy at Princeton University. He is the author of “Disagreement in Philosophy” (in The Oxford Handbook of Philosophical Methodology), “The Epistemic Significance of Disagreement” (in Oxford Studies in Epistemology), and “Disagreement and the Burden of Judgment” (in The Epistemology of Disagreement: New Essays), and more than a dozen other articles, which have appeared in journals such as The Journal of Philosophy, Philosophy and Phenomenological Research, Philosophical Studies, Midwest Studies in Philosophy, Philosophical Perspectives, Philosophy of Science, and Philosophy Compass. Arie W. Kruglanski is Distinguished University Professor at the University of Maryland. He is recipient of the National Institute of Mental Health Research Scientist Award (Career Award); the Senior Lifetime Achievement Award from the von Humboldt Foundation; the Donald Campbell Award for Outstanding Contributions to Social Psychology from the Society for Personality and Social Psychology (SPSP); the Distinguished Scientific Contribution Award from the Society of Experimental Social Psychology (SESP); the Award for Scholarship and Creativity from the Regents of the University of Maryland; and the Revesz Award from the University of Amsterdam. He has published over 200 articles, chapters, and books, and he is continuously supported by grants from the National Science Foundation, the National Institute of Mental Health, Deutsche Forschungs Gemeineschaft, the Ford Foundation and the Israeli

xiv  notes on Contributors Academy of Science. He has recently served as member of the National Academy of Sciences (NAS) panels on counterterrorism, and educational paradigms in homeland security, and served on the NAS Board for Cognitive, Social and Sensory Processes. Kruglanski also serves as a co-principal investigator at the National Center for the Study of Terrorism and the Response to Terrorism at the University of Maryland. Neil Levy is Deputy Director of the Oxford Centre for Neuroethics, and Head of Neuroethics at the Florey Institute of Neuroscience and Mental Health, University of Melbourne. He is the author of Consciousness and Moral Responsibility (Oxford UP), Hard Luck: How Luck Undermines Free Will and Moral Responsibility (Oxford UP), Neuroethics: Challenges for the 21st Century (Cambridge UP), What Makes Us Moral? Crossing the Boundaries of Biology (Oneworld), and numerous other articles, which have appeared in journals, such as Philosophy and Phenomenological Research, The Journal of Philosophy, and American Journal of Bioethics. Eric Mandelbaum is Assistant Professor at Baruch College, City University of New York. He is the author or co-author of various articles, including “Thinking is Believing” (Inquiry), Against Alief ” (Philosophical Studies), “Explaining the Abstract/ Concrete Paradoxes in Moral Psychology: The NBAR Hypothesis” (The Review of Philosophy and Psychology) and “Responsibility and the Brain Sciences” (Ethical Theory and Moral Practice). Jonathan Matheson is Assistant Professor in the Department of Philosophy at the University of North Florida. He is the author of “The Case for Rational Uniqueness” (Logos & Episteme), “Epistemological Considerations Concerning Skeptical Theism: A Response to Dougherty” (Faith and Philosophy), “Conciliatory Views of Disagree­ ment and Higher-Order Evidence” (Episteme:  A  Journal of Social Philosophy), and “Bergmann’s Dilemma:  Exit Strategies for Internalists” (Philosophical Studies)— co-authored with Jason Rogers. Duncan Pritchard is Professor of Philosophy and Chair in Epistemology at the University of Edinburgh. He has published extensively in epistemology, including Epistemic Luck (Oxford UP), The Nature and Value of Knowledge (co-authored, Oxford UP), and Epistemological Disjunctivism (Oxford UP). Charlotte de Roon is a research associate at the Centre for Regional Knowledge Development of Leiden University—Campus The Hague. She also participates as a Ph.D. candidate in the dual Ph.D. program of Leiden University. Her research focuses on organized youth participation in politics. By researching party youth wings in the political system of Western democracies she wants to clarify the relationship of these organizations with young adults in society and with their mother party.

NOTES ON CONTRIBUTORS  xv Ernest Sosa is Board of Governors Professor of Philosophy at Rutgers University. He is the author of Knowing Full Well (Princeton UP), A Virtue Epistemology: Apt Belief and Reflective Knowledge, Volume I (Oxford UP), Reflective Knowledge: Apt Belief and Reflective Knowledge, Volume II (Oxford UP), Knowledge in Perspective: Selected Essays in Epistemology (Cambridge UP), and many articles, as well as the editor or co-editor of a number of books and journals, such as Philosophy and Phenomenological Research and Noûs. Rico Vitz is Associate Professor in the Department of Philosophy at Azusa Pacific University. He is the author of Reforming the Art of Living: Nature, Virtue, and Religion in Descartes’s Epistemology (Springer, 2014) and various articles, including “The Nature and Functions of Sympathy in Hume’s Philosophy” (The Oxford Handbook of David Hume, forthcoming), “Thomas More and the Christian ‘Superstition’: A Puzzle for Hume’s Psychology of Religious Belief ” (The Modern Schoolman), “Descartes and the Question of Direct Doxastic Voluntarism” (Journal of Philosophical Research), and “Doxastic Virtues in Hume’s Epistemology” (Hume Studies). Dennis Whitcomb is Associate Professor of Philosophy at Western Washington University. He is the co-editor (with Alvin Goldman) of Social Epistemology: Essential Readings (Oxford UP), and the author of nearly a dozen articles, including “Grounding and Omniscience” (in Oxford Studies in Philosophy of Religion), “Epistemic Value” (in The Continuum Companion to Epistemology), and “Williamson on Justification” (Philosophical Studies).

Introduction Suppose someone were to ask you, “What is the ethics of belief?” If you had nothing more than a passing familiarity with the subject, there is a good chance that you would do three things. First, you would formulate your answer with reference to two widely anthologized essays, W. K. Clifford’s “The Ethics of Belief ” (1999) and William James’s “The Will to Believe” (1956). Second, you would tell a classic story from each, recounting the tales of Clifford’s shipowner and of James’s man stranded on a mountain pass. Third, you would refer to a notably contested aspect of each essay, like Clifford’s claim that “[i]‌t is wrong always, everywhere and for anyone, to believe anything upon insufficient evidence,” and James’s suggestion that a person can choose to believe certain propositions if the stakes are sufficiently high. Such an answer is fine, as far as it goes, but the ethics of belief is a topic that is both historically deeper and conceptually richer than this kind of answer suggests. Let us briefly describe these two points and the significance of each before providing an overview of the contents of this volume.

I.1. Historical Depth Philosophers were doing interesting work at the nexus between ethics and epistemology (and psychology) long before Clifford wrote his seminal article in the nineteenth century. Consider just two noteworthy examples from the Modern period. In the eighteenth century, David Hume argued that a “wise man . . . proportions his belief to the evidence” (Hume 1975, 110; 1999, 170). Throughout his seminal works, he claimed that people who fail to reason properly frequently instantiate vices like “supine indolence of the mind,” “rash arrogance,” “lofty pretensions,” and “superstitious credulity” (Hume 1975, 41; 1999, 119). Moreover, he analyzed the process of belief formation not only with respect to ethics and epistemology but also with respect to psychology. For instance, he analyzed (what he describes as) the “contagious” nature of beliefs and prescribed ways to regulate the acquisition of such beliefs virtuously.1 Similarly, in the seventeenth century, René Descartes characterized errors in the process of belief formation not merely as mistakes in reasoning but as moral errors   Vitz will discuss this particular aspect of Hume’s ethics of belief in Chapter 11.

1

2 Introduction (see, e.g., Descartes 1991 [vol. 3], 233; 1901 [vol. 4], 115) and as sins (see, e.g., Descartes 1985 [vol. 1], 55; 1908 [vol. 10], 436). Moreover, in the Passions of the Soul, he analyzed the nature of passions such as anger, identified some of the problems such passions caused in reasoning, and prescribed certain means for virtuously regulating them so that they would not adversely affect the process of belief formation (see, e.g., Descartes 1985 [vol. 1], 399–401; 1908 [vol. 10], 477–81). These are just two examples from the Modern period. If we wanted to do so, we could go on to offer examples from the Medieval and the Ancient periods, as well. For example, we could explain in detail the ways in which Descartes’s views on religious belief formation echoed that of his scholastic predecessors, like Aquinas (see, e.g., Aquinas 2006 [vol. 23], 137ff; 2006 [vol. 32], 38ff).2 Similarly, we could elucidate the ways in which the Scholastics, like Aquinas, were developing themes they found in Aristotle’s Nicomachean Ethics (1941) or in various works of Plato, like Theaetetus, Meno, or the Republic (1997). For our purposes, however, we merely want to highlight the fact that the ethics of belief is a topic with a history that goes well beyond the century in which Clifford composed his seminal essay. As some of the essays in this volume will make clear, this (all-too-frequently neglected) history provides ample resources that can enhance our contemporary discussions concerning the ethics of belief.

I.2.  Conceptual Richness To highlight the conceptual richness of the ethics of belief, let us start by considering our normal practice of praising and blaming others. If you are like (at least one of) us, then you like to censure other people. Not only do you like to censure the behaviors of other people, in general, you like to censure their beliefs as well—whether the formation and maintenance of beliefs are to be considered a kinds of behavior or not. Additionally, you may like to congratulate people for believing as they should. Either way, each of the following seems true, at least for a normal American adult living in 2013: • You should believe that Barack Obama is the president of the U.S. • You should not believe that President Obama was born in Kenya. • You should believe that the Earth is roughly spherical. • You should not believe that the Sun revolves around the Earth. • You should believe that the Chicago Cubs are a baseball team. • You should not believe that the Cubs will win the World Series in 2013. We could go on, but it seems clear that there is a class of propositions that you should believe and a class of propositions that you should not believe. No doubt, there is also a class of propositions belonging to neither category—i.e., a class of propositions such

2

  Summa Theologica I–II, Q. 62; II–II.

Introduction  3 that it is not the case that you should believe them and it is not the case that you should not believe them (e.g., the number of geese currently in Canada is odd, a Democrat will be the President of the U.S. in 2024, etc.). The boundaries between these categories are bound to be vague, but it is clear that there are such categories. But in virtue of what are propositions divided into these categories? Central to the ethics of belief is just this question: “What should I believe?” It might be thought that there is just one type of answer to this question—that the “should” in question is an all-things-considered “should.” Call this position the “Unified Question Thesis.” In contrast to the Unified Question Thesis, it might be thought that our question is multiply ambiguous, that as it stands the question is ill-formed, and that there are only answers to the disambiguated versions of it, perhaps with some of the disambiguated questions being more central than others. Call this position the “Ambiguous Question Thesis.” Among the disambiguated questions are the following: • From the moral point of view, what should I believe? • From the prudential point of view, what should I believe? • From the epistemic point of view, what should I believe? Regardless of which position one takes on whether the question “what should I believe?” is ambiguous, it is clear that there are these more fine-grained questions. In fact, some think that the central question just is one of these more fine-grained questions. In particular, it has been maintained that the ethics of belief just is epistemology, that our central question is the epistemic question. But how should we understand the epistemic question? We can better understand the question “from the epistemic point of view, what should I believe?” by getting a better understanding of what the epistemic perspective is. In light of the classic essays of Clifford and of James, there have been thought to be two goals that one has from the epistemic perspective: to believe truths and to disbelieve falsehoods. Each of these epistemic projects holds the other in check—without the other, each project would lead to an unhealthy epistemic life. So understood, the epistemic question can roughly be paraphrased as the question “with the interest of believing truths and disbelieving falsehoods, what should I believe?” Now consider the following case: Doris is very ill and the prognosis is not good. Her doctor tells Doris this information. The doctor also tells Doris that the chances of her making a recovery will improve if she believes that she will get better.

Should Doris believe that she will get better? It seems that in at least one sense, the prudential sense, that she should believe this. On the other hand it seems that in another sense, the epistemic sense, that she should not believe this. Cases like this might incline you to believe the Ambiguous Question Thesis—that there is no one comprehensive answer to what Doris should believe—or that there is one solitary fine-grained question which is relevant. According to the Unified Question Thesis, there is just one answer to what Doris should believe, and the prudential and epistemic considerations

4 Introduction both factor into that answer, though it may be less clear what that final verdict is. (It is at least less clear than the answers to the epistemic and prudential versions of the question.) The question of how we should understand this central question lies at the heart of the ethics of belief. An answer to the question “what should I believe?” invites the follow up “why should I believe that?” Having seen that there is a class of propositions that one should believe, a class of propositions that one shouldn’t believe, and a class of propositions such that it is not the case that one should believe them and it is not the case that one should not believe them one should wonder why these propositions are classified the way that they are. That is, in virtue of what is it the case that a normal American adult living in 2014 should believe that Obama is the President of the U.S. in 2014 and not believe that he was born in Kenya? Answers to this question will depend upon how one answers the question “what should I believe?” If the Unified Question Thesis is correct, then moral, prudential, and epistemic factors will explain why one should believe/disbelieve a certain proposition. In contrast, if the Ambiguous Question Thesis is correct, then only one type of consideration will factor in to the answer to each disambiguated question. This volume explores both of these questions, bringing together contemporary work which extends this traditional epistemological debate in novel ways. Traditionally, the ethics of belief, like epistemology in general, has been focused on the individual, and the individual’s perspective in particular. Perhaps we can call this the “Cartesian Model” of epistemology. Such a focus is seen in W. K. Clifford’s famous story of a shipowner in his seminal essay on the ethics of belief. Clifford describes a shipowner who knew his ship was old and not well built, and who had his doubts as to whether his ship would make the next voyage. Nevertheless, the shipowner was able to overcome these thoughts and come to believe that the ship would safely make the voyage. The ship sinks, and Clifford finds the shipowner guilty of having a belief he had no right to have—a verdict that would have held even if the ship had successfully completed its voyage. The first half of this volume focuses on the ethics of belief from this traditional Cartesian Model. Part I begins by examining the issue of doxastic control. It is natural to wonder whether there is an answer to the question “what should I believe?” unless I have some sort of control over my beliefs. It is a common adage that “ought implies can,” yet it seems that at least many of our beliefs are not under our control. For this reason, issues surrounding doxastic voluntarism have been tightly connected with the ethics of belief. This part then shifts its focus to what believers should be like and what their epistemic ends are or should be. Motivated by the “value turn” in epistemology, this next cluster of essays examines intellectual virtues and how they are related to traditional questions in epistemology including what believers should be like. The section continues the discussion by examining questions having to do with epistemic value—i.e., that which is valuable from the uniquely epistemic point of view. Several essays in this part consider problems in epistemic value and the relation

Introduction  5 between what is epistemically good and what is epistemically right. This part concludes by examining the implications for the ethics of belief of another recent trend in epistemology—the claim that knowledge is in some sense primary, and with an assessment of whether the ethics of belief falls within the domain of epistemology or ethics. Part II extends work on the ethics of belief in accordance with another recent movement in epistemology, focusing on the social dimensions of knowledge and justified beliefs. While such an emphasis is a recent development in epistemology, social epistemologists trace their project back to Plato. Following Alvin Goldman (2011), we can divide social epistemology into three separate but related enterprises: ( i) the examination of whether and how groups have beliefs or knowledge, (ii) the examination of how the individual as situated within his/her social context should go about forming beliefs, and (iii) the examination of how various social institutions affect the beliefs of those involved with them, and how they might be structured to better suit the epistemic ends of those individuals. Society plays a critical role in the formation of knowledge and justified beliefs, and social epistemology examines and evaluates these societal effects. Following this extension of epistemology from the individual to the social level, this volume extends the debate regarding the ethics of belief to the social level as well, in particular looking at how beliefs should be transmitted and integrated. This extension of the debate regarding the ethics of belief proves to be both novel and fruitful. Following this extension of epistemology from the individual to the social level, this part broadens the focus of the traditional ethics of belief by examining issues regarding the larger context of believers. Essays in this part place questions of the ethics of belief into a larger social context examining social epistemic practices like assertion as well as how social roles such as gender relations have epistemic effects. In particular, this part applies the ethics of belief to one particular type of situation currently of great interest to epistemologists—peer disagreement. Essays on this topic examine the features of disagreement and how they are relevant to what one should believe in the face of disagreement. This part also includes essays by social psychologists related to inquiry and the formation of knowledge. So, there are numerous questions related to the question “what should I believe?” These questions encompass more than has traditionally thought to have been relevant including questions pertaining to the social context of believers. Questions relevant to the ethics of belief thus include the following: • Do I have control over what I believe? • What intellectual characteristics make for better believers? • What should the epistemic ends of believers be? • How do various social institutions affect what I believe?

6 Introduction • How should various social institutions be structured to help produce better beliefs? • What should I believe when I am aware of widespread disagreement about the issue at hand? This volume focuses on questions such as these, bringing together contemporary essays from diverse perspectives. In so doing, it attempts to provide conceptual resources that aim to enhance contemporary discussions about the ethics of belief.

I.3.  Overview of Chapters Against this general background, let us turn to an overview of the individual chapters. In Chapter 1, Levy and Mandelbaum attempt to establish three theses. First, in response to recent work by Frankish, they argue that direct doxastic voluntarism (DDV) is false—i.e., that people lack the power to form beliefs at will directly. Second, drawing on recent studies in social psychology, they contend that people have a propensity to form beliefs for non-epistemic reasons. Third, they argue that although DDV is false, once people become aware of their propensities to form beliefs for non-epistemic reasons, they have obligations to avoid triggering them—which are similar in kind to the sorts of obligations people would have if DDV were true. In Chapter 2, Buckareff develops and defends an argument against DDV—i.e., the thesis that agents can consciously exercise the same sort of direct voluntary control over coming to acquire a doxastic attitude, such as belief, suspension of belief, or disbelief, that they exercise over uncontroversial basic actions, like raising one’s arm. More specifically, he argues that it is impossible for a person to succeed in exercising direct voluntary control over coming to believe that a proposition is true, on the basis of practical reasons alone. His case for this thesis proceeds in three steps. He begins by reviewing Bernard Williams’s seminal argument for a thesis similar to the one for which he will contend, and he elucidates the reasons that Williams’s argument is regarded as a failure. He continues by identifying a few constraints on belief formation. In the final section before the conclusion, he presents the case for his own view and notes the ways in which it avoids the kind of objections levied against the position taken by Williams. In Chapter 3, Battaly offers an analysis of often-neglected conceptions of vice. Her argument proceeds in three stages. In the opening section, she introduces two familiar conceptions of virtue and their corresponding conceptions vice. According to the first, virtues produce good effects, and vices produce bad effects. According to the second, virtues require praiseworthy psychologies, and vices require blameworthy psychologies. In the next section, she argues that the kind of vices on which virtue-reliabilists principally focus are vices that produce bad epistemic effects, and that the kind of vices on which virtue-responsibilists principally focus are vices that require blameworthy psychologies, understood in an Aristotelian vein. In the final section of the chapter, she argues that although virtue-responsibilists can—and usually do—rely on an

Introduction  7 Aristotelian conception of vice, they need not do so. She concludes by developing two alternative conceptions of vice that warrant further consideration. In Chapter 4, Sosa considers the question “when is it permissible to end inquiry?” One might think that the answer to this question is that one can permissibly end inquiry regarding p when one knows p. After all, if S knows p, then any additional evidence S might get regarding p would seem to be either misleading or superfluous. In this essay, Sosa argues that epistemic negligence resulting from closing off inquiry can detract from one’s epistemic performance and possibly result in the loss of knowledge. In addition, he argues that closing off inquiry can also result in the loss of rational belief since the origins of one’s belief are continually fading from view, which requires that one’s current evidence play a primary role in rational belief. In Chapter 5, Whitcomb examines the prospects of a knowledge-first account of the ethics of belief. Knowledge-first accounts claim that the concept of knowledge plays a central role in determining what we should believe from an epistemic perspective. Whitcomb finds problems for numerous formulations of a knowledge-first ethics of belief including identity theories, counterfactual theories, knowledge minus belief theories, and duplication theories. Each of these accounts either faces clear counterexamples or leaves important details to be filled in at a later date, indicating that the move to a knowledge-first account will not settle the difficulties in formulating a specific and counterexample-free account of the ethics of belief. In Chapter 6, Pritchard examines epistemic value—what is valuable from the epistemic point of view. Pritchard defends the thesis that truth is sole fundamental epistemic good. Recently it has become more commonplace reject this conception of the epistemic good for a variety of reasons, but Pritchard argues that this move has been too hasty. In this chapter, Pritchard addresses three of the most prominent arguments against truth being the fundamental epistemic good, and finds each unpersuasive. First, Pritchard takes on the swamping problem, which alleges that knowledge is more valuable than true belief, and that this cannot be accounted for if truth is the fundamental epistemic good. The second objection addressed is the challenge of accounting for why we shouldn’t focus our inquiry on trivial truths if truth is the fundamental epistemic good. Finally, Pritchard gives an account of why typically only once one has achieved knowledge or understanding is inquiry appropriately closed off. In Chapter 7, Brogaard examines an evidentialist ethics of belief, and W. K. Clifford’s proposal in particular. She argues that regardless of how one understands the notion of evidence, it is implausible that we could have a moral obligation to refrain from believing something whenever we lack sufficient evidence. Alternatively, Brogaard argues that there are wide-scope conditional requirements on beliefs but that these requirements can be met without having sufficient evidence for the belief in question. She then argues that we are epistemically, though not morally, required to form epistemically valuable beliefs. However, these beliefs, too, need not be beliefs for which we have sufficient evidence—epistemically good beliefs need not be based on sufficient evidence.

8 Introduction In Chapter 8, Dougherty scrutinizes the province of the ethics of belief. Having identified the ethics of belief with the ethics of inquiry, he argues that the ethics of belief is not properly epistemic. Rather, he claims that the ethics of inquiry or considerations of epistemic responsibility are fundamentally ethical or prudential considerations. According to Dougherty, the only distinctively ethical norms are evidential norms regarding propositional justification. In particular, these norms concern what evidence a subject possesses at a time, and does not include evidence that she easily could have obtained. He argues that this taxonomic point is not without significance and defends this claim from a number of objections. Emerging from his defense is a novel account of epistemic normativity, one that is teleological in nature. In Chapter  9, Kruglanski and Boyatzi present the findings of three decades of research on the psychological theory of “lay epistemics.” More specifically, they summarize each of the theory’s three principal components. The first element focuses on inferential processes and attempts to explain how knowledge is acquired. According to Kruglanski and Boyatzi, people acquire knowledge by following inferential rules, of which some are hard-wired and others are socially constructed. The second element deals with social relationships and attempts to elucidate the significance of people from whom knowledge is acquired. Kruglanski and Boyatzi argue that although there are differences in emphasis among certain cultures, broadly conceived (e.g., “Westerners” v. “Easterners”), people universally acquire knowledge, in accordance with inferential rules, based on the perceived reliability of epistemic authorities. The third element addresses the motivational factors that putatively explain why people attempt to gain knowledge. On this topic, Kruglanski and Boyatzi attempt to elucidate people’s frequently competing desires for truth, for cognitive consistency, and for cognitive closure. In Chapter 10, Dechesne and de Roon attempt to reconcile two competing schools of thought in social psychology. One is critical rationalism, which emphasizes the significance of attempting to validate abstract hypotheses by means of carefully crafted experiments. The other is social constructivism, or what Wilhelm Wundt called folkpsychology, which emphasizes the significance of observing people in their natural social settings, rather than in contrived experimental settings. In presenting their position, Dechesne and de Roon examine three cases. In light of their examination of the psychological study of terrorism, they suggest that the critical rationalist approach provides a helpful method for understanding the beliefs and, more importantly, predicting the behavior of terrorists and others who perpetrate mass violence. They temper their apparent support of critical rationalism, however, by considering two other cases. One is that of Robert Caldini—a prominent social psychologist who left the field due to his concern that the field’s emphasis on critical rationalism had caused it to become irrelevant to anyone outside the field. The other is that of Diederik Stapel—a prominent social psychologist who was suspended after it was revealed that he fabricated the data he used in multiple publications. Dechesne and de Roon conclude by suggesting that

Introduction  9 the lay epistemic theory of Kruglanski et al. seems to provide a way to reconcile the tension between critical rationalism and social constructivism. In Chapter 11, Vitz elucidates a Humean account of ways in which beliefs are, and ought to be, formed by means of social interactions. The chapter proceeds in three parts. In the first part, he explains, briefly, Hume’s account of belief. In the second, he (1) elucidates Hume’s conception of “contagion,” a process by which people naturally adopt the passions and opinions of others, (2) clarifies the way that beliefs are acquired by this process, and (3) explains why, in many cases, such “contagious beliefs” are extremely difficult, if not impossible, to revise. In the third part, he elucidates a Humean account of the conditions under which virtuous doxastic agents ought to resist “contagious beliefs.” In Chapter 12, Grasswick offers a rebuttal to a common criticism of feminist epistemology, according to which it either lacks normative content altogether or has the wrong kind of normative content for a proper account of knowledge. She develops her rebuttal by explaining a central theme in feminist approaches to epistemology: namely, that epistemic agents are socially situated. She then elucidates four features of the normative projects in which feminist epistemologists engage:  (1)  the prioritization of knowledge-seeking practices, (2)  the contextualization of epistemic norms, (3)  the entanglement of the epistemic and the ethical, and (4) the particular significance of the perspective of epistemic agents. In so doing, she argues that these common critiques of the normative aspect of feminist epistemology are, fundamentally, more general disagreements about the nature of epistemic normativity—on which feminist epistemologists side, in their own unique ways, with naturalized epistemologists, contextualists, and virtue epistemologists. In Chapter 13, Elgin develops a critique of a pair of widely accepted epistemic principles. According to the first, epistemic individualism, the states of an individual epistemic agent are that which constitute the agent’s “epistemic core.” According to the second, attunement, the core deliverances that justify an agent’s beliefs do so because they properly attune the agent to their objects. Elgin develops her critique by using Orwell’s 1984 as an epistemological thought experiment. She argues that the plight of the novel’s protagonist, Winston, reveals problems for each of the theses and that the support of a non-coercive community is necessary for having beliefs at all, let alone having the kind of beliefs that could amount to knowledge. She then (i) explains the relationship between the Orwellian thought experiment and concrete cases of epistemic injustice, and (ii) proposes a Kantian solution to the problems that the experiment raises. She concludes by arguing that epistemic agents are bound by a version of the third formulation of Kant’s Categorical Imperative, such that they ought to regard themselves as legislators in a commonwealth of epistemic ends. In Chapter 14, Goldberg examines the ethical dimension of the practice of assertion, both in making and in receiving an assertion, and its connection to the ethics of belief. He argues that on the plausible assumption that the speech act of assertion has an epistemic norm, we can account for the ethical dimension of the practice in

10 Introduction terms of the reasonable expectations that hearers have of asserters, and that asserters have of hearers. Further, Goldberg argues that the ethics of assertion informs the ethics of belief. Since assertion is one predominate way meeting another’s informational needs, Goldberg argues that one’s beliefs should be such that they are properly assertable should the need for such information arise. This, he argues, places constraints on both higher-order and first-order beliefs. In Chapter 15, our attention shifts to the epistemic significance of disagreement. Feldman examines what evidentialism has to say about three questions concerning peer disagreement: • (Q1): Can people involved in a peer disagreement reasonably disagree? • (Q2): How should one rationally respond to learning of a peer disagreement? • (Q3): What should a participant in a peer disagreement think about the rationality of the other peer? He notes that in peer disagreements, the parties gain evidence about their peer’s evidence regarding the disputed proposition, and goes on to examine the nature of this “evidence about evidence.” More specifically, he articulates several versions of the claim that evidence of evidence for a proposition is itself evidence for that proposition, and defends his favored formulation from objections. The chapter closes by examining what such a principle can tell us about the above questions concerning peer disagreement. In Chapter 16, Kelly addresses the question “How much weight should I give to the opinions of other people in arriving at my own opinions?” More specifically, he evaluates the Thermometer Model, according to which people who independently arrive at their beliefs are like the readings of a number of more or less accurate thermometers. This view seems to deliver the verdict that in two party cases of peer disagreement, substantial conciliation is called for by both parties, ending in states of agnosticism. Kelly subjects this account of the epistemic significance of disagreement to greater scrutiny than it has thus far received. He is particularly concerned with the question of which specific versions of Conciliationism the Thermometer Model supports and which it does not. He argues that such a model does not support the versions of Conciliationism that are most popular among leading Conciliationists, and that the versions that it does support are extremely strong views that we have independent reason to reject. In Chapter 17, Matheson examines the differences between disagreements in idealized contexts and disagreements in everyday contexts. While puzzles concerning the epistemic significance of disagreement are typically motivated by looking at the widespread and persistent disagreements in the real world, almost all of the literature on the epistemic significance of disagreement has focused on cases idealized peer disagreement. This fact might itself be puzzling since it does not seem that we ever encounter disagreements that meet the relevant idealized conditions. In this chapter, Matheson lays out the conditions of an idealized case of peer disagreement and motivates the

Introduction  11 Equal Weight View of disagreement in such circumstances. He proceeds to unpack the implications of stripping away each of the idealized conditions, and in doing, shows both why it is important to focus on idealized cases of peer disagreement as well as how the lessons from idealized cases carry over to everyday disagreements.

References Aquinas, T. (2006). Summa Theologica. Thomas Gilby et  al. (Trans.). 60 volumes. Cambridge:  Cambridge University Press. (Original edition, New  York:  McGraw-Hill, 1964–80.) Aristotle. (1941). Nicomachean Ethics, in Richard McKeon (Ed.), The Basic Works of Aristotle. New York: Random House. Clifford, W. K. (1999), “The Ethics of Belief,” in T. Madigan (Ed.), The Ethics of Belief and Other Essays. Amherst, MA: Prometheus, 70–96. (Original edition, 1877.) Descartes, R. (1964–76). Oeuvres de Descartes. Charles Adam and Paul Tannery (Eds.). 12 volumes. Paris: Libraire Philosophique J. Vrin. Original edition, Paris: Cerf, 1897–1913. Descartes, R. (1984–91). The Philosophical Writings of Descartes. John Cottingham, Robert Stoothoff, Dugald Murdoch, and (vol. 3 only) Anthony Kenny (Trans.). 3 volumes. Cambridge: Cambridge University Press. Goldman, A. (2011) “A Guide to Social Epistemology,” in Alvin Goldman and Dennis Whitcomb (Eds.), Social Epistemology: Essential Readings. Oxford: Oxford University Press, 11–37. Hume, D. (1975). An Enquiry concerning Human Understanding, in L. A. Selby-Bigge (Ed.), rev. edition, P. H. Nidditch (Ed.), Enquiries concerning Human Understanding and the Principles of Morals. Oxford: Oxford University Press. Hume, D. (1999). An Enquiry concerning Human Understanding. Tom L. Beauchamp (Ed.). Oxford: Oxford University Press. James, W. (1956). The Will to Believe and Other Essays in Popular Philosophy: Human Immortality. New York: Dover. Plato. (1997). Theaetetus, in John M. Cooper (Ed.), Plato: Complete Works. Indianapolis: Hackett. Plato. (1997). Meno, in John M. Cooper (Ed.), Plato: Complete Works. Indianapolis: Hackett. Plato. (1997). Republic, in John M. Cooper (Ed.), Plato: Complete Works. Indianapolis: Hackett.

PA RT  I

The Ethics of Belief: Individual

1 The Powers that Bind: Doxastic Voluntarism and Epistemic Obligation Neil Levy and Eric Mandelbaum

As the phrase is usually used, ‘doxastic voluntarism’ is the thesis that agents have the power to directly form beliefs for non-epistemic reasons. The thesis that we have such a power is an interesting one, one that is worthy of exploration on its own terms. However, it is often discussed because of its close connection to an even more interesting question: whether we have any epistemic obligations. The connection between the two is motivated by some version of the ought-implies-can principle; the thought is that we only have obligations to come to hold beliefs with particular contents if we have the power to form such beliefs. In this chapter, we argue for three theses: (1) we lack the power to form beliefs at will (i.e., directly); at very least, we lack the power to form at will beliefs of the kind that proponents of doxastic voluntarism have in mind;1 but (2) we possess a propensity to form beliefs for non-epistemic reasons; and (3) these propensities—once we come to know we have them—entail that we have obligations similar to those we would have were doxastic voluntarism true. Specifically, we will argue that we have obligations to avoid triggering these propensities to form beliefs that are unwarranted or even immoral. We therefore issue a warning: if you read this chapter, you will find yourself with more obligations at the end than you currently possess.2

  1  One of us believes that we lack the power to form beliefs at will while the other believes that we lack the power to form at will beliefs of the kind that proponents of doxastic voluntarism envisage. This is a dispute about the nature of beliefs, not about the nature of our powers.   2  Indeed, it may already be too late:  ceasing to read now might constitute what Smith (1983) calls a benighting act, by virtue of which you are culpable for your ignorance of your epistemic obligations.

16  Neil Levy and Eric Mandelbaum

1.1  Truth-Critical Deliberation and Voluntarism If we had the power to directly form beliefs (of the kind that proponents of doxastic voluntarism envisage; from now on we drop the qualification except when it is under discussion) for non-epistemic reasons, we might be required to justify the beliefs we form in this way. The request for justification would be significantly different from the request we routinely make of one another: rather than asking what evidence we can cite in favour of the belief ’s being true, a request for justification might ask for non-epistemic—moral or prudential, say—reasons for holding that belief. This would be an additional requirement, additional, that is, to the requirement we are sometimes under to justify our beliefs by reference to facts concerning their likelihood of being true. It would also be a more demanding requirement, inasmuch as these acts of belief formation would be voluntary. Voluntary behavior is, ceteris paribus, behavior that is apt for blame and praise, whereas non-voluntary behavior probably isn’t directly apt for blame or praise.3 It is widely agreed that we do not have the power directly to form just any beliefs. I cannot directly decide to believe that today is Wednesday, for instance. Call a belief that has no prior epistemic support an arbitrary belief. That strong voluntarism (Frankish 2007), the thesis that we have the power to directly form arbitrary beliefs, is false is more or less universally accepted. But weak voluntarism, the thesis that we have the power directly to form beliefs given certain epistemic conditions, is more controversial. We claim it too is false, and for precisely the same reasons as strong voluntarism. Strong voluntarism is false because forming an arbitrary belief with the content p requires that we simultaneously bring it about either that we forget that we have the belief only because we have decided, for non-epistemic reasons, to form such a belief, or that we change our view of the evidence so that we take the belief to be justified independent of our act of belief-formation. But we do not have the power to do either of these things directly. We cannot directly alter the contents of our memory at will, nor can we directly alter our view of the evidence at will. (No doubt we can take indirect means to alter either our memory or our view of the evidence; we might, for example, hit ourselves in the head with a brick after we form a belief hoping that the ensuing amnesia knocks out our memory of the belief formation process without altering the belief itself. Such routes are clearly not direct in the relevant sense.) It might be thought that it would be no harder to alter our view of the evidence than it would be to change our belief—our view of the evidence is just another belief, after all. But the fact that our view of the evidence is just another belief doesn’t entail that we can alter it at will: altering this belief would require altering our memory that we have done so, or altering

  3  The claim that voluntariness is a necessary condition for blame is a controversial; one of us has defended the claim at great length in a number of places (e.g., Levy 2005, 2011). For opposing views, see Adams (1985) and Smith (2005).

The Powers that Bind  17 our view of higher-order evidence, and so on. Because we cannot complete an infinite series of acts, we can’t get ourselves to believe at will (Frankish 2007). Frankish argues that the facts just mentioned entail that it is nomologically non-contingently true that strong voluntarism is false. We shall suggest that he is wrong in the following way:  Frankish is right that strong voluntarism is false, but wrong in thinking that it is non-contingently false. It is a metaphysically contingent (though perhaps psychologically necessary) fact about us that we are unable directly to form beliefs for non-epistemic reasons. More importantly, perhaps, Frankish is also wrong when he asserts that there are no grounds for thinking that weak voluntarism is false. Frankish believes that under certain conditions, we can directly bring ourselves to have a belief. These conditions are epistemic: we must have better evidence for the belief than for its negation. In other words, we can directly form a belief when, and only when, so doing takes us from fence-sitting to belief.4 We do this, Frankish suggests, by formulating a policy of relying upon the content (i.e., taking the content to be true in a non-pro-tem fashion) in what he calls truth-critical deliberations, where deliberation is truth-critical when it relies upon premises the subject is disposed to accept in contexts in which truth is of central importance. (Frankish suggests that we can identify these premises with those premises we are apt to rely on in most contexts.) So doing just is forming the belief, so in doing this we directly bring it about that we have the relevant belief. We will not forget that we have the belief as a result of adopting the policy, nevertheless we will retain it because our view of the evidence permits us to believe that p. We argue that these claims are false.5 We can certainly adopt a policy of relying upon a claim in deliberation, but in so doing we will not bring it about that we possess all the dispositions constitutive of, or entailed by, a belief; not at once, at any rate.6 Frankish appreciates the need for an act of belief formation to bring it about that the agent has the appropriate dispositions and intends his account to satisfy this condition. He believes that adopting a policy of taking p as a premise in truth-critical deliberation makes one disposed to believe that p.7 But while it is possible that adopting such a policy will bring   4  Ginet (2001) defends a similar position.   5  One of us believes that though adopting a policy of relying upon a proposition does not bring about all the dispositions typically associated with a belief, nevertheless it does cause a state that deserves to be called a belief. This is because this author thinks that merely entertaining that p causes one to believe that p (Mandelbaum 2014). On this view we, strictly speaking, do not form beliefs for reasons at all (this claim holds over perceptual beliefs but a bit more subtlety is needed for dealing with beliefs that are inferred as consequences from other beliefs). Since this view is, to put it mildly, not mainstream, we will ignore the view in the body of the chapter so as to keep contact with mainstream usage, though we will occasionally note how adopting this view of belief would affect some of our claims.   6  Of course adopting a policy might bring us sooner or later to have the correlative belief, but no one denies that we can alter our beliefs in this indirect manner.   7  Using a premise in truth-critical deliberation is sometimes referred to as ‘accepting’ a premise—e.g., Bratman (1992) and Alston (1996). We shy away from this terminology since we think it causes confusion once the Gilbert framework is on the table.

18  Neil Levy and Eric Mandelbaum it about that one has some of the dispositions characteristic of a belief, it will not bring about all of them all at once. A sufficient number of dispositions central to the dispositional stereotype associated with the belief will not immediately follow suit. In particular, the phenomenal dispositions (Schwitzgebel 2002)—the dispositions to have the appropriate affective responses—will not follow all at once. Someone who adopts a policy of taking p as a premise in truth-critical reasoning does not thereby cause themselves to be surprised if very soon after it is demonstrated to her that ~p. Such a deliberator will not exhibit any more surprise at such a demonstration than previously, when she was a fence-sitter and thought it somewhat likely that p.8 So adopting a policy of taking p as a premise in truth-critical deliberation does not directly bring it about that one has the correlative belief.9 In fact, weak voluntarism fails for precisely the same reasons as strong: we will acquire all the dispositions associated with a belief that p (as opposed to the dispositions associated with thinking that p is more likely than ~p, which by hypothesis the agent has prior to adopting the policy) only if we simultaneously bring it about either that we forget we have the belief as a result of adopting a policy or that we alter our view of the evidence. Frankish’s mistake arises due to his apparent tendency to think of beliefs as all-or-nothing states. If belief was an all-or-nothing state, then it might be possible to move from non-belief to belief by way of adopting a policy, given that prior to adopting the policy one believed that the evidence was such as to make the belief more likely than not. But beliefs are not all-or-nothing states. Rather, they come in degrees (pace Holton forthcoming). This being the case, successfully bringing it about that one believes that p occurs just in case one has caused one’s subjective probability that p to rise by some nontrivial amount. It is this that we cannot do, all at once, by behaving as Frankish recommends.

1.2  Basic and Non-Basic Actions Though we think that Frankish is wrong to think that we can go from being fence-sitters to believers all at once, in the manner he recommends, we think that the facts that ensure that we do not have this power (and which also ensure—as Frankish   8  If one were beholden to the view that entertaining causes belief, then one would have to deny that the phenomenal dispositions are in any sense constitutive of belief. One of the authors does so deny that any particular phenomenal states are even associated with, never mind constitutive of, belief.   9  Two possible objections might be raised to this claim. First, one might object that if I take p as a premise and then use it to derive a conclusion that I have antecedent reason to believe is true (but didn’t know it followed from p) that might cause me to raise my credence in p. But even if this were right, this wouldn’t count as directly raising the credence in p merely by using it as a premise in truth-critical deliberation. Second, one might object that the mere activation of a thought raises its credence (á la Mandelbaum 2014) and the more one uses a premise in deliberation the more that premise will be activated. However, this sort of evidence applies to states that aren’t the full-blown beliefs that Frankish discusses. Accepting this line of thought would be consistent with the non-mainstream view of belief that’s been mentioned in the footnotes, and not the view that Frankish maintains.

The Powers that Bind  19 recognizes—that we do not have the power to form any arbitrary beliefs) are only contingently true. Those facts, recall, are that we succeed in forming the belief that p only if we can simultaneously bring it about that we forget that we have formed the belief that p for non-epistemic reasons, or we can change our view of the evidence. We think it is a contingent fact that we cannot do these things and, therefore, a contingent fact that strong voluntarism is false.10 An agent who could do one or both of these things is a conceptual, and perhaps even a genuine empirical, possibility. As we are using the terms, people directly bring themselves to believe that p if they believe that p immediately upon performing some basic action, which they perform because they intend to bring it about that they believe that p. We maintain that in order to be successful, this basic act must bring it about that they forget how they brought about the belief, or alter their view of the evidence. As it happens, we do not know how this can be done: we have no idea what steps an agent might take to bring it about that they achieve these things. But we think it is conceptually, and perhaps even empirically, possible that there are steps that an agent can take that would bring it about that they acquire the power to believe at will. A basic action is an action performed without any intermediaries. Raising one’s hand is a basic action for most of us because we do not raise our hands by doing anything else; rather, we just raise our hands. Agents for whom raising one hand is not a basic action are actual: an agent suffering from paralysis of one arm, for instance, might only be able to raise the hand on that side by way of doing something else (grasping it with their other hand, perhaps). Now, the basic/non-basic distinction, so understood, is not the distinction between actions which are causally complex and those that are causally simple. Raising one’s hand counts as a basic action even if neurally there are many stages involved (and there are). Rather, the basic/non-basic distinction is a distinction concerning how direct the action is for the agent: subpersonal complexity does not map onto personal directness. This fact entails that precisely the same action can go from non-basic to basic as the agent becomes more skilled at performing it. Consider a recent example of how agents have performed an action by way of doing something else. Building on earlier work that showed that some patients diagnosed in a vegetative state were able to perform a task in which they could voluntarily imagine playing tennis or navigating a familiar environment (Owen et al. 2006), inasmuch as the neural activity they exhibited did not differ significantly from controls, Monti et al. (2010) were able to develop what was, in effect, an fMRI-based communication system, in which a patient, again apparently vegetative, was able to answer ‘yes’ or ‘no’ to questions by imagining playing tennis or imagining navigating a familiar environment. Responding to these questions was, for him, not performing a basic action. Rather, it was performing an action by way of performing another, an act of imagination, which was basic. However we believe that it is not merely possible but even quite   10  It should be noted, however, that some people have denied that it is false at all. See, for instance, Steup (2000).

20  Neil Levy and Eric Mandelbaum likely, given technological developments, that control over a communication device like this could become automated. Someone might eventually learn to give the correct responses so efficiently that for them it would be the basic action of . Similarly, we believe, someone might learn to control a prosthetic device using an EEG-based control system by a series of stages, beginning with discovering (say) that they can cause it to perform a desired movement by imagining a certain motor response, but ending with them moving it by performing the basic action of moving the device in the desired manner. When this occurred, a non-basic action would have become a basic action. For the transformation of non-basic actions into basic to occur, the agent must learn to act with a high degree of efficiency and reliability. At the moment, the kind of indirect control that agents exercise over their beliefs,11 is neither efficient nor reliable. Instead, it is very much a hit-or-miss affair. We change our beliefs in this indirect way12 by the kinds of means that Pascal recommended to the person who wanted to bring about belief in God: associate with believers, immerse yourself in religious writings, try to think and act like a believer; eventually, perhaps via the mechanisms of cognitive dissonance reduction, you may find yourself with the correlative belief. If and when that happens, you will not forget that you have the belief via a process of self-manipulation, but you will find yourself with a different take on the evidence. From your perspective, it will seem to you that you have manipulated yourself into holding a belief that is independently warranted. Now, if some day we hit upon a method to reliably and efficiently induce these changes in ourselves, it might become possible to automate the process. For someone who automates the process, they will be able to perform the basic action of changing at least some of their beliefs. Such a person would be like Jonathan Bennett’s (1990) Credamites, who can will themselves to have a belief, except we think it is more realistic to suppose that agents who were much like us could perform such a basic action of willing belief by bringing themselves to have a different view of the evidence, rather than by forgetting how they brought the belief about. We say that because as a matter of fact real agents can and do indirectly induce beliefs in themselves, in the way recommended by Pascal, but in these actual cases the trick is performed by changing the agent’s view of the evidence: it is this trick that is available to be automated.

1.3  Belief Acquisition on the Cheap Though we do not believe that agents have the kinds of powers needed for doxastic voluntarism to be true, we do believe that we can—and do—form beliefs for non-epistemic reasons. In this section, we want to delve into the psychological   11  At least in cases in which they can’t make a belief true or false by acting directly on the conditions that make it true or false; say, making the belief that the light is off true by turning the light off (Feldman 2001).   12  Setting aside science fiction cases involving direct stimulation of the brain or memory-erasing pills.

The Powers that Bind  21 literature concerning how people actually form beliefs. After all, doxastic voluntarism is an empirical claim: it’s a claim about whether people actually can directly form beliefs for non-epistemic reasons. Even though some theorists in the literature attempt to deal with such a claim through conceptual analysis alone,13 we think it best to interweave such analysis with empirical findings. Thus we will now turn our attention to the literature on irrational and arational belief formation. In doing so, we will illuminate what powers and propensities to form beliefs for non-epistemic reasons human beings actually have. As a warm up, consider some findings from what we might term ‘the irrational belief formation’ literature. It has long been noted that motivated reasoning can affect one’s interpretation of evidence. When motivated reasoning does so affect one’s belief acquisition capacities, the result is a belief that is formed for reasons and hence is capable of being assessed both psychologically and epistemically. However, because the belief formation processes here are motivated by non-epistemic values and goals, the end result is generally less than epistemically respectable. Perhaps the most famous study stemming from this tradition is Hastorf and Cantril’s (1954) ‘They Saw a Game: A Case Study’. Hastorf and Cantril showed Princeton and Dartmouth students a tape of a (then recent) very rough American football game. Both sets of students watched the same film yet on average Princeton students saw Dartmouth players commit twice as many infractions as the Dartmouth students saw. Moreover, perceptions of the severity of the infractions also greatly differed between the two groups. Hastorf and Cantril’s venerable finding is now part of the common background knowledge on belief formation: what people want to see greatly affects their interpretation of the events they perceive. The irrationality inherent in these findings is that the students would or could not form impartial perceptual beliefs. This type of means-end sifting through the evidence is typical of other effects in the psychological canon that can also be filed under ‘motivated reasoning’ such as the confirmation bias. The confirmation bias can be found in different guises. For example sometimes it’s seen as a form of biased assimilation, sometimes as a biased information search.14 But on either reading, the phenomenon looks to be one where people form beliefs through a biased strategy with the end of making people reaffirm their already held beliefs as opposed to objectively viewing new evidence. In Lord et al. (1979), subjects were shown mixed evidence about capital punishment. The evidence was completely equivocal—for instance, one piece of evidence consisted of data that capital punishment had positive effects on both past and future murder rates, and another piece consisted of data that pointed to the opposite conclusion, and both had equal evidentiary value. People who had antecedently believed in capital punishment claimed that the evidence presented against capital punishment had little probative   13  See for example Hieronymi (2006) and Setiya (2008).   14  Of course, sometimes it’s just the name of a positive test search (such as in Klayman and Ha 1987); that use of the phrase is orthogonal to our purposes and should be set aside.

22  Neil Levy and Eric Mandelbaum value, whereas the anti-capital punishment folks claimed that the evidence presented in favor of capital punishment was unpersuasive. Furthermore, both groups ended up with more polarized attitudes after being exposed to evidence that ran contrary to their opinions: both groups ended up believing more in their antecedently held views after encountering evidence that was problematic for their belief system! Such data comes as no surprise to those who are familiar with the literature on cognitive dissonance. Displaying the effects of selective exposure to information is one of the core tricks in dissonance theory. The selective exposure effects show that people do not sift through evidence in an objective fact-seeking way; rather, people attempt to search for information which confirms what they already believe while avoiding information that might contradict what they believe. For example, Brock and Balloun (1967) played messages for subjects that warned of the ill effects of smoking, particularly the connection between smoking and cancer. These messages were interlaced with heavy static, which could be shut off by pushing an ‘anti-static’ button. The non-smokers reliably pushed the anti-static button more than the smokers. However, when the message was changed to one that disputed the link between cancer and smoking, the smokers reliably pushed the anti-static button more than the non-smokers. The same moral held true for churchgoers and atheists when they were asked to listen to a message that attacked Christianity: the churchgoers were happy to endure the static that made the anti-Christian message unintelligible. Of course, these are just a few examples from a deluge of work showing people’s relative receptivity to information that confirms their antecedent belief and their hostility to and avoidance of counter-attitudinal information. All of the effects canvassed so far can be understood as somewhat irrational effects on belief fixation.15 They are irrational because they (a) aren’t normatively respectable and (b) are explicable at the psychological level, a level of explanation where speaking of rational and irrational inferences and tendencies makes sense. However, there is also evidence about belief fixation that operates below the psychological level, evidence which is ground zero for theories that want to talk about descriptively adequate models of belief fixation. Certain forms of belief acquisition cannot be given the honorific of ‘rational’ or ‘irrational’; in order for something to be irrational it has to have a certain type of etiology. Let’s return to our aforementioned friend, the brick. Suppose you get hit in the head with a brick and the force of the brick causes you to believe, for no reason at all, that the universe has ten planets. Now no doubt, this would not be a particularly well-justified belief, but it would be odd to condemn you for your mode of belief acquisition. After all, it isn’t bad reasoning that led you to this belief. Instead, you formed this belief in a merely brute causal way. It is this type of causal process, brute causal incursions from beneath the psychological level causing certain beliefs, that we will term arational.

 

  Lexicographic note: we use ‘belief acquisition’ and ‘belief fixation’ synonymously.

15

The Powers that Bind  23 Arational belief formation is frightening because it is, on the face of it, seemingly impossible to counteract psychologically and very difficult to counteract at all. But more frightening still is the ubiquity of arationally caused doxastic—belief-like— states. In a series of fascinating studies Dan Gilbert and colleagues have accumulated evidence showing that people acquire belief-like states in a brute causal way. In particular, the work of Gilbert et al. appears to show that we are disposed to go on to form these states corresponding to any arbitrary proposition we happen to entertain. The basic arational paradigm exploits asymmetries in people’s memory of truths and falsehoods. In a typical experiment, participants are asked to participate in a learning task while they are intermittently placed under cognitive load and are then tested about what they learned. The recurrent finding is that when the learning occurs under even slight cognitive load, people tend to misremember statements that they learned were false as true, but do not tend to misremember true statements as false. An example should illuminate the situation. In one telling experiment participants were asked to learn nonsense word meanings. They watched a computer screen where sentences of the form ‘An X is a Y’ appeared, in which the ‘X’ was a nonsense word and the ‘Y’ was a word in English (e.g., ‘A suffa is a cloud’, from Gilbert et al. 1990). Right after participants read a sentence the screen flashed either the word ‘true’ or the word ‘false’, indicating whether the previous statement was accurate or not. Participants were also told to be on guard for a tone that would occur; the tone would occasionally sound and when it did the participants were to push a button as soon as possible. The tone was introduced in order to induce cognitive load. During the critical trials, participants read six true and six false claims. While reading four of these claims (two true, two false), the participants were interrupted by the tone (these were the critical trials, since load was occurring). At the end of the trials the sentences were then turned into questions (e.g., ‘Is a suffa a cloud?’) which the participants then answered. The added cognitive load did not effect the true statements: participants reliably encoded true statements as true. However, the load did significantly affect performance on false statements: false statements were consistently incorrectly encoded as true. Lest one think that the asymmetry between remembered truths and falsehoods holds just over ‘mere memory’, perhaps one more example would help to show how this acquired information is used in a belief-like manner. In Gilbert et al. (1993) participants were asked to watch a video screen with two crawling lines of text on it, one on top of the other. The top scroll contained text reports of two unrelated crime incidents. Participants were told that they would read both true and false details about the incidents, true statements appearing in black, false statements appearing in red. The bottom crawl did not contain any text, but instead had digits that slowly moved across the screen. Half the participants—the unburdened participants—were told to ignore these digits whereas half—the burdened participants—were told to peruse the digit crawl and to push a button anytime the number 5 appeared. At the conclusion of the video, participants were asked to recommend a prison sentence for the offenses, ranging from zero to twenty years, and they were also asked

24  Neil Levy and Eric Mandelbaum to assess the criminal’s personality. In particular, participants were queried as to how much they liked him, how dangerous he was, and how much counseling would help him. The false statements the participants read during the first phase of the experiment either exacerbated or mitigated the severity of the crime. The participants in the burdened condition were significantly more likely to be persuaded by the false information. The participants in the unburdened condition recommended a sentence of six years when the false information was extenuating and seven when it was exacerbating—not a significant difference—whereas their burdened counterparts recommended five years in jail in the extenuating condition and eleven years in jail in the exacerbating, which is a statistically significant difference. Significant differences were also found across the board when looking at the defendant’s likeability, benefit from counseling, and dangerousness. Thus, it appears that the falsehoods became integrated with the participants’ beliefs and affected a robust range of their responses. If they were not yet beliefs—we doubt that Gilbert’s subjects would have had all the dispositions associated with the correlative belief 16—they were clearly on the way to becoming full-blown beliefs. They certainly affected their beliefs proper, perhaps by biasing the manner in which they processed information. The propositions that the participants encountered while under load rippled through their cognitive system. In the first part of the study the participants not only processed the lies fed to them, but they made—presumably unconscious—inferences from those states which then informed their judgments concerning the duration of the sentence and the character’s likeability. This is quite interesting because it shows that the false information that is acquired acts like beliefs in a hitherto unseen way: the information is informationally promiscuous, a hallmark of beliefs. Informational promiscuity has been previously suggested as a criterion for separating beliefs from other belief-like, sub-doxastic states, such as intramodular representational states, e.g., the representations inside one’s language module (see Stich 1978). The attitudes the participants formed infiltrated and interacted with (presumably some subset of) their web of belief in order to produce the behavior the experiment detected.17 The asymmetries we have been discussing, ones between encountering truths and falsehoods while distracted, can be seen throughout the literature: a person put under cognitive load is apt to remember statements that they are told are false as true but not statements they are told are true as false. The experiments above displays that affirming a proposition (i.e., remembering the proposition as true)18 comes much easier than   16  In particular, we doubt that they would have asserted the belief. Of course, if you think that entertaining that p causes one to believe that p then you will sever the connection between belief and assertion. So such a theorist would think that although the subjects wouldn’t necessarily assert that they believe the false propositions they encountered, they’d still act as if they believed them, as we see in the aforementioned study.   17  Note that as far as this use of inferential promiscuity is concerned, what matters is that the information was available to a whole host of other processes and not that people were running honest-to-god inferences on the information.   18  Here ‘affirm’ and cognates should not be read as entailing consciousness of the content let alone intentional or effortful mental action.

The Powers that Bind  25 rejecting a proposition (i.e., remembering the proposition as false). Affirming is easier because it is a passive process, whereas rejecting is an active one: our cognitive architecture is set up to immediately affirm propositions as true. To go further and reject those propositions takes mental effort that is not necessary for the affirmation of a proposition. That is why something like belief fixation, operationalized above as the learning of sentences, can occur under load, but the rejection of a proposition—operationalized above as remembering that something is false—stalls when one is under cognitive load. The added cognitive load helps to shortcut the active rejection, but does not interfere with passive affirmation because the passive process is automatic and load does not affect a reflex. Compare how counting backwards from one hundred by increments of five would affect seeing a crossword puzzle versus completing the puzzle. The former will not be affected while the latter will be greatly affected. Rejecting a proposition is more like thinking than seeing, while affirming is more like seeing than thinking. The observed asymmetry can be explained if we assume that when propositions are initially processed they are encoded as true by default and can only subsequently be marked as false. Evidence for this view comes from a disparate array of sources and because of space constraints we couldn’t possibly canvass all of them (though see Mandelbaum (2010) for a painstakingly thorough review). However, before we leave the topic, we will describe one other experimental paradigm that speaks in favor of the mere-entertaining-causes-affirmation view. Instead of looking at acquisition of propositions that are personally meaningless we will now move our focus to forming beliefs about our own skills. To do so, we turn our attention to studies of belief perseverance in the face of experimental debriefings. In Ross et al. (1975) experimenters asked participants to read a collection of suicide notes and to sort the real ones from the fakes. Participants encountered twenty-five pairs of notes and were told that one note from each pair was a real note, the other a fake (in fact, all were fakes). After seeing each pair participants would judge which note was real and which fake and were then given feedback on their performance. After receiving the feedback the participants were partially debriefed. During the debriefing the participants were told that all the feedback they received was fictitious, it being arbitrarily determined beforehand regardless of the participants’ responses. After the debriefing the participants were asked to estimate both how many times they actually answered correctly and how many correct answers an average person would give. Interestingly, the information in the debriefing session did not affect participants’ opinions about their ability: if the participant originally received positive false feedback (e.g., twenty-four out of twenty-five correct), they believed that they were better than average at the task, and if they received negative false feedback (e.g., seven out of twenty-five correct), they believed they were worse than average at picking out real suicide notes from fake ones. The aforementioned experiment is not generally taken to illuminate anything about belief acquisition per se. It seems that the participants formed their beliefs in a

26  Neil Levy and Eric Mandelbaum reasonable way, based on the experimental feedback. Once they are told that the feedback was non-veridical they may just have had trouble updating their beliefs. Perhaps beliefs are ‘sticky’, in that once one has a belief, that belief is hard to relinquish. If so, then the debriefing effect wouldn’t tell us anything about belief acquisition per se, but rather belief perseverance. But what happens if the subjects are briefed before they take part in the study and receive false feedback? (Call such a technique ‘prebriefing’.) What if before sorting the notes they are told that the feedback they are about to receive is bogus? It turns out that prebriefing the participants has the same effects on subjects’ beliefs as false feedback. Wegner et al. (1985) replicated the Ross study except the participants were told prior to the task that the feedback would be dubious. Even after the explicit prebriefing the participants continued to behave as if the feedback was veridical. They were unable to reject the feedback they received, even though they knew it was bogus. These perseverance effects are easily explicable if we assume that the knowledge of the feedback persists because the participants automatically affirm the feedback when they hear it, even though they know the feedback is false. Since they are engaged in a relatively fast-paced experiment, the participants lack the mental energy to override the false claims.19 Although we have discussed only a few of the results from the empirical literature on belief acquisition, we think it’s wise to conclude that our belief-fixating faculties have been set up in the following way: we are designed to initially affirm any propositions that we happen to think about. In the absence of the time and resources to reflect on these affirmations we will acquire belief-like doxastic states, and—soon enough—tend to acquire the correlative belief. Our cognitive architecture is set to automatically lead to affirmation of the propositions we happen to token. Thus, belief-like states come cheap: whatever we happen to encounter, we are disposed to affirm, and eventually to believe. Only after the initial acquisition of the proposition can we go back and reject the information we have acquired. Rejection differs from the initial affirmation in that the rejection is neither automatic nor ballistic; rather, rejecting a proposition is an active mental endeavor. One’s level of education and intelligence—whatever that exactly is, if anything— does not affect one’s proposition-affirming faculties in the first instance. Instead, all of us are set up with dispositions to acquire beliefs in brute causal ways. Education and   19  There is an interesting question whether subjects in both kinds of belief perseveration paradigms form full-blown beliefs of the type Frankish discusses, or whether prebriefing brings about a doxastic state with a narrower set of dispositions than the state that is acquired in the debriefing paradigm. One might think that the difference in these paradigms is that in the debriefing paradigm belief formation is evidence-based, whereas in the prebriefing one it is not. Perhaps this difference explains why the first results in full-strength beliefs. However, it’s unclear in what manner the states acquired in the prebriefing studies differ from the debriefing ones. If prebriefing also results in full-strength beliefs, then this explanation looks more strained (though perhaps even in this condition beliefs are formed because something that looks evidence-like is presented). The more one is apt to see the same states acquired in both paradigms, the better the non-standard entertaining-is-believing line should look to one.

The Powers that Bind  27 intelligence are tools that can help the rejecting process, for example by giving us more motivation and greater levels of concentration needed to reject certain propositions,20 but do not affect the initial process, for the process works below the psychological level as it were (and since it works below the psychological level, the process appears to be arational). Evolution has conspired to make us initially gullible, a decent strategy for creatures like us who have more or less veridical perceptual faculties. But the design that worked so well in the Pleistocene is less than optimal in our current environs, where one is much more likely to encounter misinformation than in the environment of evolutionary adaptiveness. Today, as Keith Stanovich (2010) notes, we live in an environment in which other agents may start to arrange the cues to belief in ways that benefit them and not us.

1.4 Obligations Where does this leave us? In the first section of this chapter, we argued that the thesis that we can at will acquire beliefs, of the kind that proponents of doxastic voluntarism have in mind, is false. The evidence reviewed above show why the italicized qualification is necessary: we can certainly acquire doxastic states that resemble, and may actually qualify as, beliefs, more or less at will. The recipe for acquiring such states is simplicity itself: entertain the proposition that p and you will acquire a doxastic state with the content p. If states like this count as beliefs, then we can acquire beliefs at will. However, it is apparent that these kinds of doxastic states are not the beliefs that philosophers like Frankish think we can acquire in conducive circumstances. They are too unstable and fleeting to be states of the kind that have been at issue in the debate concerning doxastic voluntarism. Frankish suggests as a criterion for beliefs of the latter kind that they serve as premises in conscious ‘truth-critical deliberation.’ States apt to play that role are, inter alia, states that the agent is willing to assert, while the doxastic states acquired automatically are not (always) apt for assertion. However, these ‘thin’ states and our propensity to acquire them, when taken together with the knowledge that we have such propensities, do entail that we have obligations, even though doxastic voluntarism—understood in the traditional manner—is false. We can acquire doxastic states of a thin kind at will, and these states affect our behavior.21 Knowing how we acquire these states imposes obligations. We now know how to acquire thin doxastic states with content p: entertain the proposition; the mere fact of having done so will cause you to acquire a doxastic state with a corresponding content, and this state will, in turn, dispose you to come to have what all sides would accept is a

  20  For example, there is evidence that those with a higher ‘need-for-cognition’ score, do better at rejecting propositions than those with lower scores. See Mandelbaum (2014), particularly the discussion of ‘yea-sayers’ and ‘nay-sayers’.   21  In particular, they are likely to bias us toward gradually acquiring states with correlative (if not identical) contents that are beliefs on any plausible view.

28  Neil Levy and Eric Mandelbaum fully-fledged belief with a matching content. You can increase the likelihood that you will come to believe that p by making sure that you are properly distracted so that you don’t have time to consider and reject the proposition. The more one encounters the proposition under the requisite load, the more inferential tentacles the doxastic state will acquire. One can increase their credence in that belief even full well knowing that one is doing that simply by setting up one’s environment in a certain way and repeating the above procedure.22 This set up is unlike Bennett’s Credamites set up for we don’t need to have anterograde amnesia: we can full well know what we are doing, as long as we are properly distracted. One can have a perfectly well-functioning memory as long as one also has a perfectly well-functioning smart phone to serve as a distracter. So wherein lies our obligations? We assume here that we can’t have obligations over what we cannot control. Since doxastic voluntarism is, strictly speaking, false it appears that we cannot have obligations over what we believe, at least not in any simple or direct way. However, our walk through the empirical literature on belief acquisition pointed to a locus of control we do appear to have over our beliefs: if we are disposed to believe whatever propositions we encounter, then although we may not have direct control over what we believe we do often have control over what ideas we happen to encounter. For example, if we have control over anything, then we have control over what television channel we happen to put on. Suppose you want to watch Fox News because you are interested in seeing how certain types of media portray certain events. Even though this is a benign enough endeavor, you are putting yourself at risk of catching certain beliefs not because the beliefs are worth acquiring epistemically speaking, but rather simply because you encounter them; you run the risk of catching these beliefs in a similar way in which one catches a cold. And just as you can control whether or not you catch a cold to a certain degree—for example, by not kissing someone who has a cold—so too can you control whether or not you encounter, and hence believe, certain propositions. If the forgoing is correct, we—those of us who know about our propensities to acquire doxastic states through merely entertaining propositions—do have epistemic obligations which arise in the same kind of way in which they would arise were doxastic voluntarism true. We have obligations that arise from the kind of control we actually have over our belief formation process, limited and patchy though it is. Of course it is often necessary to engage with claims with which one does not agree; even, sometimes, to engage with claims that we know beforehand are dangerously and outrageously wrong. Political scientists, journalists, and cultural critics may all need to watch Fox News for somewhat similar reasons to why physicians expose themselves to infections: for the good of us all. Just as physicians can reduce their risks with proper infection controls, so those who deliberately expose themselves to Fox News   22  Of course one could accelerate the process by having certain affective variables line up the right way— after all, as the dissonance literature shows we are more inclined to believe what makes us feel better about ourselves (see, e.g., Thibodeau and Aronson 1992).

The Powers that Bind  29 can take steps to reduce the risk that they acquire the beliefs that Fox disgorges. As we noted above, affirmation leads to belief more reliably when we lack the time and resources to effortfully reject claims. Proper infection control requires that steps be taken to ensure that we are not under cognitive load, stressed, or tired when we enter the quarantine zone. Unfortunately there are no guarantees that these infection controls will succeed. Given that (a) the affirmation of claims is automatic and (b) affirmed claims immediately bias information processing, we can expect even the most fastidious Fox News watcher to acquire attitudes that are influenced by the pollutants they ingest. The viewer may acquire beliefs with the same content as the propositions they encounter or they may instead acquire beliefs that bear the taint of those propositions, either by being entailed by or associated with them.23 Effortful processing is too slow to keep up with the pace of claim generation.24 Moreover, though we have picked on Fox as a particularly egregious example of a source of mental contamination, contaminants are ubiquitous. Everywhere in contemporary society there are people attempting to persuade us of claims, to cause us to buy their products or their ideas. We encounter many of these messages when we are under load: stressed, tired, or distracted (when we are commuting and quickly pass by a billboard, for instance).25 Further, even if the agent is able to avoid mental contamination near the time of exposure, the danger has not passed. Even after we have evaluated claims, with whatever degree of success, we remain vulnerable to psychological mechanisms that leave us with unjustified beliefs. The ‘sleeper effect’ produces a delayed increase in the persuasiveness of a claim (Pratkanis et al. 1988). Sometimes a message is presented together with a discounting cue—e.g., message: ‘global warming is a myth’; discounting cue: source of message is the oil industry. Subjects who evaluate the message may initially give it little weight, but their confidence in its truth tends to rise over time. Why? One possibility is that when the message is recalled, the discounting cue is not, because there is only a weak association between message and cue; the message and the cue may be stored in different memory networks such that their inferential connections, and thus their decay rate, will differ.26 Since messages are often much more   23  In fact, the situation where one acquires a belief that is related but not identical to the content of the original perception may be more dangerous than one where the original content is believed straight away, for the former situation increases the likelihood of the subject failing to recall the source of the belief (thus increasing the ‘sleeper effect’; see later in this section).   24  It is also worth noting that some of the techniques Fox and other news organizations employ, such as the use of simultaneous scrolling text and unrelated news delivered verbally, could not be better designed to induce cognitive load; indeed they closely resemble standard methods used in social psychology to this end (see Mandelbaum 2010 for the gory details). This fact will make efforts at good cognitive hygiene all the more likely to fail.   25  It is important to note how light the required load can be. Merely self-regulating one’s own behavior is often enough load to be distracting and accelerate the quick acquisition effect (Gilbert 2002). Of course, in social situations (including the classroom!) one is, often enough, trying to self-regulate for fairly mundane reasons.   26  A similar style of explanation can be used to explain the ‘source monitoring’ errors underwriting the (false) recovered memory phenomenon (Schacter et al. 1997). Source monitoring (i.e., recalling the source

30  Neil Levy and Eric Mandelbaum vivid, they tend to be more accessible and available for recall. Consequently memory for the discounting cue can decay more quickly than memory for the claim, since the discounting cue is apt to not be as integrated and activated as the message is (Kumkale and Albarracín 2004).

1.5 Conclusion Though doxastic voluntarism is false, we nevertheless have epistemic obligations generated in much the same kind of way in which its truth would generate epistemic obligations. Were doxastic voluntarism true—i.e., were we to have the kind of direct control over the content of our belief that it entails—then we would have obligations to use this power well, and might reasonably be praised or blamed for (some of) the beliefs we formed. We do not have direct and immediate control over the content of our beliefs in the way envisaged, but we do have some degree of control: we make it likely that we will acquire beliefs by mere exposure to them. Just as we have obligations to take risks into account when we act, we have obligations to take the risk of forming unjustified and, worse, immoral beliefs into account when we expose ourselves to them. We can do various things to reduce the risks, but we cannot reduce them to zero except by avoiding exposure altogether. Of course when we decide how to act we need to weigh the risks against benefits: that an action carries with it a potential risk of harm does not entail that the action is impermissible, or even inadvisable. Everything depends on the magnitude of the harms, the probabilities of avoiding them, and the

of a signal) is particularly important in cases of recovered memories of abuse. In these cases, a therapist cues patients and prods them to remember (or ‘remember’) traumatic experiences that they have forgotten (or ‘forgotten’). Although it’s unclear whether any of these cases of recovered traumatic memory are veridical, it is clear that many of the supposed cases of recovered traumatic memory are not veridical. In these cases, the patients create, rather than recall, the event. The patient comes to ‘recall’ the event only after a therapist’s suggestion; because they fail to appropriately monitor the source of the memory, they take suggestion for recall (the Gilbert style experiments can be interpreted as presenting cases where subjects forget the source tags of ‘true’ and ‘false’). With regard to recovered ‘memories’, the effects may be potentiated by other features of the context. For instance, load can be brought about by the mere intensity of the situation (being asked to recall traumatic events). The problem then metastasizes because of the stereotypes normally invoked in this recall and the involvement of episodic memory. The patient generally has some negative feelings built up generally towards an older male figure, like a father, uncle, or priest. These figures have quite stereotypical traits that are easily conjured up. The combination of stereotype activation and cognitive load make for a volatile situation. In a study on stereotypes and source monitoring Sherman and Bessenoff (1999) found that when under cognitive load, participants are apt to default to judgments that fit a stereotype even if they were just shown that the stereotype does not hold for the case at hand. The interaction between stereotypes and cognitive load in recovered memories situations is exacerbated because episodic recollection is more demanding and effortful than semantic recollection (Tulving 1983). When patients are asked to recall traumatic memories, they are being asked to recall episodic memories and are thus put under additional load, making faithful memory search quite difficult. Semantic recollection, on the other hand, is much less effortful and can occur under load. Thus, when people are put under load they are apt to resort to the stereotypes that are stored in semantic memory while lacking access to their actual episodic memories.

The Powers that Bind  31 magnitude and probabilities of potential benefits. We think that there is an obligation on us to take all these facts into account when we act.

References Adams, R. M. (1985). ‘Involuntary Sins’, Philosophical Review 94: 3–31. Alston, W. (1996). A Realist Conception of Truth. Ithaca, NY: Cornell University Press. Bennett, J. (1990). ‘Why Is Belief Involuntary?’ Analysis 50: 87–107. Bratman, M. (1992). ‘Shared Cooperative Activity’, The Philosophical Review 101: 327–41. Brock, T., and Balloun, J. (1967). ‘Behavioral Receptivity to Dissonant Information’, Journal of Personality and Social Psychology 6(4): 413–28. Feldman, R. (2001). ‘Voluntary Belief and Epistemic Evaluation’, in M. Steup (Ed.), Knowledge, Truth, and Duty: Essays on Epistemic Justification, Responsibility, and Virtue. New York: Oxford University Press, 77–92. Frankish, K. (2007). ‘Deciding to Believe Again’, Mind 116: 523–47. Gilbert, D., Krull, D., and Malone, M. (1990). ‘Unbelieving the Unbelievable: Some Problems in the Rejection of False Information’, Journal of Personality and Social Psychology 59(4): 601–13. Gilbert, D. (2002). ‘Inferential Correction’, in T. Gilovich, D. Griffin, and D. Kahneman (Eds.), Heuristics and Biases: The Psychology of Intuitive Judgment. New York: Cambridge University Press, 167–84. Gilbert, D., Tafarodi, R. and Malone, P. (1993). ‘You Can‘t Not Believe Everything You Read’, Journal of Personality and Social Psychology 65 (2): 221–33. Ginet, C. (2001). ‘Deciding to Believe’, in M. Steup (Ed.), Knowledge, Truth, and Duty: Essays on Epistemic Justification, Responsibility, and Virtue. New York: Oxford University Press, 63–76. Hastorf A., and Cantril, H. (1954). ‘They Saw a Game: A Case Study’, The Journal of Abnormal and Social Psychology 49(1): 129–34. Hieronymi, P. (2006). ‘Controlling Attitudes’, Pacific Philosophical Quarterly 87(1): 45–74. Holton, R. (forthcoming). ‘Intention as a Model for Belief ’, in M. Vargas and G. Yaffe (Eds.), Rational and Social Agency: Essays on the Philosophy of Michael Bratman. Oxford: Oxford University Press. Klayman, J., and Ha, Y. (1987). ‘Confirmation, Disconfirmation, and Information in Hypothesis Testing’, Psychological Review 94(2): 211–28. Kumkale, G. T., and Albarracín, D. (2004). ‘The Sleeper Effect in Persuasion: A Meta-Analytic Review’, Psychological Bulletin 130(1): 143–72. Levy, N. (2005). ‘The Good, the Bad and the Blameworthy’, Journal of Ethics and Social Philosophy 1: 1–16. Levy, N. (2011). Hard Luck. Oxford: Oxford University Press. Lord, C., Ross, L., and Lepper, M. (1979). ‘Biased Assimilation and Attitude Polarization: The Effect of Prior Theories on Subsequently Considered Evidence’, Journal of Personality and Social Psychology 37(11): 2098–109. Mandelbaum, E. (2010). ‘Thinking is Believing: An Essay on the Unbearable Automaticity of Believing’, Ph.D. Dissertation, University of North Carolina, Chapel Hill. Mandelbaum, E. (2014). ‘Thinking is Believing’, Inquiry 57(1): 55–96. Monti, M. M., Vanhaudenhuyse, A., Coleman, M. R., et al. (2010). ‘Willful Modulation of Brain Activity in Disorders of Consciousness’, New England Journal of Medicine 362: 579–89.

32  Neil Levy and Eric Mandelbaum Owen, A. M., Coleman, M. R., Boly, M., Davis, M. H., Laureys, S., and Pickard, J. D. (2006). ‘Detecting Awareness in the Vegetative State’, Science 313: 1402. Pratkanis, A. R., Greenwald, A. G., Leippe, M. R., and Baumgardner, M. H. (1988). ‘In Search of Reliable Persuasion Effects: III. The Sleeper Effect is Dead. Long Live the Sleeper Effect’, Journal of Personality and Social Psychology 54(2): 203–18.  Ross, L., Lepper, M., and Hubbard, M. (1975). ‘Perseverance in Self-Perception and Social Perception: Biased Attributional Processes in the Debriefing Paradigm’, Journal of Personality and Social Psychology 32(5): 880–92.  Schacter, D. L., Norman, K. A., and Koutstaal, W. (1997). ‘The Recovered Memories Debate: A Cognitive Neuroscience Perspective’, in M. Conway (Ed.), False and Recovered Memories. New York: Oxford University Press. Schwitzgebel, E. (2002). ‘A Phenomenal, Dispositional Account of Belief ’, Noûs 36: 249–75. Setiya, K. (2008). ‘Believing At Will’, Midwest Studies in Philosophy 32(1): 36–52. Sherman, J., and Bessenoff, G. (1999). ‘Stereotypes as Source-Monitoring Cues:  On the Interaction between Episodic and Semantic Memory’, Psychological Science 10(2): 106–10.  Smith, A. M. (2005). ‘Responsibility for Attitudes: Activity and Passivity in Mental Life’, Ethics 115: 236–71. Smith, H. (1983). ‘Culpable Ignorance’, Philosophical Review 92: 543–71. Stanovich, K. (2010). Rationality and the Reflective Mind. Oxford: Oxford University Press. Steup, M. (2000). ‘Doxastic Voluntarism and Epistemic Deontology’, Acta Analytica 15: 25–56.  Stich, S. (1978). ‘Beliefs and Subdoxastic States’, Philosophy of Science 45: 499–518.  Thibodeau, R. and Aronson, E. (1992). ‘Taking a Closer Look:  Reasserting the Role of the Self-Concept in Dissonance Theory’, Personality and Social Psychology Bulletin 18(5): 591–602. Tulving, E. (1983). Elements of Episodic Memory. New York: Oxford University Press.  Wegner, D., Coulton, G., and Wenzloff, R. (1985). ‘The Transparency of Denial: Briefing in the Debriefing Paradigm’, Journal of Personality and Social Psychology 49(2): 338–46. 

2 Deciding to Believe Redux Andrei A. Buckareff

The ways in which we exercise intentional agency are varied. I take the domain of intentional agency to include all that we intentionally do versus what merely happens to us. So the scope of our intentional agency is not limited to intentional action. One can also exercise some intentional agency in omitting to act and, importantly, in producing the intentional outcome of an intentional action. So, for instance, when an agent is dieting, there is an exercise of agency both (1) with respect to the agent’s actions and omissions that constitute her dieting behavior and (2) with respect to the agent’s achieving the intended outcome of losing weight. In our mental lives we exercise intentional agency both by performing mental actions and by intentionally producing certain outcomes at which our mental actions are aimed.1 The nature and scope of our intentional agency with respect to controlling the acquisition of mental states such as beliefs, desires, and intentions is a topic that is of interest in its own right. In this essay, I will focus solely on our control over the acquisition of beliefs. Understanding what sort of control we have over our beliefs has far-reaching implications. For instance, theorizing about self-deception and wishful thinking is aided by theorizing about what if any intentional agency we can exercise with respect to acquiring beliefs. Another often-mentioned concern that motivates thinking about doxastic agency comes from religion (when conversion requires a change of belief). We also hold persons morally and epistemically responsible for beliefs they have or fail to have.2 Finally, some deontological theories of epistemic justification require that   1  For more on mental action generally, see the essays in O’Brien and Soteriou (2009). See also Buckareff (2005, 2007).   2  There are exceptions to the widespread view that ascriptions of epistemic responsibility are tied to the sort of doxastic control agents may exercise. For instance, David Owens argues that exercises of doxastic control are not subject to epistemic norms. Rather, “the norms of practical reason determine the rationality of the deed” (2000: 82). Such control, he notes, “underwrites a conditional responsibility to conform our beliefs to epistemic norms when it is prudent (or morally obligatory) so to do” (2000: 83). Things are no better, according to Owens, if the exercise of doxastic control is like the control exercised over basic actions. He writes that, “the truth or falsity of the claim that we can induce beliefs by means of basic acts has no significance for epistemology, internalist or otherwise” (Owens 2000: 85).

34  Andrei A. Buckareff agents be able to exercise a robust form of doxastic control.3 Fruitful work on any of these problems requires that we have an account of our intentional agency in acquiring beliefs. There are at least three loci of doxastic control. The first is over acquiring beliefs. The second is over maintaining beliefs. The third is over how we use our beliefs.4 I am chiefly concerned with the first locus of doxastic control in this essay, but I will say something about the second locus along the way. Also, I will only consider one way we might exercise control over acquiring beliefs.5 Specifically, I will present an argument against direct doxastic voluntarism (DDV). By “DDV” I mean the thesis that agents can consciously exercise the same sort of direct voluntary control over coming to acquire a doxastic attitude—such as belief, suspension of belief, or disbelief—that they exercise over uncontroversial basic intentional actions. An agent S’s action A at a time t is a basic action if there is no other action B at or before t that S performs by which S performs A.6 For instance, if I kick a ball, my kicking a ball is something I do by moving my leg. If moving my leg occurs in direct response to my intention to kick the ball, then moving my leg is a basic action.7 If DDV is correct, then coming to believe can be a basic action-type just like moving my leg. DDV, or something very close to it, was Bernard Williams’ target in his 1973 paper, “Deciding to Believe.” Williams’ argument is widely regarded as having been a failure. But I think that Williams was on to something. Hence, in this chapter, while I do not attempt to resurrect Williams’ argument, I develop and defend a revised argument for a thesis that is quite close to Williams’. I will proceed as follows. First, I will discuss Williams’ (1973) failed attempt at showing that DDV is conceptually impossible. This will be followed by a discussion of some constraints on our belief-forming activities. I will then clarify my target a bit more than Williams does in his original paper. Finally, I will present my own argument against the conceptual possibility of DDV.

2.1  A Failed Argument As I mentioned in the introduction, the core of the debate over DDV with which I am interested in this chapter has been over whether it is conceptually possible for an agent to decide to believe that p and come to believe as a direct result of the intention

  3  Alston (1989), Steup (2000), Feldman (2001), Ginet (2001), Ryan (2003), Nottelmann (2006), and many others, motivate an examination of the tenability of versions of doxastic voluntarism solely by appealing to the debate over deontological conceptions of epistemic justification.   4  For discussion of the difference between the first and third loci, see Bishop (2007: 28–41).   5  See Hieronymi (2006) for a useful discussion of different types of control we may exercise over mental states. I discuss some different varieties of doxastic control in Buckareff (2004, 2011).   6 The locus classicus for a defense of the distinction between basic and non-basic actions is Danto (1965).   7  While I use the example of an overt action, I do not wish to suggest that there are no mental actions that are basic actions.

Deciding to Believe Redux  35 formed by making a decision—i.e., “at will.”8 In other words, as previously mentioned, the debate with which I am concerned is over whether having the ability to exercise the same sort of control over coming to believe that we exercise over basic intentional actions is conceptually possible. That is, I am solely interested in whether it is conceptually possible that coming to believe is a basic action-type in its own right (Winters 1979: 243–4).9 Most importantly, the debate over doxastic agency that I am focusing on in this essay is over whether it is conceptually possible for one to exercise basic voluntary control while lacking any concern for the epistemic reasons that favor p in deciding to believe that p. Barbara Winters writes regarding this condition as follows: This condition is necessary because the philosophical controversy about belief at will concerns whether the model of free basic action can be applied to belief acquisition. If I were free to hold any belief I chose, I could decide to acquire only beliefs which were held by my peers or which I found pleasing. I might even select my beliefs randomly. The salient point is that the issue of a proposition’s truth or falsity does not play any role in my deliberations about whether or not to believe it; to answer the question, Should I believe that p is true? I do not ask, Is p true? (1979: 244, emphasis added)

I take it this means that the relevant question in the debate over DDV has been over whether coming to believe can be genuinely conceived of as a basic action that occurs in response to practical reasons rather than epistemic reasons.10 Any putative defense of a version of DDV that relies on a conception of doxastic agency being exercised in response to epistemic reasons and/or does not treat the sort of doxastic control as basic voluntary control is a defense of an entirely different thesis. Hence, such theories can be ignored for the purposes of this essay since their defenders are focusing on a different thesis about doxastic agency than the one with which I am concerned.11 Williams famously argued, in “Deciding to Believe” (1973), that the truthdirectedness of belief makes it conceptually impossible for us to have the ability to

  8  See Williams (1973), O’Shaugnessy (1980), and Scott-Kakures (1994) for seminal essays arguing against the conceptual possibility of direct doxastic voluntarism. See Govier (1975), Winters (1979), Bennett (1990), Pieper (1997), and Funkhouser (2003) for critiques of Williams’ argument. For a critical response to O’Shaugnessy, see Scott-Kakures (1994: 81–3). For a reply to Scott-Kakures, see Radcliffe (1997). Finally, Pojman (1985), Clarke (1986), Adler (2002), Church (2002), Frankish (2007), and Setiya (2008) also offer arguments for the conceptual impossibility of direct doxastic voluntarism.   9  Since the debate is often framed in terms of deciding to believe, it may be argued that deciding is a basic action and, hence, the debate is not over whether coming to believe can be a basic action. This is not obviously the case. If the goal of deciding is to settle on what to do in cases of practical uncertainty (see Mele 2003), then an agent has achieved her goal once she has settled on A-ing rather than not-A-ing and hence acquired an intention to A. What follows from the intention would be a basic action since it is not done by anything else even if caused by the intention.   10  See Bennett (1990: 90) and Audi (2001: 100–4).   11  So the varieties of doxastic voluntarism defended in Steup (2000, 2008), Ryan (2003), McCormick (2011), and McHugh (forthcoming), to the extent that they can be characterized as defenses of direct doxastic voluntarism, are providing defenses of far more modest variants of direct doxastic voluntarism than the version on which I am focusing.

36  Andrei A. Buckareff exercise the relevant sort of doxastic control at issue in the debate over DDV. It is worth quoting Williams at some length: [I]‌t is not a contingent fact that I cannot bring it about, just like that, that I believe something, as it is a contingent fact that I cannot bring it about, just like that, that I’m blushing. Why is this? One reason is connected with the characteristic of beliefs that they aim at truth. If I could acquire a belief at will, I could acquire it whether it was true or not; moreover I would know that I could acquire it whether it was true or not. If in full consciousness I could will to acquire a ‘belief ’ irrespective of its truth, it is unclear that before the event I could seriously think of it as a belief, i.e., as something purporting to represent reality. At the very least, there must be a restriction on what is the case after the event; since I could not then, in full consciousness, regard this as a belief of mine, i.e., something I take to be true, and also know that I acquired it at will. With regard to no belief could I know—or, if all this is to be done in full consciousness, even suspect—that I had acquired it at will. But if I can acquire beliefs at will, I must know that I am able to do this; and could I know that I was capable of this feat, if with regard to every feat of this kind which I had performed I necessarily had to believe that it had not taken place? (1973: 148)

This argument captures intuitions some philosophers have about the conceptual impossibility of exercising direct voluntary control over belief in response to an intention formed by an act of deciding to believe for practical reasons. But Williams’ argument has been widely regarded as unsuccessful. Winters (1979) offered one of the most formidable critiques of Williams’ argument. Many other criticisms have followed similar lines of reasoning against Williams.12 I will focus on Winters’ case given that her argument is one of the best known and is often cited. Winters claims that with a few alterations, the following argument emerges from the above quoted paragraph from Williams (1979: 252–3). 1. Necessarily, if I acquire a belief at will, then I will in full consciousness to acquire it irrespective of its truth. 2. Necessarily, if in full consciousness I will to acquire a belief irrespective of its truth, then after the event it is impossible that I in full consciousness regard it as a belief of mine and also believe that I acquired it at will. 3. Therefore, with regard to no belief could I believe that I had acquired it at will. 4. If, with respect to every acquisition of a belief at will I had performed, it is necessary that I not believe it took place, then I could not know that I was able to acquire beliefs at will. 5. Therefore, I cannot know that I am able to acquire beliefs at will. 6. If I can acquire beliefs at will, I must know that I am able to acquire beliefs at will. 7. Therefore, I cannot acquire beliefs at will. Winters claims that the problems with Williams’ argument lie with (2), (4), and (6). Because of the troubles, Williams’ argument is not sound.   12  See Govier (1975), Bennett (1990), Pieper (1997), and Funkhouser (2003) for additional arguments against Williams.

Deciding to Believe Redux  37 Regarding (2), Winters claims that it is too strong. (2)  can be contrasted with a weaker principle that I call “SW” for “sustained at will” (1979: 253):13 (SW) Necessarily, it is not the case that someone believes in full consciousness that she believes that p and that her belief that p is sustained at will. Winters notes that “(2) substitutes ‘acquired at will’ for [SW’s] ‘sustained at will’ ” (1979: 253). Winters claims that since a belief can be maintained for reasons different from those involved in its acquisition it is possible that one could believe that some belief was acquired at will so long as one no longer believes that the belief is sustained at will. For instance, one may now have evidence for the truth of the proposition believed (1979: 253). Winters notes, “These possibilities are incompatible with (3), which claims that there could be no belief of mine [past or present, as the antecedent of (4) clarifies] that I could regard as having been acquired at will” (1979: 253).14 According to Winters, the failure to establish (3) because of troubles with (2) suffices to undermine Williams’ argument for the conceptual impossibility of exercising direct voluntary control over belief. (4) and (6) are worth considering in the interest of assessing whether the rest of the argument would have been successful if (2) had been better supported. According to Winters, (4) claims that if in any instance it is necessarily false that an agent believes she has exercised the ability to believe at will, then an agent cannot have knowledge of having such an ability. Winters notes that Williams seems to assume that (4) follows from believing at will being an ability one necessarily cannot know one has exercised. She claims that such a principle is false because “I can become aware of capacities I have as a result of extrapolation from other data I have about myself or through the reliable testimony of others” (1979: 254). Furthermore, even if one is never aware of having believed at will, one could know one has the ability to believe at will from observing others coming to believe at will, and by having others report when one does it (Winters 1979: 255). Finally, regarding (6), Winters notes, “In (6) Williams maintains that having the ability to acquire beliefs at will requires awareness of its possession” (1979: 255). She takes it that the salient feature of believing at will that motivates (6) may be the requirement that believing at will be a basic action performed in full consciousness. But she adds that, even if this is the case, it does not follow that one must be aware of having this ability if one can acquire beliefs at will. One may want to believe that p and not be aware of one’s wanting to believe and similarly acquire an intention that causes one to believe. Upon introspection, one may become aware of epistemic reasons for believing. But it was one’s ability to believe at will that caused one to come to believe, even if one’s epistemic reasons are sufficient to now fully explain why one continues to have   13  SW appears as follows in Winters (1979):  “Necessarily, ~(∃x)(∃p) (x believes in full consciousness ⌈x believes p & x’s belief of [sic.] p is sustained at will⌉)” (253).   14  The bracketed material is in the original.

38  Andrei A. Buckareff the belief after it is acquired at will. So having an ability does not require that one be aware of one’s ability. The upshot is that because of problems with (4) and (6) the transition between (3) and (7) fails. Further considerations Winters entertains are not relevant for my purposes here, so I ignore them in the interest of being succinct. Winters’ critique of Williams’ argument for the conceptual impossibility of believing at will as well as those of a number of others are widely regarded as successful.15 It does not follow from the success of these critiques, however, that we must accept that it is conceptually possible that we can exercise direct voluntary control over coming to believe. In the remainder of this essay, I offer and defend an argument that, I believe, is impervious to criticisms such as those Winters and others level against Williams.

2.2  Belief, Truth, and Reasons Following Williams and others, I  will begin with the following, plausible assumption: there is a meaningful sense in which the concept of a belief is of a cognitive state that aims at truth or at conformity to truth.16 There are at least two features of belief that are often discussed in relation to the truth-directedness of belief. I address both of these in this section along with another, closely related feature. Together, these three features place constraints on what is conceptually possible with respect to our belief-acquiring abilities. First, a propositional attitude is a belief if it represents its content as true in a way that other propositional attitudes (including other cognitive states) do not. This may be due in part to the so-called direction of fit of belief. Belief has a mind-to-world direction of fit, the mind conforming itself to the world as it is presented to the agent; conative states, like desire and intention, have a world-to-mind direction of fit, the functional role of such attitudes would include the conformity of the world to the mind.17 Second, reasons for believing that p (and not merely motivational and explanatory reasons why a belief that p is worth acquiring/was acquired) are truth-conducive reasons—i.e., broadly, considerations that favor the truth of p. Regarding this second feature, it is reasonable to assume that epistemic reasons differ from broadly practical reasons insofar as epistemic reasons bear directly on the truth of p and count in favor of the truth of p. Practical reasons do not bear directly on the truth of p and do not count in favor of the truth of p, but they may count in favor of treating p as true (see Shah 2006).   15  I do have some worries about Winters apparently being committed to the view that it is conceptually possible that one have the ability to perform a certain type of action, yet never be aware of manifesting that ability. The concept of intentionally exercising an ability would seem to require that one can be and has been aware of the general ability in question. But I will not focus on this point in this chapter.   16  Unless otherwise indicated, the following claims should be understood as conceptual claims even if not explicitly articulated as such.   17  See Anscombe (1963/2000), Searle (1983), and Velleman (2000) for more on direction of fit.

Deciding to Believe Redux  39 Third, from the first-person perspective, the question “Do I believe that p?” is indistinguishable from the question “Is it true that p?” (see Edgely 1969; cf. Shah 2003). If I answer that it is not true that p, then I am not in a good position to answer honestly that I believe that p. Going further, if the last claim about the transparency of belief is correct, then we have a toehold for the following claim: If an agent believes that p, then she is disposed to believe that there is some evidence for p. This is not to say that the transparency thesis entails that to believe that p is to believe that there is some evidence for p. And I am not claiming that an agent who believes that p actually has adequate evidence for p. I am only claiming that an agent who believes that p would be disposed to take it to be the case that there are considerations that count in favor of the truth of p. Features of belief suggest that this last claim is justified. Recall what I said about the direction of fit of beliefs from mind-to-world. In believing that p an agent takes an attitude toward p that conforms to how the world is presented to her. If the agent’s attitude conforms to the way she takes the world to be presenting itself, then the considerations she acquires her attitude in response to are her reasons for believing that p. If this is right, then reasons for believing that p are reasons for holding that it is the case that p. And any reasons for holding that it is the case that p would be considerations that count for the truth of p and, hence, would be evidence for p. Of course, human agents are so constituted that non-epistemic reasons or other influences may actually play a role in explaining why an agent believes as she does. This is the case if the agent’s assessment of her evidence is biased or rendered more salient by her motivational states (which are non-epistemic reasons) at the time she evaluates her reasons for belief (as in textbook cases of “straight” self-deception).18 But this does not mean that from an agent’s conscious first-person perspective a belief could somehow be acquired in response to non-epistemic reasons (or that it could be sustained upon introspection). From this we get a constraint on belief. In order for an agent to sustain a belief that p, she should be able to identify some evidence for the truth of p (see Jones 2002; 2004). I take it this constraint on belief is effective upon an agent’s becoming aware that she believes that p. If she does not identify any evidence, continuing to believe will become very difficult, if not impossible. For failing to identify evidence places the agent in the position of believing that p is true and being disposed to assert that it is the case that p   18  There is psychological evidence that non-epistemic reasons often make evidence for the proposition an agent comes to believe more salient, causing the agent to acquire a belief in response to her putative epistemic reasons. For recent accessible treatments of the nature of both motivated and unmotivated biased belief, and the way heuristics in unbiased belief affect beliefs in ways that suggest that even beliefs acquired in the face of contrary evidence nonetheless seem to be shaped by evidence (because of motivated or cold biasing), see Nisbett and Ross (1980) and Mele (2001: ch. 2). Such shaping of belief by evidence, along with belief ’s being truth-directed is even evident in cases of persons who suffer from conditions such as Capgras delusion—the chief symptom of which is that sufferers believe that one or more close relatives have been replaced by an impostor. For a fascinating discussion of delusional beliefs that suggests that such beliefs, even when they fail to cohere with an agent’s related beliefs, aim at truth and are shaped by evidence, see Stone and Young (1997: 327–64).

40  Andrei A. Buckareff while simultaneously both recognizing that there are no epistemic reasons to believe that p and being disposed to assert that there is no reason to believe that p. The transparency of belief has consequences for acquiring beliefs as well. From the standpoint of conscious first-person doxastic deliberation, deliberating about whether to believe that p must include considering reasons for believing that p is true. Admittedly, one may engage in practical deliberation about bringing it about that one comes to believe that p, considering non-epistemic reasons that favor believing that p. But such practical reasoning aimed at bringing it about that one comes to believe that p is different from what one is doing when engaging in doxastic deliberation. Nishi Shah contends that from the first-person doxastic deliberative standpoint “the question whether to believe that p seems to collapse into the question whether p is true” (2003: 447; see also Shah 2006). He writes that, “A corollary to this hypothesis is that reasons for an affirmative answer to the question whether to believe that p must be considerations that are taken as relevant of the truth of p” (2003: 449). Shah argues that it is not merely a psychological fact about human agents that there is a “seamless shift in focus from belief to truth.” He asserts that it is “something demanded by the nature of first-personal doxastic deliberation” (2003: 447). Given this feature of doxastic deliberation from the first-person standpoint, if an agent is considering whether to believe that p, then she is considering whether p is true. Thus, if an agent is in a position to answer affirmatively either the question whether to believe that p or whether p is true, then she must take herself to have some evidence for the truth of p and she must regard her belief that p as based on her relevant putative evidence. It may be argued at this point that what I am claiming about belief acquisition cannot be correct. What about acquiring beliefs on the basis of a hunch? Certainly a hunch is not evidence. So what I am claiming about evidence and belief-acquisition cannot be right.19 I do not wish to claim that a hunch really is evidence or that it can never count as evidence. But if an agent can successfully acquire a belief on the basis of a hunch, then I take it that the agent would have to regard her hunch as having evidential value (most likely more evidential value than it actually has). This may occur in cases where someone is credulous. Someone who is more skeptical may afford a hunch no evidential value at all. Another case where a hunch may be taken to have more evidential value than it really has would be a case of motivated belief. Suppose that in such a case what is believed on the basis of a hunch is something the agent wants to be true. The hunch, accordingly, is assigned evidential value that it may in fact lack. What matters is that the agent takes the hunch to count in favor of the truth of p. But if a hunch is taken to lack any evidential value by an agent, then honest reflection on whether to believe that p on the basis of the hunch will result in a failed attempt at coming to believe that p.

 

  This objection was raised by Rico Vitz.

19

Deciding to Believe Redux  41 I am sure there are more objections that can be raised to what I have asserted in this section. For now, however, I will simply take it for granted that the claims I have made about concept of belief and constraints on the acquisition of belief are at least prima facie plausible.

2.3  Contra DDV Assuming the conceptual constraints on belief and belief-acquisition discussed in the previous section, in this section I develop an argument for the conceptual impossibility of exercising direct voluntary control over coming to believe by simply deciding to believe for practical reasons alone.20 More specifically, I shall argue that it is conceptually impossible for an agent to possess the ability to consciously exercise direct voluntary control over coming to believe. The relevant sort of doxastic agency would require that an agent be able to consciously exercise direct voluntary control over coming to believe on the basis of an intention formed by making a practical decision to believe, where the decision and subsequent event of coming to believe are motivated by practical reasons alone. I hope to show that, given the conceptual constraints on belief and the acquisition of belief, having any such ability is conceptually impossible.

2.3.1  Deciding and Judging In what follows, I should be understood as referring to practical decisions when I use ‘deciding’ and its cognates. I will take a practical decision to be a momentary mental action of actively forming an intention in response to practical reasons in order to resolve some practical uncertainty that arises when deliberating about what to do in some situation.21 So in asking whether it is conceptually possible for an agent to have the ability to consciously decide to believe, the challenge is to determine whether the following process is conceptually possible. First, an agent consciously deliberates about practical considerations that favor believing that p versus not believing that p. Next, the agent, being unable to settle on what to do, consciously decides to believe that p, thereby consciously forming an intention to believe that p. Finally, the agent is capable of being aware of coming to believe that p straightaway being caused by the intention in response to the practical reasons that favor acquiring a belief that p. It is the possession and exercise of any such ability in the sort of process described that I am claiming is conceptually impossible.

  20  It is worth noting that my argument is similar to arguments offered by Adler (2002: c­ hapter 2), Church (2002), Clarke (1986), O’Shaugnessy (1980), Scott-Kakures (1994), and others. I will not rehearse the differences between their arguments and the one offered here. I acknowledge the similarities between the arguments while also noting that critics of our strategies cannot simply offer a one-size-fits all response to any of the arguments against the conceptual possibility of direct doxastic voluntarism. The differences, as one should expect, between authors are subtle in places and more obvious in others.   21  See Mele (2003: 197–214) for a defense of such an account of deciding.

42  Andrei A. Buckareff I will assume that there is an analogous phenomenon to making a practical decision when we enter the theoretical sphere where doxastic deliberation occurs. Marking out their differences and similarities will be useful for what is to follow. First, regarding decision-making, making a decision is required in cases where an agent lacks the time and cognitive resources necessary to resolve some practical uncertainty by more extended methods such as further evaluation of practical reasons, etc. An example of this may be an agent having to get settled on what she will order when out for dinner with a group of people when she is asked what she will have from the menu. When one makes a decision and forms an intention by doing so the acquisition of an intention is the intentional outcome of the decision. But it is not the case that the intention to A is the intended outcome—i.e., the specific intention acquired with its content was not intended. Rather, the outcome of acquiring an intention (whether to A or refrain from A-ing) is intentional. Of course, the agent intends to make up her mind (acquire an intention). And she partially executes this intention by deciding (along with deliberating). But the agent, in deciding to A, does not decide to acquire the intention to A; and the reasons for deciding are simply reasons for A-ing. But that the intention itself is not the intentional object of the mental action of deciding does not make it any less the intentional outcome of the decision. It is no less intentional than each step one takes when walking to the store in order to get some bread. One does not intend to take each step, but each step is an intentional movement.22 When trying to settle on what to believe, judging (or, if you prefer, a doxastic-decision) may play a similar functional role in the economy of making up our minds about what to believe that practical decisions play in making up our minds about how to act. There may be some important differences between judging that p and deciding to A. But there are some similarities worth noting. For instance, judging is something an agent does by which she terminates doxastic deliberation and comes to acquire a belief in cases where the agent must make up her mind and get settled on what to believe. Judging is something done in response to theoretical reasons just as practical decisions are performed in response to practical reasons. In judging, an agent (at least partially) executes her intention to make up her mind about whether it is the case that p. Additionally, an agent may acquire a belief by judging just as she may acquire an intention by deciding. Finally, if we suppose that an agent can judge in cases of feeling theoretically uncertain just as one can decide in cases of feeling practically uncertain, we can regard the strength of the attitude acquired in either case to be determined by the strength of the reasons that cause the formation of the attitude. So an agent may acquire a partial belief in response to

  22 I  am here assuming that the Simple View of intentional action is incorrect and that the Single Phenomenon View is correct. For defenses of versions of the Single Phenomenon View, see Mele (1992), Brand (1997), and Bratman (1997). For defenses of the Simple View, see Adams (1986), Garcia (1990), and McCann (1998).

Deciding to Believe Redux  43 judging and a partial intention in response to a decision. In either case, the strength of the attitude (in a rational agent) is fixed by the reasons that count in favor of the attitude taken (see Holton 2008). So it is not necessary that a token of either type of mental action result in full belief or full intention. In both cases, the attitude formed can only be formed in response to reasons and its strength reflects the strength of the reasons for which it is acquired. In fact, it is more conceptually and psychologically realistic to regard either type of mental action as resulting in an attitude that is weaker than an attitude that is acquired on the basis of deliberation that terminated naturally with the agent feeling psychologically certain about what she will do or what she believes. Finally, there is an important difference between deciding and judging. When we judge, it is in order to make up our minds about what to believe or what to take as true. When we decide, we do so in order to make up our minds about what to do. We do not decide in order to make up our minds about what to intend. This difference may prove to be important in what follows.23

2.3.2  The Case Against DDV If an agent engages in doxastic deliberation, then an agent is deliberating about what sort of doxastic attitude to take towards a proposition. Assuming that the claims made above about belief, truth, and reasons are correct, then the reasons an agent consciously considers in doxastic deliberation are what the agent takes to be epistemic reasons, that is, the agent considers (putative) evidence for the truth of a proposition. So the belief formed by doxastic deliberation will be based on what an agent takes to be epistemic reasons. However, if an agent is engaging in practical deliberation about how to decide with respect to what to believe, then the agent is considering practical reasons in his deliberation. Practical reasons are non-evidential considerations. They are not reasons that count in favor of the truth of p. So, in the practical case, an agent is not deliberating about what to believe. At best, he is deliberating about what it would be best to come to believe for practical reasons. That is, he is considering reasons for the desirability of believing that p and for deciding to believe that p, but he is not deliberating about reasons for believing that p. The differences between cases of doxastic and practical deliberation are highlighted by contrasting two cases of deliberation—one doxastic and one practical. I propose comparing a case of doxastic deliberation versus practical deliberation where the purpose is the same. Namely, in both cases, the agent is trying to make up his mind whether to believe that p or to believe that ~p. By contrasting these cases, I hope to show that in the practical case, coming to believe on the basis of settling some practical deliberation by making a practical decision is conceptually impossible.

 

  In Buckareff (2011) I discuss forming beliefs and forming intentions in more depth.

23

44  Andrei A. Buckareff 2.3.2.1  Doxastic Deliberation over Whether to Believe that p Consider two scenarios. In the first, an agent is deliberating over whether to believe that p or to believe that ~p. The agent considers the evidence he has that counts in favor of p and the evidence that favors ~p. He concludes that the evidence favors p over ~p. Thus, the agent will simply end his deliberation by acquiring the appropriate belief. In the second scenario, the agent deliberates and reaches an impasse, with the preponderance of his evidence not clearly favoring p over ~p. I take it there are a few things that the agent can do. First, the agent may simply suspend belief by terminating deliberation and not acquiring any belief. Second, the agent may be unable to acquire a belief and, perhaps owing to practical pressures, non-doxastically accept that p, where non-doxastically accepting that p is a way of regarding p as true without believing p.24 Finally, the agent may judge that p (in the relevant sense of “judge” articulated in section 2.3.1, above). In that case, the agent may take the evidence to favor p slightly over ~p, but in order to acquire a belief the agent must make the mental effort involved in the mental action of judging that p. In the judging case, the agent will acquire a belief the strength of which is proportionate to the reasons the agent takes himself to have for believing that p. In what follows, I will ignore the first and second options. That is, I will not consider cases where an agent either suspends belief or non-doxastically accepts that p. I will focus solely on cases where the agent judges either that p or ~p, since, as I contended in section 2.3.1, I take judging to be the theoretical analogue to practical deciding. Now in both the first and second scenarios, the agent is deliberating about what to believe. In the first scenario, I assume that the belief acquired is full-belief, with the agent taking his evidence to clearly support p over ~p. In the second scenario, the agent’s evidence does not decisively favor p over ~p. Hence, if the agent judges that p and acquires a belief as a result, the belief acquired will be significantly weaker than in the first scenario. Importantly, if the agent in either scenario was self-aware, being conscious of what he was doing in deliberating about what he takes to be his epistemic reasons, then the entire process would be one in which the agent is sensitive to what considerations favor p over ~p. The agent would be able to indicate what reasons apparently favor p over ~p when deliberating and would be aware of why he settles on p over ~p. Finally, the belief would be one that the agent would be able to immediately recognize as having resulted from a process of considering what epistemic reasons the agent took himself   24  My own views on belief and non-doxastic acceptance most closely approximate the views of Stalnaker (1987), Bratman (1992), Velleman (2000c), and Shah and Velleman (2005). Other accounts of the distinction have been offered by Cohen (1992), Alston (1996), Engel (1998), Tuomela (2000), Bishop (2002, 2007), and Frankish (2004). There are significant differences between these authors. However, they all defend a distinction between belief and non-doxastic acceptance. The differences they claim for each are close enough for their respective theories to bear a family resemblance to one another. My position has changed over time, but I am still in basic agreement with the main features of the general account of non-doxastic acceptance I outline in Buckareff (2004, 2005).

Deciding to Believe Redux  45 to have for believing either statement. That an agent can possess and exercise such an ability to come to believe via such a process is not conceptually impossible. The belief is consciously acquired in response to considerations the agent takes to count in favor of the truth of p over ~p. 2.3.2.2  Practical Deliberation over Whether to Believe that p Now suppose that an agent is deliberating about whether to believe that p or to believe that ~p. In this case, the agent is only considering practical reasons. The agent does not consider any epistemic reasons. This scenario would be like deliberating about whether or not to have another glass of wine with one’s meal. The focus would be solely on practical considerations that favor believing that p versus believing that ~p. So what is being deliberated about are not considerations that favor the truth of p versus ~p. Consider two scenarios. Suppose in the first scenario that the agent concludes that she will believe that p on the basis of her deliberation, simply acquiring the intention to believe that p without having to make a decision. In the second scenario, the agent is faced with some practical uncertainty. Hence, she must make up her mind about whether she will believe that p or believe that ~p. She decides to believe that p, thereby forming the intention to believe that p. If we suppose that the agent can get as far as acquiring an intention to believe that p, whether or not she must decide to settle some practical uncertainty, can she successfully come to believe? That is, can she successfully execute her intention to believe that p? Remember, the agent is aware of what she is doing throughout the process being described above. Hence, I take it that she is aware of the sort of reasons for which she would be coming to believe. Assuming that the process is one of which she is continuously aware, even if she could successfully form the intention to believe, she would be unable to execute her intention without performing some other actions to produce the belief. Assuming any success in coming to believe as a result, the agent would not be exercising basic voluntary control. It would be indirect control. Any success in immediately coming to believe in this case would require that the agent successfully execute a present-directed intention to believe that p now while being unaware of any epistemic reasons supporting p. She would have to be able to come to explicitly represent p as true in the way fitting for a belief being fully aware that she is doing so only for reasons that do not bear at all on truth of p. I suggest that this is conceptually impossible given the characteristics of the concept of belief and the conceptual constraints on belief-acquisition discussed in section 2.2. What I believe I have shown thus far can be summarized as follows. In the case of practical deliberation, what is assumed is that no epistemic reasons are considered. In this case, the entire process involves considering reasons that count in favor of believing that p being worthwhile. In the case of doxastic deliberation, the entire process involves thinking about apparent epistemic reasons that the agent takes to count for or against the truth of p. In other words, in the case of doxastic deliberation, the process is directed at determining whether p or ~p is true and what reasons favor either

46  Andrei A. Buckareff option. In the case of practical deliberation, the agent is not concerned with determining whether p or ~p is true; and there is no concern for the reasons that favor the truth of p or ~p. To the extent that there is any consideration of epistemic reasons in the practical deliberation case, the agent is only considering what counts in favor of it being true that believing that p or believing that ~p is the more worthwhile option. When the agent forms the intention to believe that p, she is in no position to answer whether or not p is true. The considerations that count in favor of the truth or falsity of p were not what she was thinking about when engaging in practical deliberation. She can answer that she thinks that believing that p is worthwhile. But since she cannot honestly answer whether she thinks p is true, she is in no position to answer that she believes that p. Absent some sort of sudden awareness of apparent epistemic reasons for believing that p, any attempt at immediately believing that p on the basis of her intention to believe that p would be unsuccessful. And, ex hypothesi, such a sudden awareness of putative epistemic reasons is ruled out since the attempt at coming to believe that p is supposed to be made solely for practical reasons. If the foregoing is correct, then coming to believe is not a basic action-type. It is conceptually impossible to have the ability to exercise conscious direct voluntary control over coming to believe just like one would exercise over an uncontroversial basic action. It is conceptually impossible for an agent to get past acquiring an intention to believe that p because of the conceptual constraints on belief-acquisition articulated in section 2 of this chapter. Any success an agent may enjoy in executing the intention to come to believe immediately would require changing things so much as to have something that could no longer be accurately described as a basic action performed solely for practical reasons. For instance, someone may argue that an agent could execute an intention to come to believe that p now if there is an intervener who, being aware of the agent’s intention, somehow manipulates the agent so she comes to believe. In any such case, the agent’s coming to believe would not be an action, much less a basic action. This is so because any such intervention will result in coming to believe being the consequence of a deviant causal process that makes what happens in coming to believe not count as a basic action. Events and processes that result from causal deviance, including causal chains involving an intervener, are taken to fail to count as actions in the action theoretic literature.25 And what holds in the case of an intervener is true of any process that bypasses the normal causal route from an intention to some behavior. Causal deviance of any sort threatens the actional status of a mental event or process. So coming to believe that results from a deviant causal process cannot be a basic action. It may also be suggested that the agent could execute her intention if she suddenly became aware of epistemic reasons immediately upon forming her intention to believe   25  Exploring why causal deviance is a problem for actions is beyond the scope of this essay. I suggest that interested readers consult the introduction and essays in Bishop (1989) as well as Aguilar and Buckareff (2010) for more on causal deviance.

Deciding to Believe Redux  47 now. But if the agent comes to believe for a mixture of practical and epistemic reasons, as she would in this case, where the epistemic reasons are acquired immediately upon forming the intention to believe that p, then we have a case where coming to believe is no longer based entirely on practical reasons. Hence, we would no longer be dealing with the sort of case of exercising DDV-type control with which I am concerned in this essay. The argument that emerges from the foregoing is the following. I trust that I have provided ample reasons for accepting each of the premises. 1. Necessarily, if I can successfully consciously execute an intention to immediately come to believe that p, then my coming to believe that p could be caused in full consciousness solely by an intention to believe that p formed by deciding to believe that p on the basis of practical reasons alone. 2. Necessarily, if my coming to believe that p could have been solely caused in full consciousness by an intention to believe that p formed by deciding to believe that p on the basis of practical reasons alone, then I can consciously come to believe on the basis of practical reasons alone. 3. Necessarily, I  cannot consciously come to believe on the basis of practical reasons alone. Therefore, 4. Necessarily, I cannot successfully consciously execute an intention to immediately come to believe that p. That this argument resembles Williams’ and others that have been offered should be apparent to those familiar with the literature. But my strategy differs in two important ways. First, unlike Williams, I  am not simply emphasizing coming to believe without having any concern for the truth of p or without any concern for any considerations that count in favor of the truth of p. I am arguing that it is conceptually impossible to execute a present-directed intention to believe that p where the only reasons considered are practical reasons and no epistemic reasons that bear on the truth of p are considered at any point in the process. Absent some sudden awareness of apparent or actual epistemic reasons for believing that p or some sort of intervention, such an intention cannot be executed. In the event of any such awareness of epistemic reasons or the intervention of some third party, the case will no longer be one of successfully executing an intention to believe on the basis of practical reasons alone or it will fail to be a case of a genuine basic action due to the deviant causal chain that would obtain between the intention and coming to believe. Second, I part company with Williams by not allowing for the possibility of an agent coming to believe and then losing the belief. I am arguing that successfully consciously executing an intention to come to believe solely for practical reasons is conceptually impossible. For the reasons articulated in this essay, I am not even allowing for the conceptual possibility of acquiring and then losing the belief. My claim is that an agent

48  Andrei A. Buckareff could not get past acquiring an intention, much less successfully execute it only to lose the belief. If I  am right, then my argument is immune to the objections Winters raises in response to Williams’ argument that I discussed in section 2.1 of this essay. Remember that I have argued that it is conceptually impossible for an agent to have the ability to consciously execute a present-directed intention to believe that p solely for practical reasons. Because of this, my argument is immune to Winters’ counterexample that involves an agent sustaining a belief for epistemic reasons that was acquired for practical reasons. This is the case because the question with which I am concerned is over whether a belief can be consciously acquired in the first place via such a process involving practical reasons alone. I have argued that consciously executing the relevant intention to believe that p in response to practical reasons would be conceptually impossible. Furthermore, my argument does not fall prey to Winters’ two counterexamples involving an agent (1) exercising the relevant ability without knowing she has the ability when exercised and (2) exercising the ability without being aware of doing so. Again, this is the case because I am focusing exclusively on cases where an agent is conscious of what she is intentionally attempting. Of course, the argument I have offered may be vulnerable to some other objections besides those offered against Williams’ argument. However, it is immune to the standard objections offered in the literature to arguments like Williams’ for the conceptual impossibility of believing at will.

2.4 Conclusion As promised in the introduction, I have not resurrected Williams’ argument in this essay. But I have defended a claim similar to his own. If I have succeeded in achieving the main goal of this essay, then I have shown that it is conceptually impossible to successfully consciously exercise direct voluntary control over coming to believe on the basis of practical reasons alone. My argument, of course, is in many ways similar to Williams’ and the arguments of others who have argued for a thesis similar to his. But the conclusion I have reached is a bit more modest, and my strategy has differed in important ways that I hope are obvious to those familiar with Williams’ 1973 paper. Notwithstanding the differences between Williams’ argument and my own, if I am right, then while I have not vindicated Williams, I have shown that he was on to something.26

  26  I wish to thank Earl Conee, Andrew Cullison, Ward Jones, Daniel Mittag, Rico Vitz, and especially Richard Feldman for helpful comments on this chapter.

Deciding to Believe Redux  49

References Adams, F. (1986). “Intention and Intentional Action:  the Simple View,” Mind and Language 1: 281–301. Adler, J. (2002). Belief ’s Own Ethics. Cambridge, MA: MIT Press. Aguilar, J. and Buckareff, A. (Eds.) (2010). Causing Human Actions: New Perspectives on the Causal Theory of Action. Cambridge, MA: MIT Press. Alston, W. (1989). “The Deontological Conception of Epistemic Justification,” in Epistemic Justification: Essays in the Theory of Knowledge. Ithaca, NY: Cornell University Press, 115–52. Anscombe, G. E. M. (1963/2000). Intention, 2nd edn. Cambridge, MA: Harvard University Press. Audi, R. (2001a). “Doxastic Voluntarism and the Ethics of Belief,” in M. Steup (Ed.), Knowledge, Truth, and Duty. New York: Oxford University Press, 93–111. Bennett, J. (1990). “Why is Belief Involuntary?” Analysis 50: 87–107. Bishop, J. (1989). Natural Agency: An Essay on the Causal Theory of Action. New York: Cambridge University Press. Bishop, J. (2007). Believing by Faith: An Essay in the Epistemology and Ethics of Religious Belief. New York: Oxford University Press. Brand, M. (1997). “Intention and Intentional Action,” in G. Holstrom-Hintikka and R. Tuomela (Eds.), Contemporary Action Theory, Vol. 1:  Individual Action. Dordrecht: Kluwer, 197–217. Bratman, M. (1997). “Two Faces of Intention,” in A. Mele (Ed.), The Philosophy of Action. New York: Oxford University Press, 178–203. Brown, J. (2008). “Subject-Sensitive Invariantism and the Knowledge Norm for Practical Reasoning,” Noûs 42: 167–89. Buckareff, A. (2004). “Acceptance and Deciding to Believe,” Journal of Philosophical Research 29: 173–91. Buckareff, A. (2005). “How (Not) to Think about Mental Action,” Philosphical Explorations 8: 83–9. Buckareff, A. (2006). “Compatibilism and Doxastic Control,” Philosophia 34: 143–52. Buckareff, A. (2007). “Mental Overpopulation and Mental Action: Protecting Intentions from Mental Birth Control,” Canadian Journal of Philosophy 37: 49–66. Buckareff, A. (2011). “Action-Individuation and Doxastic Agency,” Theoria 77: 312–32. Church, J. (2002). “Taking it to Heart: What Choice Do We Have?” The Monist 85: 361–80. Clarke, D. (1986). “Doxastic Voluntarism and Forced Belief,” Philosophical Studies 50: 39–51. Cohen, L. (1992). An Essay on Belief and Acceptance. New York: Oxford University Press. Danto, A. (1965). “Basic Actions,” American Philosophical Quarterly 2: 141–8. Edgely, R. (1969). Reason in Theory and Practice. London: Hutchinson. Engel, P. (1998). “Believing, Holding True, and Accepting,” Philosophical Explorations 1: 599–616. Fantl, J. and McGrath, M. (2002). “Evidence, Pragmatics, and Justification,” The Philosophical Review 111: 67–94. Feldman, R. (2001). “Voluntary Belief and Epistemic Evaluation,” in M. Steup (Ed.), Knowledge, Truth, and Duty. New York: Oxford University Press, 77–92. Frankish, K. (2007). “Deciding to Believe Again,” Mind 116: 523–47. Funkhouser, E. (2003). “Willing Belief and the Norm of Truth,” Philosophical Studies 115: 179–95. Garcia, J. (1990). “The Intentional and the Intended,” Erkenntnis 33: 191–209. Ginet, C. (2001). “Deciding to Believe,” in M. Steup (Ed.), Knowledge, Truth, and Duty. New York: Oxford University Press, 63–76. Govier, T. (1975). “Beliefs, Values, and the Will,” Dialogue 15: 642–63.

50  Andrei A. Buckareff Hawthorne, J. (2004). Knowledge and Lotteries. New York: Oxford University Press. Hieronymi, P. (2006). “Controlling Attitudes,” Pacific Philosophical Quarterly 87: 45–74. Jones, W. (2002). “Explaining Our Own Beliefs:  Non-Epistemic Believing and Doxastic Instability,” Philosophical Studies 111: 217–49. Jones, W. (2004). “Pragmatic Believing and its Explanation,” Critíca 36: 3–36. McCann, H. (1998). The Works of Agency: On Human Action, Will, and Freedom. Ithaca, NY: Cornell University Press. McCormick, T. (2011). “Taking Control of Belief,” Philosophical Explorations 14: 169–83. McHugh, C. (forthcoming). “Exercising Doxastic Freedom,” Philosophy and Phenomenological Research. Mele, A. (1992). Springs of Action:  Understanding Intentional Behavior. New  York:  Oxford University Press. Mele, A. (2001). Self-Deception Unmasked. Princeton, NJ: Princeton University Press. Mele, A. (2003). Motivation and Agency. New York: Oxford University Press. Nisbett, R. and Ross, L. (1980). Human Inference: Strategies and Shortcomings. Englewood Cliffs, NJ: Prentice Hall. Nottelmann, N. (2006). “The Analogy Argument for Doxastic Voluntarism,” Philosophical Studies 131: 559–82. O’Brien, L. and Soteriou, M. (Eds.) (2009). Mental Action. New York: Oxford University Press. O’Shaugnessy, B. (1980). The Will, Vol. 1. New York: Cambridge University Press. Owens, D. (2000). Reason Without Freedom:  The Problem of Epistemic Normativity. London: Routledge. Pieper, L. (1997). “Self-Knowledge in ‘Deciding to Believe’,” Dialogue 36: 493–510. Pojman, L. (1985). “Believing and Willing,” Canadian Journal of Philosophy 15: 37–55. Radcliffe, D. (1997). “Scott-Kakures on Believing at Will,” Philosophy and Phenomenological Research 57: 145–51. Reed, B. (2010). “A Defense of Stable Invariantism,” Noûs 44: 224–44. Ryan, S. (2003). “Doxastic Compatibilism and the Ethics of Belief,” Philosophical Studies 114: 47–79. Scott-Kakures, D. (1994). “On Belief and the Captivity of the Will,” Philosophy and Phenomenological Research 54: 77–103. Searle, J. (1983). Intentionality:  An Essay in the Philosophy of Mind. New  York:  Cambridge University Press. Setiya, K. (2008). “Believing at Will,” Midwest Studies in Philosophy 32: 36–52. Shah, N. (2003). “How Truth Governs Belief,” Philosophical Review 112: 447–82. Shah, N. (2006). “A New Argument for Evidentialism,” The Philosophical Quarterly 56: 481–98. Stanley, J. (2005). Knowledge and Practical Interests. New York: Oxford University Press. Steup, M. (2000). “Doxastic Voluntarism and Epistemic Deontology,” Acta Analytica, 15: 25–56. Steup, M. (2008). “Doxastic Freedom,” Synthese 161: 375–92. Stone, T. and Young, A. (1997). “Delusions and Brain Injury: the Philosophy and Psychology of Belief,” Mind and Language 12: 327–64. Velleman, J. D. (2000). “On the Aim of Belief,” in The Possibility of Practical Reason. New York: Oxford University Press, 244–81. Williams, B. (1973). “Deciding to Believe,” in Problems of the Self. New  York:  Cambridge University Press, 136–51. Winters, B. (1979). “Believing at Will,” Journal of Philosophy 76: 243–56.

3 Varieties of Epistemic Vice Heather Battaly

Elizabeth Anscombe’s 1958 call for a revival of virtue in ethics has (arguably) been answered. Contemporary virtue ethics has since produced substantial and extensive scholarship on the topic of moral virtue. In the 1980s, virtue concepts and properties also began to take root in epistemology.1 The field of virtue epistemology has since generated a burgeoning literature on epistemic virtue. In both fields, analyses of virtue have been ubiquitous. But what about analyses of moral and epistemic vice? These have largely been overlooked.2 This chapter elucidates two different concepts of epistemic vice. According to Concept 1, ends or effects are what matter. An epistemic vice is a stable quality that reliably produces bad epistemic effects. Producing bad epistemic effects is both necessary and sufficient for epistemic vice. According to Concept 2, motives matter. Producing bad epistemic effects is insufficient for epistemic vice. At a minimum, epistemic vice (also) requires a blameworthy psychology—bad epistemic motives, or a false conception of the epistemic good. Section 3.1 introduces two familiar, and corresponding, concepts of virtue: virtues produce good effects; and virtues require praiseworthy psychologies. It argues that Virtue-reliabilists conceive of epistemic virtues as qualities that produce good epistemic effects; whereas Virtue-responsibilists conceive of epistemic virtues as qualities that require praiseworthy psychologies. Accordingly, Section 3.2 contends that Reliabilist vices are best analyzed as qualities that produce bad epistemic effects. Responsibilist vices are best analyzed as qualities that require blameworthy psychologies. Responsibilists typically model their analyses of epistemic vice on Aristotle’s analysis of moral vice.

  1  Sosa (1980).   2  For a recent analysis of moral vice, see Adams (2006: ­chapter 3). For a recent analysis of epistemic vice, see Swank (2000). Swank uses intuitions about demon-world cases to argue that epistemic vice should not be analyzed in terms of truth.

52  Heather Battaly Section 3.3 argues that Responsibilists can, but need not, endorse Aristotle’s analysis of vice. On Aristotle’s view, vice requires one to: (a) consistently perform acts that are in fact bad, (b) have an (objectively false) conception of the good, (c) be motivated to perform said acts because one (falsely) believes that they are good, (d) lack competing motivations and beliefs, and (e) feel no regret. Though Aristotle’s analysis of vice is one plausible variety of Concept 2, it is not the only plausible variety. Aristotle’s conditions on vice are jointly sufficient, but some of those conditions are not necessary. There are at least two additional ways to be vicious. First, vicious people can have correct conceptions of the bad, and be motivated to perform bad acts because they are bad. In ethics, this captures the super-villain’s vice of cruelty. Second, vicious people can negligently fail to have any conception of the good or bad; they can fail to have any evaluative beliefs about the matter. This captures the vices of laziness, apathy, and conformity. This section argues that Responsibilists can endorse epistemic analogues of any of these three varieties of Concept 2, while still being able to distinguish epistemic vice from epistemic akrasia and explain why the former is more blameworthy than the latter.

3.1  Two Concepts of Virtue What is a virtue? What is a vice? For starters, virtues are qualities that make one an excellent person. A  person can be excellent in a variety of ways:  for instance, she can be excellent insofar as she has good vision, or insofar as she is skilled at logical problem-solving, or insofar as she is open-minded, just, or benevolent. Broadly, virtues are excellences. In contrast, vices are defects. Vices are qualities that make us defective people. Analogously, a person can be defective in a variety of ways: for instance, she can be defective insofar as she has bad vision; or insofar as she lacks logical skills; or insofar as she is dogmatic, unjust, or cruel. Which qualities make a person excellent? Arguably, different qualities can make one an excellent person in different ways. The historical and contemporary literature on virtue emphasizes two key ways in which a quality can make one an excellent person.

3.1.1  Concept 1: Ends Matter—Virtues Attain Good Ends or Effects First, a quality might make us excellent by enabling us to reliably attain good ends or effects, like true beliefs, or the welfare of oneself or others. Accordingly, virtues just are qualities that consistently attain good ends or effects. If we adopt teleology, they are qualities that enable a person to attain her end or perform her function well; if we reject teleology, they are qualities that enable a person to attain good effects. This concept of virtue begins with the intuition that good ends or effects (e.g., true beliefs, the welfare of others) are what ultimately matter; qualities are valuable insofar as they get us those good ends or effects. Hence, virtues are qualities that effectively attain the good. Success in attaining good ends or effects need not

Varieties of Epistemic Vice  53 be perfect, but it must be reliable. That is, reliably attaining the good is necessary for virtue. This means that people who try but consistently fail to help others do not have the virtue of benevolence. Likewise, people who try but consistently fail to get true beliefs do not have epistemic virtues. Bunglers are not virtuous, even if their motives are pristine. Nor are agents consistently subject to bad luck (e.g., demon-victims). According to this concept, reliably attaining the good is also sufficient for virtue; any stable quality that reliably attains good ends or effects counts as a virtue, whether it is a hardwired capacity (e.g., vision), an acquired skill (e.g., in logic), or an acquired character trait (e.g., open-mindedness). Good motives are not required. So, venture capitalists who consistently succeed in helping others via charitable donations have the virtue of benevolence, even if they are motivated by selfish concerns. Likewise, students who reliably arrive at true beliefs as a result of their logical skills have epistemic virtues, even if they are solely motivated to get good grades.3 Virtue-reliabilism. Led by Ernest Sosa, Virtue-reliabilism argues that the epistemic virtues are reliable belief-forming faculties or skills. The paradigms of epistemic virtue include reliable vision, memory, induction, and deduction. Vision, etc., are virtues insofar as they reliably attain good ends or effects; specifically, true beliefs. Hence, Virtue-reliabilism employs Concept 1. Sosa makes this explicit: “there is a ‘sense’ of virtue . . . in which anything with a function . . . [has] virtues” (1991:  271). He argues that since attaining true beliefs is one of our primary epistemic ends (functions), epistemic virtues will be qualities that enable us to reliably attain true beliefs. In short, for Virtue-reliabilism, ends matter. There are four important features of Virtue-reliabilism. First, the epistemic virtues are reliable: they are dispositions to produce more true beliefs than false ones.4 Second, according to Virtue-reliabilism, epistemic virtues can be hard-wired or actively acquired. In Sosa’s words, “much or our intellectual competence comes with our brains, but much is due to learning” (2007: 86). Vision, if reliable, is a hard-wired virtue: our brains are wired to produce beliefs based on visual experience. In contrast, interpreting the images of a particle accelerator would be an actively acquired epistemic virtue—a skill. It takes time and practice to reliably produce true beliefs about the presence of various types of particles based on such images. We have no control over, and thus are not responsible for, the presence or absence of hard-wired capacities, like reliable vision. But, we can be praised or blamed for the presence or absence of acquired epistemic virtues that require effort to attain, like the skills of logical problem solving. Third, epistemic virtues do not require good intellectual motives. Hard-wired virtues like vision do not require learned intellectual motives of any kind, and thus   3  This concept of virtue is employed by Plato in Republic book I and by Aristotle in Nicomachean Ethics books I and VI. In contemporary virtue ethics, it is employed by Driver (2001). See Battaly (forthcoming: Chapter 1).   4  Sosa (1991: 139; 2007: 83–4).

54  Heather Battaly do not require the learned motive to care appropriately about truth.5 Nor do such hard-wired virtues require intentional intellectual actions. Intellectual actions are (roughly) acts that an agent intentionally performs in acquiring beliefs: e.g., considering objections, searching for evidence, testing hypotheses, etc. As long as our brains and eyes are functioning well, hard-wired virtues like vision will reliably produce true beliefs without our intentionally doing anything. Finally, Virtue-reliabilists conceive of the epistemic virtues as instrumentally valuable. True belief, or some subset thereof, is what ultimately matters; it is intrinsically good. The virtues are valuable because they are reliable means to getting true beliefs.6 In sum, for Virtue-reliabilists, the epistemic virtues just are stable qualities that reliably produce true beliefs. If a stable quality fails to reliably produce true beliefs (e.g., 20/200 vision, crystal-ball gazing), it is not an epistemic virtue; and if it succeeds in reliably producing true beliefs (e.g., 20/30 vision, induction from an ample track record), it is an epistemic virtue.

3.1.2  Concept 2: Motives Matter—Virtues Require Good Motives The historical and contemporary literature on virtue emphasizes a second way in which a stable quality can make one an excellent person. One might contend that getting good ends does not make a person excellent, or is not the only thing that does. It also matters why one attains, or tries to attain, good ends. In short, good motives matter; they are what make a person excellent. According to this second concept, good motives are necessary for virtue. To have a virtue, one must consistently have good motives. Reliable success in getting good ends is not enough for virtue, and may not even be required.7 Hence, venture capitalists who consistently succeed in helping others, but who only care about protecting their own business interests, do not have the virtue of benevolence. Likewise, students who reliably arrive at true beliefs, but who only care about getting good grades (not about truth), do not have epistemic virtues. Why would good motives be necessary for virtue if one was already producing good effects? There are two guiding intuitions that underlie this view. The first is that virtues are praiseworthy—they are a credit to us—and praise should only be attached to things within our control. Arguably, we have greater control over our motives than we do over our effects in the world. If our entrenched motives are acquired over time via practice and effort, then we have some control over whether we end up caring about helping   5  Caring appropriately about truth plays a key role in Responsibilist analyses of epistemic virtue. Sosa has recently argued that epistemic virtue does require aiming at truth. See Sosa (2011: 16, 22, 33). But he thinks aiming at truth is part of our hard-wired set-up, and is shared by children and animals.   6  Sosa argues that the virtues are instrumentally valuable in his (1991: 141, 225). Conceiving of the virtues as instrumentally valuable is the main impetus for the value problem. In response to the value problem, Sosa argues that epistemic virtues are instrumentally and constitutively valuable. See Sosa (2007: 88).   7  Montmarquet (1993), Zagzebski (1996), and Slote (2001) agree that virtues require good motives. But, Montmarquet and Slote argue that having good motives and performing appropriate acts is sufficient for virtue; attaining good ends or effects is not necessary. In contrast, Zagzebski contends that attaining good ends or effects is necessary.

Varieties of Epistemic Vice  55 others or only about helping ourselves. But, we have no control over bad luck, which can prevent us from producing good effects in the world. Due to no fault of our own, our donations to charitable organizations can end up in the hands of thieves instead of the people who need help, and our beliefs about medium-sized objects can turn out to be false because of an evil demon. Relatedly, the second guiding intuition is that virtues express character; they reveal what we care about, and what we value. Here, too, what we care about and value—our entrenched motives and conceptions of the good—are actively acquired. But attaining good ends need not reveal what we care about or value. Good ends may be attained by good luck (e.g., in an angel-world) or by hard-wired capacities, like vision, which operate in animals and children in the absence of any acquired values. Good ends may even be attained despite what we care about and value, as demonstrated by the opportunistic motives of the venture capitalist above.8 Virtue-responsibilism. Led by Linda Zagzebski (1996), Virtue-responsibilism argues that the epistemic virtues are acquired character traits over which we have some control and for which we are (partly) responsible. The paradigms of epistemic virtue include open-mindedness and intellectual courage. These qualities are virtues insofar as they involve praiseworthy motives that are acquired over time; specifically, entrenched motives for truth, or for understanding. These virtues get (some, or all, of) their positive value from the motive for truth.9 Hence, Virtue-responsibilism employs Concept 2. For Responsibilists, motives matter. Responsibilists agree that good intellectual motives are required for virtue. To illustrate, imagine that Watson and Crick cared only about winning the Nobel Prize, not at all about truth. If so, they lacked epistemic virtue. Their motives were sub-par, even though their discoveries were momentous. Responsibilists disagree over whether good intellectual motives, in combination with appropriate intellectual actions, are sufficient for virtue. James Montmarquet argues that they are. Zagzebski argues that they are not. There are four corresponding features of Virtue-responsibilism. First, Res­ ponsibilists disagree about whether the epistemic virtues require reliably getting true beliefs. Montmarquet argues that they do not—getting true beliefs is too dependent on luck. On his view, if we consistently care about truth and perform appropriate intellectual actions, then we have epistemic virtues, even if we unluckily end up in a demon-world where all of our beliefs are false (1993: 20). Zagzebski disagrees. She thinks that both good motives and good effects are required for virtue possession. On her view, part of what makes the epistemic virtues valuable is their reliable production of true beliefs. So, if we end up in a demon-world, the qualities we have are not epistemic virtues, even if our motives and actions are impeccable (1996: 185).

  8  This concept of virtue is famously employed by Aristotle in much of NE. In contemporary virtue ethics, it is employed by Hursthouse (1999) and Slote (2001). See Battaly (forthcoming: Chapter 1).   9  Montmarquet (1993) thinks the epistemic virtues get  all of their value from the motive for truth. Zagzebski thinks they get their value from the motive for truth and from the true beliefs they produce.

56  Heather Battaly Second, according to Virtue-responsibilism, epistemic virtues are praiseworthy and thus actively acquired. In Zagzebski’s words, “the . . . kind of praiseworthiness that applies to virtue . . . reflects the fact that the virtuous person might have been vicious instead . . . The fact that the person could have gone either way . . . distinguishes virtue from . . . other excellences, particularly all those that are natural or inborn” (1996: 105). To be praiseworthy, epistemic virtues must be (partly) under our control and thus actively acquired. They cannot be hard-wired capacities, like vision. Third, epistemic virtues require acquired intellectual motives and acquired dispositions of intellectual action. Virtue-responsibilists think that we are not hard-wired to care appropriately about truth. We may innately care about truth too little, or too much, and must learn to care appropriately. Such learning requires effort, time, and practice. Likewise, we must learn which intellectual actions are appropriate: e.g., whom to trust about which topics and when, when to look for evidence and when to stop looking, and when to consider or ignore alternative ideas. To illustrate, Zagzebski argues that the virtue of open-mindedness involves an acquired motive to get truth and an acquired motive to consider alternative ideas. It also involves an acquired habit of considering alternative ideas when one should, and reliably attaining true beliefs as a result. Finally, Virtue-responsibilists conceive of the epistemic virtues as intrinsically valuable insofar as the motivation for truth is intrinsically valuable. Zagzebski thinks the epistemic virtues are also instrumentally valuable insofar as they reliably produce true beliefs.

3.2  Two Concepts of Vice So, which qualities make a person defective? Which qualities are vices, and why? While analyses of virtue have been ubiquitous in the contemporary literature, analyses of vice have been few and far between. But, we can remedy this. If there are indeed two key concepts of virtue as described above, then we can use those concepts, and their underlying rationales, to develop corresponding concepts of vice. Accordingly, vices will be qualities that: (1) attain bad ends or effects, or (2) involve bad motives. I will argue that the vices that correspond to Reliabilist virtues are best analyzed as qualities that attain bad ends or effects; whereas, the vices that correspond to Responsibilist virtues are best analyzed as qualities that require bad motives. Given their respective analyses of virtue above, this result is hardly surprising. It is, perhaps, more surprising that testimonial injustice, as described by Miranda Fricker (2007), is best analyzed alongside the Reliabilist vices, as a quality that produces bad ends or effects.

3.2.1  Concept 1: Ends Matter—Vices Attain Bad Ends or Effects First, a quality might make us defective insofar as it consistently produces bad ends or effects, like false beliefs, or the harm of oneself or others. Here, as in Concept 1 above, ends or effects are what ultimately matter. According to Concept 1, virtues get their positive value from the intrinsically good ends they produce. Correspondingly, vices

Varieties of Epistemic Vice  57 will get their negative value by producing intrinsically bad ends or effects. In short, vices will be qualities that reliably produce the bad. Reliably producing the bad will be both necessary and sufficient for a quality’s being a vice. Thus, a ruler who intends to harm his subjects but consistently bungles the job, reliably producing their welfare instead, does not have the vice of cruelty. Likewise, the agent who wants to believe whatever is easiest (call him “Schmalileo”10), but who has the good luck of being in a world where this motive consistently produces true beliefs, does not have an epistemic vice. Since vices get their negative value from the bad ends or effects they produce, reliably attaining bad ends or effects is necessary for vice. Bad motives are not sufficient. Reliably attaining the bad is also sufficient for vice:  any stable quality that reliably attains the bad counts as a vice, be it a hard-wired faculty (vision), an acquired “anti-skill” (doing invalid derivations), or an acquired character trait (dogmatism). To illustrate, the ruler who wants to help his subjects but, being in a demon-world, consistently harms them instead, has the vice of cruelty. Similarly, the logic student who cares about truth, but who nevertheless consistently produces false beliefs, has epistemic vices—e.g., the vice of doing invalid derivations. In sum, one way for a person to be defective is to have a quality that consistently produces bad ends or effects in the external world. Having a quality that consistently produces false beliefs makes us defective, even if our internal psychologies are exemplary. 3.2.1.1 Reliabilist Vices The vices that correspond to Reliabilist virtues include hard-wired faculties (e.g., bad vision) and acquired “anti-skills” (e.g., employing the ad populum fallacy). There are two reasons why Reliabilist vices are best analyzed as qualities that attain bad ends or effects. First, we have no control over the presence or absence of hard-wired faculties, like bad vision. We cannot be blamed for bad vision; nor does bad vision reveal anything about our motives or values. Even animals can have bad vision. Consequently, what makes a hard-wired faculty, like bad vision, a vice must be something other than a blameworthy psychology. Second, recall that Reliabilist virtues (e.g., 20/30 vision) are instrumentally, not intrinsically, valuable. It is ends or effects (e.g., true beliefs) that are intrinsically good, not the virtues themselves. Virtues like 20/30 vision get their positive value from the intrinsic value of the true beliefs they produce. In short, 20/30 vision is a virtue because it is reliable; it produces a preponderance of intrinsically good effects (e.g., true beliefs) in the external world. Correspondingly, the negative value of Reliabilist vices will also be instrumental. It is ends or effects (e.g., false beliefs) that are intrinsically bad, not the vices themselves. Accordingly, vices like 20/200 vision will get their negative value from the negative value of the false beliefs they produce. So, 20/200 vision will be a vice because it is unreliable; it produces a preponderance of intrinsically bad effects (e.g.,

 

  Montmarquet (1993: 20).

10

58  Heather Battaly false beliefs). In sum, if Reliabilist virtues are instrumentally good, then Reliabilist vices will be instrumentally bad. To be instrumentally bad, a quality must produce more intrinsically bad effects (e.g., false beliefs) than intrinsically good ones (e.g., true beliefs). Hence, Reliabilist vices must be unreliable. Else, we could not explain their negative instrumental value. Virtue-reliabilists say relatively little about epistemic vices, but what they do say confirms the analysis above. Thus, Alvin Goldman claims that “processes like guessing, wishful thinking, and ignoring contrary evidence are deemed vicious because they (are deemed to) produce a low ratio of true beliefs” (1992: 160). In other words, processes like guessing, etc., are vices because they are unreliable. Sosa does not often use the term “vice”, but he does claim that agents who ignore contrary evidence have “flawed epistemic character” (1991:  241). He would likely classify agents who are color-blind, myopic, or superstitious as similarly flawed. He implies that such qualities are flaws because of their “cost in truth”: they are unlikely to produce true beliefs (Sosa 1991: 242). If the above analysis of Reliabilist vices is correct, then a quality’s unreliability is both necessary and sufficient for its being an epistemic vice. Hence, unlucky demon-victims, whose seemingly positive qualities consistently produce false beliefs, have epistemic vices; whereas lucky Schmalileo, whose seemingly negative qualities consistently produce true beliefs, does not. Sosa’s stated views are consistent with these claims. Though Sosa does not explicitly argue that demon-victims are vicious, he does argue that they are not virtuous relative to the demon-world, since in that world they are unreliable.11 With respect to agents like Schmalileo, Sosa thus seems committed to claiming that they are virtuous relative to their world, since in their world they are reliable. Hence, Sosa’s view yields (approximately) the same commitments as Concept 1: demon-victims are, at the very least, not virtuous, and Schmalileo is not vicious.12 3.2.1.2  Testimonial Injustice Miranda Fricker argues that prejudices in society infect our perceptions of who is credible and who is not. For instance, she contends that in the novel The Talented Mr. Ripley, gender prejudice prevents Herbert Greenleaf from seeing Marge as a source of knowledge about the disappearance of his son (2007: 87). Fricker conceives of the virtue of testimonial justice, which corrects these prejudiced perceptions, along Responsibilist lines. Thus, testimonial justice is actively acquired via deliberate reflection on, and revision of, one’s credibility judgments of speakers.13 There is also more to this virtue than   11  Sosa (1991: 289) thinks demon-victims are virtuous relative to the actual world.   12  Sosa (2007: 22) may treat Schmalileo the same way he treats the Gettier victim; in which case, he would claim that Schmalileo has true, virtuously produced beliefs, that do not constitute knowledge. To constitute knowledge, beliefs must be apt (accurate because adroit): agents must arrive at true beliefs because they are virtuously produced; they cannot arrive at true beliefs because of luck. Schmalileo, like the Gettier victim, arrives at true beliefs because of luck.   13  Fricker (2007: 91–6). Fricker thinks it is possible, but unlikely, for one to possess the virtue of testimonial justice without active effort (2007: 100).

Varieties of Epistemic Vice  59 the actions of reflecting and revising. In Fricker’s words, “virtue is equally concerned with the motivational states that lie behind actions” (2007: 127). Following Zagzebski’s model, Fricker argues that the virtue of testimonial justice requires entrenched motivations to get truth and justice and to neutralize prejudice in one’s credibility judgments.14 Like other Responsibilist virtues, the virtue of testimonial justice is praiseworthy and reveals the agent’s entrenched motives and values. So, it is surprising that testimonial injustice is best analyzed alongside Reliabilist, rather than Responsibilist, vices. What makes testimonial injustice a vice? Fricker argues that testimonial injustice— prejudiced perception of who is, and is not, credible—is not hard-wired,15 nor must it be actively acquired. Instead, this disposition can be, and often is, passively acquired: it is inherited from the community in which one is raised.16 In Fricker’s words, “human societies have prejudices in the air,” and these prejudices infect our perceptions of speakers, whether we want them to or not (2007: 96). On her view, such prejudices operate below the level of awareness, and without our permission (2007:  39). This means that testimonial injustice is pervasive; anyone who grows up in a racist or sexist society is initially infected with dispositions of prejudiced perception. In short, we have no initial control over the presence of testimonial injustice; we cannot prevent ourselves from inheriting it. In this respect, the testimonial injustice of a child has the same status as her bad vision; she is not blameworthy for either quality. Nor must testimonial injustice express one’s entrenched motives or values, though it can in, e.g., the KKK. Fricker argues that credibility prejudice often operates below the level of belief and motive. She thinks one’s prejudiced perceptions can even be “flatly inconsistent with [one’s] beliefs” (2007: 36). To illustrate, a card-carrying feminist, at the level of belief and motive, might still take the ideas of her male colleagues more seriously than those of her female colleagues due to prejudiced perception. So, one can have the quality of testimonial injustice even when one’s motives and beliefs are impeccable. Consequently, insofar as passively acquired testimonial injustice is a vice, what makes it so must be something other than a blameworthy psychology. It shares this feature with bad vision. So, what does make testimonial injustice a vice? From where does testimonial injustice get its negative value? Arguably, it gets its negative value from the bad effects that it produces; testimonial injustice is a vice because it consistently produces bad ends or effects. Although Fricker does not explicitly argue for this claim, her stated views offer evidence in its favor. For starters, she argues that testimonial injustice produces both epistemic and moral harm. Epistemically, it prevents the hearer from getting knowledge from the speaker. Greenleaf does not get knowledge from Marge, though Marge knows that Ripley killed his son. Morally, it produces both primary and secondary

  14  Fricker (2007: 99, 122).   15  But, Fricker (2007: 116) does seem to think that we have testimonial injustice in the state of nature. If so, it may be hard-wired.   16  Perhaps, this is also true of the “anti-skills” above.

60  Heather Battaly harms. In Fricker’s words, “the primary harm is the essential harm that is definitive of epistemic injustice in the broad. In all such injustices, the subject is wronged in her capacity as a knower” (2007: 44). That is, degrading a person as a knower, and thus as a human, is essential to, and definitive of, testimonial injustice. This harm causes secondary harms. For instance, the speaker may consequently lose confidence in her beliefs, and may even lose knowledge or be prevented from acquiring intellectual virtues (2007: 49). So, Fricker conceives of testimonial injustice as a quality that is defined by the harm it produces. Even though she does not explicitly call testimonial injustice a “vice”, it fits the analysis of vice offered by Concept 1. 3.2.1.3 Concept 1* One might object that there is a related, but distinct, concept of vice, Concept 1*, that is even better suited to analyze Reliabilist vices and testimonial injustice. According to Concept 1*, any quality that fails to be a virtue constitutes a vice: a quality need not produce bad ends or effects to be a vice; it need only fail to produce good ends or effects. Here, the underlying intuition is that vices are defects, and one way to be defective is to fail to get external goods—to fall short. Concepts 1 and 1* are in the same family and enjoy significant overlap. The primary difference between the two is that Concept 1 treats virtues and vices as contraries, whereas Concept 1* treats them as contradictories. In short, Concept 1 allows for qualities that are neither virtues (because they do not produce good effects) nor vices (because they do not produce bad effects); whereas Concept 1* does not. According to Concept 1*, any quality that falls short of getting external goods is a vice—a defect— even if it doesn’t produce bad effects. To put the same point differently, Concept 1 designates as vices all and only those qualities that produce bad effects. Whereas, Concept 1* designates as vices both (a) those qualities that produce bad effects, and (b) those qualities that fail to produce good effects without producing bad effects. To illustrate, consider the epistemic virtue that is featured on the television series Lie to Me, which dramatizes the work of Paul Ekman (2009). Characters in the series reliably detect when people are lying by observing their micro-facial-expressions. Let’s assume that this epistemic skill is an acquired Reliabilist virtue. There are two ways to fail to have this Reliabilist virtue. One might either (a) be unreliable—one might produce more false than true beliefs about when people are lying (based on their micro-expressions)—or (b) be non-reliable—one might fail to produce any beliefs about when people are lying (based on their micro-expressions) by abstaining. According to Concept 1, only (a) is indicative of vice because only (a) produces a preponderance of falsehoods. According to Concept 1*, both (a) and (b) are indicative of vices because both miss out on valuable truths. It is important to note that in cases where abstaining is not an option, there is only one way to fail to be reliable: by being unreliable. In such cases, Concepts 1 and 1* are co-extensive. To explicate, consider hard-wired belief-forming processes. People who have the hard-wired process of vision cannot prevent themselves from producing visual

Varieties of Epistemic Vice  61 beliefs. The hard-wired process ensures that visual beliefs are produced automatically. Consequently, such people cannot fail to have the Reliabilist virtue of vision by being non-reliable: by failing to produce visual beliefs. The only way they can fail to have the virtue of vision is by being unreliable: by producing more false visual beliefs than true ones. In short, given that one has hard-wired belief-forming process x, the only way to fail to reliably produce x-type beliefs is to be unreliable. This means that when it comes to hard-wired Reliabilist vices, Concepts 1 and 1* are co-extensive, though they offer different analyses of what makes a quality a vice.17 Something analogous can be said of testimonial injustice. According to Fricker, we cannot prevent ourselves from making credibility judgments; we make them involuntarily. Hence, we can’t fail to have the virtue of testimonial justice (i.e., fail to make unprejudiced credibility judgments) by failing to make credibility judgments. The only way to fail to make unprejudiced credibility judgments is to make prejudiced ones. Here, too, Concepts 1 and 1* are co-extensive but offer different analyses of what makes testimonial injustice a vice.18 Is Concept 1* better suited to analyze Reliabilist vices and testimonial injustice than Concept 1? Whether we think Concept 1* is better, worse, or equally good will likely depend on our intuitions about the deciding cases—cases where we can fail to produce beliefs. The problem is that in the deciding cases, our intuitions are likely to differ. Consider three deciding cases. First, suppose I go through life without acquiring the Reliabilist virtue of detecting lies based on micro-expressions. Consequently, I fail to form beliefs in this way, and am sometimes ignorant of deception. Do I thereby have a Reliabilist vice? Some will think that I do, since I am missing out on a sub-set of valuable truths. But, some will think that I do not; perhaps, because this process and its truths do not seem quite valuable enough. Do our intuitions change when we consider processes and truths that are valuable enough? They may. Second, consider people who go through life without acquiring the virtues associated with critical thinking. Assume that the failure to acquire such virtues prevents them from producing beliefs about a wide range of vital issues. Though they are not producing false beliefs about these issues, they are ignorant.19 Do they have Reliabilist vices? Here, we may be more likely to opt for Concept 1*. Finally, suppose you are in a demon-world, but you manage to abstain from forming some type (y) of beliefs. Do you thereby have a Reliabilist vice? According to Concept 1, abstaining is the ticket to avoiding vice in the demon-world. If you abstain, you are not vicious since you are not forming false y-type beliefs.20 But, according to Concept 1*, you do have a vice since you are still missing out on valuable truths. For 1*, epistemic vice is   17  Concept 1 claims 20/200 vision is a vice because it is unreliable; 1* claims it is a vice because it fails to be reliable.   18  Concept 1 claims it is a vice because it produces harms; 1* claims it is a vice because it fails to produce goods.   19  Similarly, see the film Idiocracy.   20  Of course, in the demon-world, you would still have plenty of other epistemic vices. See Battaly (forthcoming: Chapter 4).

62  Heather Battaly unavoidable in the demon-world. In short, I will not pretend to resolve the debate between Concepts 1 and 1* here. Each concept has its advantages. Concept 1 allows children, who have not yet acquired the virtues needed to form beliefs on vital issues, to count among the non-virtuous, rather than the vicious. Concept 1* allows adults who are ignorant of a wide range of valuable truths to count among the vicious, rather than the non-virtuous. Though I have argued that Concept 1 provides the best analysis of Reliabilist vices and testimonial injustice, Concept 1* also provides a plausible analysis. Concepts 1 and 1* are both far better suited to analyze Reliabilist vices than Concept 2.

3.2.2  Concept 2: Motives Matter—Vices Require Bad Motives Alternatively, one might contend that producing bad external effects does not make a person defective or is not the only thing that does. What (also) makes a person defective are features of her internal psychology. In other words, it (also) matters why a person produces the effects she does. Psychological features, like motives, matter—here, motives are intrinsically good or intrinsically bad. According to Concept 2 above, virtues get (some, or all, of) their positive value from good motives. Correspondingly, vices will get (some, or all, of) their negative value from bad motives. In short, psychological features, like bad motives, are required for vice: a quality will be a vice only if it includes a disposition to have bad motives, or other blameworthy psychological features. Why would bad motives be required for vice if one was already producing bad effects? The first underlying intuition is that vices are blameworthy and blame should only be attached to things within our control. Again, it is thought that we have greater control over our motives, which we acquire via practice, than we do over our effects in the world, which are subject to luck. In short, if we can produce bad effects due to no fault of our own (e.g., in the demon-world), then producing bad effects cannot be sufficient for vice. Vice requires something for which we can be blamed. Arguably, we can be blamed for our motives. We can be blamed for caring only about ourselves (and not others), or for caring only about getting comforting beliefs (and not true ones), provided that we have some control over developing these dispositions. We can also be blamed for our dispositions to act and our conceptions of the good. It is blameworthy features of our psychologies—e.g., bad motives, false conceptions of the good, dispositions to perform bad actions—that are required for vice. The second intuition is that vices, like virtues, express character. They reveal what we care about and value; i.e., our motives and conceptions of the good (or the bad). But producing bad effects need not reveal what we care about or value. We will produce bad effects in the demon-world even if we have impeccable motives and values. And animals and children can produce bad effects (e.g., via bad vision) in the absence of any acquired motives or values. To sum up, one way for a person to be defective is to have a blameworthy psychology. Producing bad effects is not sufficient for vice, and may not even be necessary.21  

  See Battaly (forthcoming: Chapter 4).

21

Varieties of Epistemic Vice  63 3.2.2.1 Responsibilist Vices The vices that correspond to Responsibilist virtues include qualities like dogmatism, epistemic self-indulgence, epistemic malevolence, and epistemic conformity. It may be obvious that Responsibilist vices are best analyzed as qualities that require blameworthy psychologies. After all, these qualities are acquired character traits, rather than hard-wired faculties. Unlike hard-wired faculties, which we share with animals, qualities like dogmatism reveal what kind of people we are. They express what we care about and value—the motives and conceptions of the good (or bad) that we have acquired. They are also blameworthy. We rightfully blame, e.g., politicians for their dogmatism and the electorate for its epistemic conformity. Arguably, blame requires a degree of control that we do not have over our effects in the world, but do have over our motives, actions, and conceptions of the good (or bad). Virtue-responsibilists say more about vice than their Reliabilist counterparts, confirming the analysis above. Thus, Zagzebski thinks that epistemic vices are structurally analogous to Aristotelian moral vices; i.e., they are character traits. Her list of epistemic vices includes, e.g., epistemic idleness, conformity, cowardice, wishful-thinking, and closed-mindedness (Zagzebski 1996: 152). She argues that moral and epistemic vices are blameworthy: “Virtues are qualities that deserve praise for their presence and blame for their absence. Even greater blame is due to a person who has the contrary of a virtue, namely, a vice . . .” (1996: 104). She even suggests that such blame may be primarily attached to bad motivations. In developing her “motivation-based theory,” she claims that “the main reason we criticize [a person who] guess[es] is that his guessing reflects poorly on his motivation . . . His belief-forming procedure shows a lack of motivation for knowledge” (1996: 207). Zagzebski also argues that epistemic vice is distinct from, and more blameworthy than, epistemic akrasia. Using Aristotle’s distinction between moral vice and akrasia as a model, she argues that unlike the epistemically vicious person, the epistemically akratic person has “the desire to be intellectually virtuous,” though she fails to act on that desire (1996: 154). Instead, the akratic acts on a bad motive: e.g., she believes whatever is comfortable, despite the fact that she knows she shouldn’t, or she lapses into ways of thinking which she knows are inappropriate (1996: 154). The epistemically vicious and the epistemically akratic perform the same actions and produce the same external effects. What distinguishes epistemic akrasia from epistemic vice are features of the agent’s psychology. The epistemically akratic person has competing motives—some good, some bad—whereas the motives of the vicious person are uniformly bad. The epistemically akratic person also has a correct conception of the epistemic good—her beliefs about what is epistemically good are true—whereas the epistemically vicious person has a false conception of the epistemic good—her beliefs about what is good are false. I will address Aristotelian vice and akrasia in section 3.3. The current point is simply that Responsibilists like Zagzebski analyze epistemic vices in terms of bad motivations and other blameworthy features of our internal psychologies.

64  Heather Battaly 3.2.2.2 Concept 2* One might object that there is a related, but distinct, concept of vice, Concept 2*, that is even better suited to analyze Responsibilist vices. According to Concept 2*, vices do not require that one possess a blameworthy psychology; e.g., objectively bad motives or a false conception of the good. They only require that one fail to possess a praiseworthy psychology; e.g., objectively good motives or a true conception of the good. This lowers the standard of vice, making it easier to be vicious. Here, the underlying intuition is that vices are defects, and one way to be defective is to fail to have a praiseworthy psychology—to fall short of good motives and values. Concept 2* is not a plausible analysis of Responsibilist vices if it entails that virtues and vices are contradictories—that any quality which falls short of good motives or values is a vice. On this strong reading of Concept 2*, children are vicious since they have not yet acquired praiseworthy psychologies. But the concept of vice employed by Responsibilists precludes this possibility. Responsibilist vices are restricted to stable character traits that have been acquired over time. Children fail to have both Responsibilist virtues and Responsibilist vices because they are in the process of acquiring character traits but have not yet reached stable dispositions. Responsibilist vices also warrant blame for their presence. But children cannot be blamed for failing to have praiseworthy psychologies. There are weak readings of Concept 2* that are plausible, one of which will be pursued in section 3.3.2.2.22 It is “weak” in the sense that it treats virtues and vices as contraries. According to the Aristotelian variety of Concept 2, vice requires a false conception of the good. But one can plausibly argue that Aristotle is incorrect. Vice does not require a conception of the good (true or false) since one can be vicious by negligently failing to have a conception of the good. On this reading of Concept 2*, vices will be stable, acquired character traits that express one’s failure to have a conception of the good and, correspondingly, one’s failure to care about the good, where these failures are blameworthy. It is possible to fail to have a conception of the good by failing to have any conception of value at all, simply because one doesn’t care enough to develop one. Adults who utterly fail to care about morality never bother to develop conceptions of moral value. They never evaluate their desires or ask tough questions, like: “What things are morally good or bad?” or “What sort of person should I be?” Healthy adults can be blamed for these failures and on this weak reading of Concept 2*, are vicious.23   22  Concept 2 claims that vice requires a blameworthy psychology. But, there are different psychological features for which we can be blamed, including: objectively bad motives; false conceptions of the good; and motives to pursue what one believes to be bad (for its own sake). Since each of these psychological features is slightly different, there will be slightly different varieties of Concepts 2 and 2*. One can have objectively bad motives without being motivated to pursue what one believes to be bad; e.g., when one wants to drink scotch for breakfast (one has an objectively bad motive), but wants to do so because one (falsely) believes it is good. One can also be motivated to pursue what one believes to be bad without having objectively bad motives; e.g., when one’s beliefs about the bad are false.   23  Julia Annas (2011: 103) proposes an analysis of moral vice that is consistent with this weak reading of Concept 2*. She thinks vices are acquired character traits that fail to commit to goodness.

Varieties of Epistemic Vice  65 Arguably, there is no single “real” concept of virtue or vice; arguments to that effect will be unproductive. Both concepts of virtue (1 and 2) identify ways to be an excellent person. One can be excellent either by reliably getting the good, or by possessing good motives or a praiseworthy psychology. Likewise, both concepts of vice (1 and 2) identify qualities that make us worse people. We can be defective either by producing bad effects—e.g., false beliefs—or by having bad motives or blameworthy psychologies— e.g., wanting to believe whatever is easiest. Getting truths and avoiding falsehoods matters, so do our motives in forming beliefs. Some varieties of Concepts 1* and 2* are also plausible. Section 3.3 argues that Aristotle’s analysis of vice is one, but not the only, plausible variety of Concept 2. There are other options open to the Responsibilist that warrant exploration, including Concept 2*.

3.3  Varieties of Concept 2 3.3.1  Aristotle’s Analysis of Vice In Nicomachean Ethics VII, Aristotle famously argues that moral vices are acquired character traits that require blameworthy psychologies. For starters, he thinks that moral vice requires dispositions to have motives and to perform actions, that are in fact bad. In addition, Aristotle contends that moral vice is distinct from akrasia, which also requires motives and actions that are in fact bad. The difference between the two lies in other blameworthy features of our psychologies. Unlike the akratic, the vicious person chooses her acts because she believes (albeit falsely) that they are good. She does so with conviction and without competing motivations or regret. Here, I elucidate Aristotle’s notions of vice and akrasia and contend that vice, so construed, is indeed more blameworthy than akrasia. In Aristotle’s words, “akrasia is not vice . . . for akrasia is contrary to choice while vice is in accordance with choice” (NE.1151a6–7). Similarly, “There is a sort of man . . . whom passion masters so that he does not act according to the right rule, but does not master to the extent of making him ready to believe that he ought to pursue such pleasures without reserve; this is the akratic man” (my emphasis, NE.1151a21–4). What distinction is Aristotle drawing? I’ll follow Aristotle in using the moral vice of self-indulgence to illustrate the distinction. For Aristotle, choice requires rational desire (boulesis). Roughly, one rationally desires, e.g., physical pleasure when one wants it because one believes it is good: when one is motivated to pursue it because one values it. Choice, and thus vice, require having a conception of the good. The morally self-indulgent person and the akratic person perform the very same voluntary actions—e.g., both have scotch for breakfast. But the self-indulgent person chooses to do so: she is motivated to do so because she believes that pleasure is far and away the most important good. The self-indulgent person’s conception of the good life is positively dominated by the value she places on pleasure. When she has scotch for breakfast, she is “led on in accordance with [her] own choice,

66  Heather Battaly thinking that [she] ought always to pursue the present pleasure” (NE.1146b22–3). She is, however, objectively wrong. On Aristotle’s view, she over-values pleasure: her conception of the good is false. Consequently, her motives and actions are objectively bad. She is also unaware of any of this. In Aristotle’s words, “vice is unconscious of itself ” (NE.1150b36). In sum, the morally self-indulgent person has scotch for breakfast because she thinks it is good, and does so without competing motives, regret, or awareness that she is vicious. Her beliefs, motivations, and actions are consistent and integrated with one another. In contrast, though Aristotle’s akratic person does drink scotch for breakfast, he does not choose to do so. Like Aristotle’s vicious person, his akratic person has a conception of the good. But, unlike the vicious, the akratic’s conception of the good is true; he places appropriate value on pleasure. Accordingly, the akratic believes that it is bad to have scotch for breakfast and rationally desires not drinking the scotch. Nevertheless, he voluntarily drinks the scotch, despite his rational desires, because of appetite. His acts are voluntary because he has options and knows exactly what he is doing. Unlike the vicious person, the akratic person is conflicted and aware of it. He has competing motivations: a rational desire not to drink the scotch and an appetite for scotch. Because he is aware that his action conflicts with his rational desire, he feels regret. In sum, on Aristotle’s view, vice requires one to: (a) consistently perform acts that are in fact bad; (b) have an (objectively false) conception of the good; (c) be motivated to perform said acts because one (falsely) believes that they are good; (d) lack competing motivations and beliefs; and (e) feel no regret. In contrast, Aristotle’s notion of akrasia requires condition (a), while denying (b) through (e). Aristotle’s akratic performs acts that are in fact bad. But, he has a true conception of the good and is motivated not to perform said acts because he (correctly) believes that they are bad. He also has competing motivations, which bring about the acts, and feels regret.24 Aristotle’s analysis of vice shows us why vice is distinct from, and worse than, akrasia. Both vice and akrasia express what the agent cares about and values. The vicious agent values things that are in fact bad—her conception of the good is false. Her motives are also in fact bad—the motive to drink scotch for breakfast is objectively bad. Since her psychology is integrated, she has no competing motives that are objectively good. In contrast, the akratic agent has two points in his favor. He values things that are in fact good—his conception of the good is true. He also has motives that are objectively good—he rationally desires not drinking scotch for breakfast. He just can’t get himself to act in accordance with those beliefs and motives. He voluntarily performs bad acts due to his competing motives. In short, the vicious person is worse than the akratic

  24  We can construe akrasia more broadly, as voluntarily acting contrary to one’s beliefs about what is good or bad, whether those beliefs are true or false. Aristotle considers a sort of akrasia, according to which the akratic acts contrary to false beliefs about the bad (NE.1146a26–30). This akratic performs acts that are in fact good due to motives that are in fact good (e.g., Huck Finn). For present purposes, it does not matter which of these accounts of akrasia we adopt.

Varieties of Epistemic Vice  67 because the former values things that are in fact bad and only has motives that are in fact bad. For Aristotle, both of these features of the vicious agent’s psychology are blameworthy. Hence, the vicious person is more blameworthy than the akratic. At least the akratic has a true conception of the good and some good motives.25 3.3.1.1  Epistemic Akrasia Aristotle’s analysis of vice is a variety of Concept 2. In the next section, I argue that Responsibilists can adopt an epistemic analogue of Aristotle’s analysis of vice. But that would mean that there is also an epistemic analogue of Aristotelian akrasia. And epistemic akrasia sounds suspicious. After all, one of the primary differences between action and belief is that the former is uncontroversially voluntary but the latter is not. To put such worries to rest, I propose an analysis of epistemic akrasia that is conservative in its scope. It applies to voluntary intellectual actions and to the voluntary use of intellectual processes, but does not apply directly to beliefs. It, too, requires (a), while denying (b)–(e). Accordingly, the epistemic akratic has a correct conception of which intellectual actions and processes are epistemically good or bad. He correctly believes that (say) it is bad to ignore evidence and employ wishful thinking, and is thus motivated to avoid such actions and processes. But he can’t get himself to act in accordance with his rational desires. He voluntarily performs actions and employs processes that he has decried, due to competing epistemic motivations. Since he is conscious of acting contrary to his rational desires, he feels regret. To illustrate, suppose that an agent correctly thinks that she should seek out alternative views when conducting inquiries. She is thus motivated to seek out alternatives and presumably motivated to get true beliefs. And, yet, she consistently passes up opportunities to seek out alternatives because she is also motivated to believe whatever is easiest. Conscious of acting contrary to her rational desires, she feels regret.26 This agent performs akratic intellectual actions. Agents can also use intellectual processes akratically. In Amelie Rorty’s words, a person can voluntarily “follow . . . a pattern of inference that he regards as illicit” (1983: 179). He can opt to employ wishful thinking, knowing that it is unreliable, because he wants to believe what is comfortable. This analysis of epistemic akrasia recognizes that we may only have indirect control over our beliefs, through the intellectual actions and processes we employ. Once we employ those actions and processes, there may be little we can do to prevent the   25  Aristotle argues that vice is worse than akrasia because it is harder to change the vicious person. The akratic person can be changed by habituating his appetites. Aristotle also agrees with the above: the akratic is better than the vicious person because “the best thing in him, the first principle, is preserved” (NE.1151a25); i.e., he knows what is good.   26  Relatedly, suppose that an agent wants to get a true belief about the specific matter at hand; but also wants to end her current inquiry and pursue true beliefs about other matters. Further, suppose that the agent correctly believes that, in the given context, she should not end her current inquiry. (Imagine she is close to an answer.) In the given context, wanting to end her current inquiry and pursue true beliefs about other matters is a bad motive. And, yet, she voluntarily ends her current inquiry; and regrets doing so. Adapted from Owens (2002: 393–4). Owens argues against akrasia of belief, but for akrasia of intellectual action.

68  Heather Battaly formation of belief. Consequently, this analysis of epistemic akrasia should be no more problematic than analyses of moral akrasia. In Christopher Hookway’s words, “If activities can be carried out akratically and inquiries . . . are activities, then . . . [inquiries] can also be carried out akratically” (2001: 179). 3.3.1.2  Applying Aristotle’s Analysis to Epistemic Vice Virtue-responsibilists argue that epistemic virtues are structurally analogous to Aristotelian moral virtues. Does this mean that Responsibilists must endorse an epistemic analogue of Aristotle’s analysis of vice? I argue that Responsibilists can, but need not, endorse Aristotle’s analysis of vice. Aristotle succeeds in identifying one way to be vicious, but his is not the only way. His conditions on vice are jointly sufficient, but some of those conditions are not necessary. What would an epistemic analogue of Aristotelian vice look like? Arguably, the vice of epistemic self-indulgence is well-suited for an Aristotelian analysis. Very roughly, to say that a person is epistemically self-indulgent is to say that he consistently: (ESI1) chooses to desire, consume or enjoy epistemic objects that are in fact inappropriate; or (ESI2) chooses to desire, consume, or enjoy (appropriate or inappropriate) epistemic objects too much or too frequently.27 Epistemic objects include, for instance, true beliefs and false beliefs, about any topic, from Derek Jeter’s batting average to the current state of the world economy. Here, I provide two illustrations of epistemic self-indulgence. The character Mr. Casaubon, from George Eliot’s Middlemarch, is epistemically self-indulgent because he consistently chooses to desire, consume, and enjoy truths that are in fact inappropriate (ESI1).28 Arguably, not all truths are equally appropriate; trivial truths about sports, or the whereabouts of celebrities, are less valuable than truths about science or the world economy. Though it is notoriously difficult to distinguish trivial from non-trivial truths, the truths Casaubon pursues are clearly trivial. Casaubon compiles volumes of notes for a book he plans to write, but never does, The Key to all Mythologies. Of course, there are non-trivial truths about mythology. But the truths Casaubon pursues are not among them. Instead, he pursues “possible arguments to be brought against his . . . view of the Philistine God dragon and other fish-deities” rather than truths about the frescoes of Raphael (Eliot 1984: 191). He is described as “an elaborator of small explanations about as important as the surplus stock of false antiquities kept in a vendor’s back chamber” (1984: 199), and as one who was “lost among small closets and winding stairs” (1984: 192). Evaluating his own research, Casaubon remarks, “I have been led farther than I had foreseen, and various subjects for annotation have presented themselves, which, though I have no direct need of them, I could not pretermit” (1984:  194). Casaubon pursues such inquiries reflectively, and with conviction, thinking they are valuable. He has a conception of the epistemic good, in    

  Battaly (2010 and forthcoming: Chapter 4).   Eliot (1984). I assume Casaubon’s inquiries produced true (rather than false) beliefs.

27

28

Varieties of Epistemic Vice  69 which his inquiries in mythology play the starring role. (They are more valuable than inquiries about Raphael’s paintings, which he dismisses.) Casaubon thereby satisfies Aristotle’s conditions for vice. He (a)  consistently desires and consumes epistemic objects that are in fact inappropriate. He (b) has a (false) conception of the epistemic good and (c) is motivated to desire and consume such epistemic objects because he (falsely) believes that they are good. Moreover, he is entirely without regret, or awareness of his vice, and even makes Dorothea edit his notes (1984: 457–8).29 Within epistemology, skeptics about knowledge and justification are arguably epistemically self-indulgent. Such skeptics choose to avoid both actual falsehoods and possible falsehoods. As epistemologists, they have conceptions of the epistemic good, which they adamantly defend. They are epistemically self-indulgent because their conceptions are false: they over-value the importance of avoiding falsehoods and under-value the importance of getting truths. They care too much about avoiding falsehoods (ESI2). Of course, skepticism is sometimes appropriate; e.g., when the stakes are high. But the stakes are not always high. Accordingly, skeptics satisfy Aristotle’s conditions for vice. They (a) consistently avoid possible falsehoods even when they shouldn’t. They (b) have (false) conceptions of the epistemic good and (c) are motivated to avoid possible falsehoods because they (falsely) believe that they are doing the right thing. 30 They do so without regret, trying to convince non-skeptics to change our ways. Casaubon and the skeptic are more blameworthy than the epistemic akratic. The akratic can be praised for acquiring a correct conception of the epistemic good and for having motives that are in fact good. Accordingly, the akratic would be aware that her inquiries into mythology were trivial, and that her denials of knowledge were wrong; and would feel regret. One might object that Aristotle’s conditions on vice are not sufficient—that he has not identified a way to be vicious. Suppose an agent consistently performs acts that turn out to be bad and has a conception of the good that turns out to be false. But suppose she simultaneously has good internal reasons for believing that her conception of the good is correct. One might argue that this agent is not vicious, despite satisfying Aristotle’s conditions, since she is internally justified and, thus, blameless.31 To illustrate, imagine that an agent ends up with a false conception of the good, and ends up performing objectively bad acts, due entirely to bad luck. She is not negligent in any way. She has good reasons for her conception of the good, and has acquitted herself admirably in searching for reasons. She has done everything well. She just has the bad   29  There is another way for truths to be inappropriate—by being epistemically harmful, or misleading. Suppose that some of the data in The Bell Curve is true; e.g., suppose that Black-Americans typically score lower on IQ tests than White-Americans. Such truths might still be inappropriate because misleading: though themselves true, they tend to produce a preponderance of false beliefs (e.g., the discrepancy is genetic; Blacks do not have as much innate intelligence, etc.) Agents who consistently choose to consume truths that turn out to be misleading would also be epistemically self-indulgent. I am grateful to Sandy Goldberg and Kate Elgin for this point.   30  See Battaly (2010: 223). Avoiding possible falsehoods is an intellectual action.   31  Thanks to Sarah Wright for this point.

70  Heather Battaly luck of being in a demon-world, or more realistically, of being wrong. Her conception of the good is false due solely to powers beyond her control, not due to any fault of her own. This agent satisfies Aristotle’s conditions (a)–(e), as stated above. But because she is also simultaneously blameless, she is not vicious. This objection highlights an important aspect of Aristotelian vice:  being blameworthy. Aristotle never intended his analysis of vice to include people who produce bad effects due purely to bad luck. He explicitly excludes “brutes” and the mentally ill, who have the bad luck of being naturally bad or diseased, from the categories of vice and akrasia (NE.VII.5).32 Instead, he intends his analysis of vice to include agents like Casaubon, who are blameworthy for producing false conceptions of the good. Casaubon does not over-value his inquiries into mythology due simply to bad luck. He does so due to faults of his own. Likewise, Hitler is blameworthy for producing a false conception of the good because he can be blamed for discounting conflicting evidence, failing to consider alternatives, etc. This demonstrates that we must make explicit another condition of Aristotle’s analysis of vice—one that Aristotle presupposes all along. Roughly, vice requires one to be (f) blameworthy for (a) consistently performing bad acts and/or for (b) having a false conception of the good. Given this newly explicit condition, neither demon-victims nor brutes satisfy Aristotle’s conditions for vice since they cannot be blamed for the bad effects they produce. Neither demon-victims nor brutes are vicious according to Concept 2, though they are vicious according to Concept 1.

3.3.2  Alternative Analyses of Epistemic Vice 3.3.2.1  Motivation to Pursue the Believed Bad Aristotle identifies one way to be vicious. His analysis allows us to count dictators and skeptics, who (falsely and negligently) think that they are doing good, among the vicious. But his is not the only way to be vicious. Indeed, one might think that the sort of vice captured by Aristotle isn’t insidious enough. Though Aristotle’s vicious person consistently performs acts that are in fact bad, she (b) (falsely) thinks such acts are good, and (c) is motivated to perform them for that reason. She pursues what she believes to be good. Arguably, (b) and (c) are not necessary for vice. They are holdovers from Socratic intellectualism. Conditions (b) and (c) prevent ruthless dictators and super-villains, who pursue what they believe to be bad, from counting among the vicious. But, if anyone is vicious, surely such dictators and villains are. Provided that motivational internalism is false and we can and do desire the bad for its own sake, we need an analysis of vice that reflects this. Accordingly, an alternative way to be vicious is to (a) consistently perform acts that are in fact bad, (b’) have an (objectively true) conception of the good and bad, (c’) be motivated to perform said acts because one (correctly) believes that  

  Thanks to Gregory Velazco y Trianosky for this point.

32

Varieties of Epistemic Vice  71 they are bad, (d) lack competing motivations and beliefs,33 (e) feel no regret, and (f ’) be blameworthy for (a) consistently performing bad acts and/or for (c’) being motivated to perform said acts because one (correctly) believes that they are bad. Cruelty is a moral vice that lends itself to such analysis. Michael Stocker (1979) has argued that we can, and sometimes do, pursue the harm of others simply because we know it is bad and want what is bad. We do not have ulterior motives to get the good. In a similar vein, Robert Adams contends that “[p]‌eople sometimes hate other people, or themselves, and want bad things to happen to those they hate, not so that good things may come of the bad, but just because they hate” (2006: 40). Accordingly, we can conceive of the cruel dictator or villain as one who (a) consistently harms others, thus performing acts that are in fact bad, (b’) knows that it is bad to harm others, and (c’) wants to harm others simply because it is bad, without any (d) competing motives or (e) regret. Moreover, he is (f ’) blameworthy for wanting and doing what is bad. Presumably, the cruel dictator is conscious of his vice. Since his psychology is thoroughly integrated, he performs his acts with conviction and defiance, perhaps even revelry; not regret. This analysis of cruelty captures just how bad cruelty is. The cruel person knowingly pursues the pain of others for its own sake, reveling in his own vice. Adams argues for a similar analysis of cruelty. On his view, cruelty is a vice of opposition, whereby the vicious agent opposes goods and endorses evils for their own sake (2006: 41). Jason Baehr has applied Adams’s view to Virtue-responsibilism, arguing that the vice of epistemic malevolence entails opposing the epistemic good for its own sake. According to Baehr, epistemic malevolence can take an impersonal form of opposition to knowledge as such or a personal form of opposition to a specific person’s or group’s share in knowledge or “to her epistemic well-being” (2010: 203). On Baehr’s view, the malevolent person’s opposition to the epistemic good must be deep; i.e., it must reflect his “fundamental cares and concerns” (2010: 190). In short, the epistemically malevolent person must have a conception of the epistemic good and bad. Baehr thinks that some epistemically malevolent agents will block the knowledge of others simply because they oppose what is epistemically good, and not for any ulterior reason (2010: 192). As Baehr points out, the evil demon may well be such an agent (2010: 207). Clearly, the demon prevents his victims from getting true beliefs, thus satisfying (a) above. If the demon does so because he knows that blocking true beliefs is epistemically bad, and wants to do what is epistemically bad, then the demon also satisfies (b’)–(f ’) above. He is an epistemic villain. It is important to note that Baehr employs an (epistemically) internalist reading of “opposition to the good.” The analysis of vice according to (a)–(f ’) above is externalist in comparison with Baehr’s view. Baehr argues that an agent is malevolent if and only if she is opposed to what she has good reason to regard as good. On his view, being   33  Contra motivational internalism, this vicious agent is not motivated to pursue what she (correctly) believes to be good.

72  Heather Battaly opposed to what is in fact good is neither necessary nor sufficient for malevolence. So, for Baehr, the malevolent agent need not (a) consistently perform acts that are in fact bad; nor need she (b’) have a correct conception of the epistemic good and bad. What makes her vicious is her motivation: she is motivated to oppose what she has good reason to think is good, and to pursue what she has good reason to think is bad. So, she satisfies an internalist version of (c’). She is also blameworthy for these motives, satisfying a corresponding version of (f ’). Baehr arguably succeeds in identifying another way to be vicious. Suppose agent S does as Baehr suggests: she opposes what she has good reason to believe is (epistemically) good. Imagine that S has good reasons for thinking that x is good and has acquitted herself admirably in searching for reasons. Due to no fault of her own, S’s belief that “x is good” turns out to be false—x is in fact epistemically bad. Consequently, S ends up opposing something (x) that is in fact bad and being motivated to pursue something (~x) that is in fact good. Agents like S cannot be blamed for their false conceptions of the (epistemic) good. But they are still internally defective since their motives are, indeed, blameworthy: S wants what she thinks is bad and opposes what she thinks is good. In Baehr’s words, she “intends to be undermining a genuine good and firmly believes . . . that she is doing so . . . it is completely a matter of luck, relative to [her] perspective . . . that she is not opposed to a genuine good” (2010: 196–7, his emphasis). It is sheer accident that S ends up pursuing what is in fact good. So, even though Baehr’s agent is not blameworthy for having a false conception of epistemic value, she is blameworthy for wanting what she thinks is bad. She wants what she thinks is bad due to faults of her own, not bad luck, thus satisfying a version of (f ’). This is one way to be internally defective. Suppose that Baehr’s conditions are sufficient for epistemic malevolence and for vice in general. Still, they are not necessary. As argued in 3.3.1.2, there is a different way to be defective, which accounts for the vices of agents like Hitler. Such agents are defective because they consistently perform acts that are in fact bad, while falsely and negligently believing that they are pursuing the good. Unlike agent S, they can be faulted for their false conceptions of the good. But, arguably, there is nothing wrong with Hitler’s motives relative to his own perspective. Unlike agent S, agents like Hitler (arguably) have motives to pursue what they think is good. Thus, they do not satisfy Baehr’s necessary condition—they are not motivated to pursue what they have good reason to think is bad—and yet they are clearly vicious, and clearly epistemically malevolent. Hitler destroyed the (epistemic and overall) well-being of millions because he negligently and wrongly thought that doing so was good. Embracing pluralism about vice allows us to count agents like S (who satisfies internalist versions of (c’) and (f ’)), agents like the evil demon (who satisfies (a)–(f ’)), and agents like Hitler (who satisfies (a)–(f)) as vicious and epistemically malevolent. One might worry that Baehr’s vice of epistemic malevolence is not uniquely epistemic. Rather, it appears to be a variety of moral malevolence that is applied to epistemic objects, like knowledge. So, are there uniquely epistemic vices, whereby

Varieties of Epistemic Vice  73 agents are motivated to pursue the epistemically bad? This warrants exploration. For instance, are there “epistemic rebels” who knowingly and defiantly perform bad intellectual actions (e.g., embrace inconsistent claims) and use unreliable processes (e.g., invalid arguments), without ulterior motives? Perhaps, there are radical versions of post-modernism which view themselves this way. If not, butchered versions of post-modernism, adopted by angst-ridden students, may fit the bill. As an alternative approach, dogmatism is a uniquely epistemic vice, if anything is. Though dogmatism lends itself to an Aristotelian analysis, are there versions of it in which agents knowingly and defiantly ignore alternatives because doing so is epistemically bad? Finally, even if we ultimately decide that there are no uniquely epistemic vices whereby agents knowingly pursue bad intellectual activities, it would still be important to know why that was the case. Why would epistemic vices be different from moral vices in this respect? 3.3.2.2  Vice Without a Conception of Value Both of the aforementioned analyses of Concept 2, in 3.3.1 and 3.3.2.1, make vice difficult to attain. In order to be vicious, one must have a conception of what is good or bad. One must have evaluative beliefs about the matter. (Condition (b) claims that the conception must be true; (b’), that it must be false.) But isn’t vice easier to attain than this? What if one never bothers to develop a conception of the good or bad? Consider, for instance, Roger Sterling, a character on the television show Mad Men. Sterling consistently performs the same acts that a morally self-indulgent person would perform—he consumes alcohol and sex excessively—but he appears not to care about morality at all. One of the most disturbing features of Sterling’s character is his utter lack of a conception of what is good or bad. He does not seem to pursue acts either because he thinks they are good or because he thinks they are bad. He simply acts on his preferences, without bothering to evaluate those preferences. He just doesn’t care about morality. Does this mean Sterling isn’t morally vicious? This seems too easy. Arguably, one can be vicious even when one fails to have a conception of what is good or bad, provided that this failure is itself blameworthy. In Sterling’s case, as in the case of healthy adults, we can assume that the failure is blameworthy. Sterling is clearly capable of reflecting on and evaluating his preferences, and of asking tough questions about morality. Those who fail to ask tough questions are negligent. In Terence Irwin’s words, “Vice differs from virtue because the virtuous person answers a question that the vicious person does not even ask. The virtuous person has asked what sort of person he ought to be . . .” (2001: 92). One way to be vicious is to never ask such questions. Instead, one simply acts on one’s desires without ever evaluating them.34 The moral vices of laziness, apathy, and conformity are likely to fit this description.   34  In NE.III.5, Aristotle flirts with an analysis along these lines: “Perhaps a man is the kind of man not to take care. Still they are themselves by their slack lives responsible for becoming men of that kind. . .” (NE.1114b4). On Sterling, see Battaly (forthcoming: Chapter 4).

74  Heather Battaly Accordingly, the vicious agent (a) consistently performs acts that are in fact bad, while (~b) failing to have a conception of the good or bad, and thus (~c) failing to have motivations that are informed by a conception of the good or bad. She is also (f '') blameworthy for (~b) failing to have a conception of value and/or for (~c) failing to have motivations that are informed by a conception of value, and (e) feels no regret. Regret would indicate that she did have a conception of the good or bad. With respect to (d), the agent has no evaluative desires or beliefs to be integrated with her preferences. That said, her preferences themselves need not be entirely integrated with one another. Still, they must be largely integrated, if she is to consistently perform acts that are in fact bad. Here, too, vice is a stable, acquired character trait. It is marked by dispositions of action and expresses who one is as a person. Specifically, it expresses one’s failure to care about the good and one’s consequent failure to have a conception of the good. Thus, it is a variety of Concept 2*. This variety of Concept 2* warrants exploration by Virtue-responsibilists. Epistemic vices of this sort may be even more prevalent than their moral counterparts. Failing to develop a conception of epistemic value may turn out to be fairly common, but nonetheless blameworthy. For starters, consider healthy adults who constantly and unreflectively consume celebrity and sports trivia, or rely on the horoscopes, or employ fallacies. They perform intellectual actions that are in fact bad, satisfying (a). More importantly, the best explanation of their actions may be their utter failure to ask questions about epistemic value, satisfying (~b) and (f ’’). They may have failed to consider whether some processes, or truths, are more epistemically valuable than others. They may not have cared enough about truth to ask questions like “What sort of thinker should I be?” or “Which processes should I employ?” Their failures are no less blameworthy than the failures of their moral counterparts. Like Roger Sterling, they act in accordance with their (epistemic) preferences, without ever bothering to evaluate those preferences. If epistemic laziness, epistemic apathy, and epistemic conformity are vices, then they are likely to be vices of this kind. Finally, consider the recent debate over the value of the Humanities. It has been argued, both on and off university campuses, that the Humanities are not valuable because they do not contribute to the economy. That premise, however misguided, is not the most frustrating aspect of the debate. Its key frustration is that proponents of the above argument never bother to consider whether truths in the Humanities are epistemically valuable, perhaps even more epistemically valuable than other truths. This thought never seems to cross their minds.35 So, perhaps, they have the best claim to being the Roger Sterlings of the epistemic world. Briefly, these alternatives to Aristotle’s analysis of vice are still distinct from, and worse than, akrasia. The epistemically malevolent agent and the akratic agent both have true conceptions of the epistemic good and bad. But, the epistemically malevolent

 

35

  See Battaly (2013).

Varieties of Epistemic Vice  75 agent defiantly desires the epistemic bad for its own sake and does so without competing motives or regret. In contrast, at least the akratic wants to do what he knows is epistemically good and feels regret when he fails. Finally, the epistemically lazy agent negligently fails to have any conception of what is epistemically good or bad. She does not care enough to evaluate her epistemic processes or actions. In contrast, at least the akratic has asked and correctly answered evaluative questions about her processes and actions. She cares enough to have developed a conception of epistemic value, even if she cannot get herself to act in accordance with it. In sum, I  have argued that there are several viable varieties of epistemic vice. Reliabilist vices are best analyzed as qualities that produce bad epistemic ends or effects. Responsibilist vices are best analyzed as qualities that require bad epistemic motives, or other blameworthy psychological features. Responsibilists can, but need not, adopt an epistemic analogue of Aristotle’s analysis of vice. I have sketched two alternatives to Aristotle’s analysis that warrant further exploration.36

References Adams, R. M. (2006). A Theory of Virtue. Oxford: Oxford University Press. Annas, J. (2011). Intelligent Virtue. Oxford: Oxford University Press. Anscombe, G. E. M. (1958). “Modern Moral Philosophy,” Philosophy 33(124): 1–19. Aristotle. (1992). Nicomachean Ethics, D. Ross (Trans.). Oxford: Oxford University Press. Baehr, J. (2010). “Epistemic Malevolence,” in H. Battaly (Ed.), Virtue and Vice, Moral and Epistemic. Malden, MA: Wiley-Blackwell, 189–213. Battaly, H. (2010). “Epistemic Self-Indulgence,” in H. Battaly (Ed.), Virtue and Vice, Moral and Epistemic. Malden, MA: Wiley-Blackwell, 215–35. Battaly, H. (2013). “Detecting Epistemic Vice in Higher Education Policy,” Journal of Philosophy of Education 47(2), 263–80. Battaly, H. (forthcoming). Virtue. Cambridge: Polity Press. Driver, J. (2001). Uneasy Virtue. Cambridge: Cambridge University Press. Ekman, P. (2009). Telling Lies, 3rd edn. New York: W.W. Norton. Eliot, G. (1984). Middlemarch. New York: Modern Library. Fricker, M. (2007). Epistemic Injustice. Oxford: Oxford University Press, 2007. Goldman, A. I. (Ed.) (1992). “Epistemic Folkways and Scientific Epistemology,” in Liaisons. Cambridge, MA: The MIT Press, 155–75. Hookway, C. (2001). “Epistemic Akrasia and Epistemic Virtue,” in A. Fairweather and L. Zagzebski (Eds.), Virtue Epistemology. Oxford: Oxford University Press, 178–99. Hursthouse, R. (1999). On Virtue Ethics. Oxford: Oxford University Press. Irwin, T. (2001). “Vice and Reason,” The Journal of Ethics 5(1): 73–97.

  36  I am grateful to Jon Matheson for comments. I would also like to thank the participants of the University of North Florida’s conference on the Ethics of Belief, especially Jon Matheson, Rico Vitz, Mitch Haney, Kate Elgin, Heidi Grasswick, Sandy Goldberg, Ernest Sosa, and Sarah Wright; and the philosophers at Cal State Northridge, especially Gregory Velazco y Trianosky.

76  Heather Battaly Montmarquet, J. A. (1993). Epistemic Virtue and Doxastic Responsibility. Lanham, MD: Rowman & Littlefield. Owens, D. (2002). “Epistemic Akrasia,” The Monist 85(3): 381–97. Plato. (1992). Republic, G. M. A. Grube (Trans.). Indianapolis: Hackett Publishing. Rorty, A. (1983). “Akratic Believers,” American Philosophical Quarterly 20(2): 175–83. Slote, M. (2001). Morals from Motives. Oxford: Oxford University Press. Sosa, E. (1980). “The Raft and the Pyramid,” Midwest Studies in Philosophy 5: 3–25. Sosa, E. (1991). Knowledge in Perspective. New York: Cambridge University Press. Sosa, E. (2007). A Virtue Epistemology. Oxford: Oxford University Press. Sosa, E. (2011). Knowing Full Well. Princeton: Princeton University Press. Stocker, M. (1979). “Desiring the Bad,” Journal of Philosophy 76(12): 738–53. Swank, C. (2000). “Epistemic Vice,” in Guy Axtell (Ed.), Knowledge, Belief, and Character. Lanham, MD: Rowman & Littlefield, 195–204. Zagzebski, L. T. (1996). Virtues of the Mind. Cambridge: Cambridge University Press.

4 Knowledge and Time: Kripke’s Dogmatism Paradox and the Ethics of Belief * Ernest Sosa

How is knowledge affected with the passage of time? We shall consider some surprising ways. First we take up a paradox of dogmatism in Saul Kripke’s collected papers, Philosophical Troubles. Kripke first presented his paradox decades ago at Cambridge University. Soon afterwards, Gilbert Harman in his book Thought drew from that talk a certain puzzle, to which he then offered a solution. But now Kripke observes that this was not really the puzzle of main interest to him. The two are closely related, but not quite the same. In section 4.1 we first take up Kripke’s version, and then Harman’s version, along with his solution. Such problems of dogmatism are in a family with other issues of broad import in epistemology. Pragmatic encroachment, for example, is discussed in section 4.1, which considers how present knowledge can properly affect future inquiry, how it can affect whether our minds should be open, ajar, or closed. Section 4.2 takes up the divides between externalism and internalism, and between reliabilism and evidentialism. It also considers how we can be torn by opposing pulls. Past evidence can pull us diachronically to believe, even while the present reasons synchronically pull us against believing. What then is one to do all epistemic things considered? We find reasons to distinguish first-order animal knowledge from a reflective knowledge that relies mainly on our second-order perspective. Reasons emerge why reflective knowledge deserves pride of place. Finally, section 4.3 highlights two morals: first, the importance of epistemic negligence; and, second, how the fragility of memory bears on our need for an open mind.   *  Large portions of earlier versions of this chapter were presented at the University of North Florida conference, and at a conference to celebrate the publication of Saul Kripke’s Philosophical Troubles, CUNY, September 15–16, 2011, and the chapter has benefited from both occasions.

78 Ernest Sosa

4.1  Kripke’s Dogmatism Paradox As night falls on your cabin in the woods you do well to shut the door. Who knows what creature might come through that doorway to do you harm? Especially should you do so if you know that creatures would enter, bringing nothing that matters except possibly harm. On that basis, you should form the intention to shut that door, and you should act on your intention. Kripke argues compellingly for his paradox by analogy with that sort of example. Once you know that p, you can deduce, with extreme competence and simplicity, that any evidence contrary to p would be misleading, whereas positive evidence would probably do you little good. After all, by hypothesis you already know that p! Given this, you should close your mind to any new potential evidence on the question whether p. If positive, the evidence will do little for you; if negative, it will harmfully pull you away from the truth, and may even cost you the knowledge that you have. This is a powerful argument for turning dogmatically close-minded once you know that p. And we do often act on similar reasoning. We may ignore the National Inquirer as we stand in line at a supermarket counter. Reading that tabloid will do us no good, we reason, so why spend the time? Here I assume what is of interest to be exclusively epistemological. One aims only to acquire interesting or important information. If one wants a good laugh, that’s another matter. In ignoring the tabloid we properly disregard a source. Alternatively, we may focus not on the source but on the question. Thus, we take ourselves to know that astrology is ridiculously false, and refuse to waste our time listening to any advocate. Any defense of it is not worth hearing out. Most of us would endorse that strategy to protect our time and energy. You add up a restaurant bill, for another example, and figure out what each diner owes, which gives you reason to turn a deaf ear, refusing to double-check. Indeed, having attained knowledge one should never try to make sure, since that would only risk being misled by counter-evidence. The replication of experiments after one knows also comes under a cloud. Even when we barely know, further replication might just mislead. It should not be done, some will say; indeed, the scientific community should institute rules against such replication. Still we do not adopt any general policy of dogmatically closing our minds on the many questions to which we know the answers. We do not try to block incoming evidence that might have bearing. And we would seem deplorably close-minded if we did try. Even though universal dogmatism is thus unacceptable, one does still have prima facie reason to close one’s mind, as explained earlier. If the reason is only prima facie, however, one may still prefer to continue inquiry. One may want to make sure, for

Knowledge and Time  79 example, just for the fuller satisfaction of curiosity, or because it’s epistemically better to know with certainty. One might thus have reason to double-check and to seek further evidence, reason stronger than the prima facie reason to close one’s mind. Our earlier rejection of dogmatic mind-closure derived from understanding this as peremptory. Even if peremptory closing is unacceptable, however, knowledge still plausibly provides prima facie reason to close our minds. What distinguishes cases where such prima facie reason graduates into ultima facie reason? When are we right all things considered to close our minds? This question is closely related to an issue of current interest in epistemology, that of pragmatic encroachment. According to a thesis of pragmatic encroachment, whether S knows that p constitutively depends on what is at stake. Put aside contextualism, which raises its own issues not for this occasion. We focus rather on a thesis of subject-sensitive encroachment, according to which S’s knowledge turns on what is at stake for S. Whether S knows or not is said to involve how much depends for S on whether he gets it right. As the stakes rise, so does the standing required of one’s belief if it is to constitute knowledge. We can leave open what sort of standing is involved, whether it be degree of reliability or degree of evidence-based justification, or something else with degrees. The point is that in order to know you need higher standing as the stakes rise. Whether you are right to close your mind might thus depend on what is at stake. The gossip that fills the National Inquirer, for example, is of slight importance for the ordinary reader. So it is easy to close one’s mind on such subject matter. Little evidence is required for the conclusion of inquiry. The question may even matter so little that no further attention would be justified. Consider more important matters affecting someone’s health, however, or their livelihood, or one’s good performance on the job, or in the marketplace, or in reaching a scientific or technological conclusion with important payoff. On any such question, one should rarely close one’s mind against any further evidence that might have bearing. The matter is too important for such a blind or blinkered attitude. The advocate of pragmatic encroachment denies that a subject really knows unless he can properly act on that knowledge to attain his objectives. We do aim to avoid being misled. If we really know that p, therefore, this grounds a policy of avoiding any further ostensible evidence bearing on the question whether p. Otherwise the supposed knowledge would be merely ostensible, not real. Real knowledge is actionable knowledge. Or so we are told. Opponents of encroachment see a gap between knowing something and the propriety of acting on that knowledge. If nothing too important hangs in the balance, we can act on our knowledge without further ado. When the stakes rise, however, we may need to make sure. And this need to make sure can justify the search for further evidence. In this way, even after one attains ordinary knowledge, one might still properly inquire, despite the risk of being misled by misleading evidence.

80 Ernest Sosa Beliefs can vary in degree of confidence and in epistemic quality even when they all constitute knowledge. You might therefore reasonably close your mind when you know for certain, even though you should not do so when you barely know. Concerning how open-minded and how inquisitive to be on a question even when one knows the answer, two sorts of factors are important:  First, how high are the stakes? Second, how certain are we, and how properly certain? Kripke’s paradox concerns a prospective intention to close one’s mind. Harman’s version, by contrast, concerns an antecedent belief: namely, that any further evidence will be misleading if negative. Once you attain knowledge, you virtually know that any contrary evidence will be misleading and is best ignored. When negative evidence does come forth at some later time, should you then ignore it in accordance with what you already know? Harman resolves this problem by noting that knowledge can be lost with change of evidence. Later total evidence may deny you the knowledge that until then you enjoyed. The two problems of dogmatism are closely related, since the Harman belief might sustain the Kripke intention. If at a given time you know that p, you might form and implement an intention to close inquiry, regardless of how much is at stake. Even with the belief-related problem resolved as it is by Harman, the intention-related problem might still puzzle us. Why is it not appropriate to base on our knowledge an intention to avoid epistemic harm by closing our minds? Harman’s insight retains some bearing, nevertheless, since a properly formed intention is itself subject to modification with the inflow of evidence! So, even if at an earlier time you properly form that intention, this does not entail that at a later time you can still sustain it. After all, you may no longer be properly certain that p. Although you intended to close your mind, some evidence may have barged in anyhow. What is more, rising stakes may now require greater confidence. So, you may need to seek further evidence, rather than closing your mind. Nevertheless, resolving the belief-related puzzle does not fully resolve the intention-related puzzle, which still bears further thought. So far we have considered the bearing of pragmatic concerns in determining when we should continue inquiry despite already knowing the answer. A further question remains, as to when one is justified epistemically in closing one’s mind, given that one already knows, and properly takes oneself to know. Even this question divides. One question now asks what might best deliver the epistemic goods, whether it is closing one’s mind or continuing inquiry. A less consequentialist question asks rather what is epistemically better in itself. What should the knower do with his mind: close it or keep it open? How does one proceed more competently? One would of course be negligent if one should inquire further but fails to do so. Consider the choice whether to affirm, deny, or suspend on a certain question. Even if we put aside which option will bring the greatest intellectual gain, one might still wonder how best to respond to the question, with regard to knowledge on that

Knowledge and Time  81 question specifically, with regard to the proper satisfaction of one’s curiosity on that question. Our threefold option thus concerns a given question, say the question whether p. One might affirm that p, one might deny it, or one might suspend judgment. Which is the best choice all things considered, and more specifically which is epistemically the best choice? Even more specifically, one might wonder which choice is best vis-à-vis one’s epistemic attitude to that question, to the question whether p. An example clarifies this distinction. One might act in a way that best promotes acquisition of knowledge by trusting a book guide to sources of knowledge. The book might detail the best experiments to perform, the best places to find hidden documents, the best spots for archaeological digs, etc. That book might still fail to deserve our trust, however; the best of reasons might strongly counsel distrusting it. Believing what that book says is then no way to attain knowledge in so believing, despite how much truth and knowledge such trust might yield over time. Despite how epistemically helpful they may be, beliefs derived from the book are not themselves justified or competent in the way required for constituting knowledge. Let us linger over this status of justification or competence. I mean the normative status that a belief must itself have in order to constitute knowledge. Suppose we close our mind on a certain belief. What effect might this have on the epistemic standing of that belief? Can we close our mind in such a way that we thereby lose our knowledge? Might we lose it through a kind of diachronic epistemic negligence? Even if our belief was earlier competent, its status might change as time goes by and our mind remains closed. Sooner or later it might cease to qualify as a competently held belief, since it now remains in place only through negligence. Surely that can happen. Consider the overall epistemic competence that can holistically yield knowledge. If we are to give negligence its proper weight, we must include in such overall epistemic competence more than just how probably true our belief is if based on the total evidence. Overall epistemic competence must also include the proper conduct of inquiry. Negligent belief would not count as fully competent and would not constitute true knowledge. These results pose a threat to simple process reliabilism. Suppose we grant to the avoidance of negligence its proper status. What determines the epistemic justification (competence) of a belief seems then to surpass the reliability of the process that produces that belief. If so, the following questions will need attention. First, is epistemic negligence a causal factor in the formation of a belief? If not, how can it still bear on the relevant reliability of that process? Second, suppose a believer to be guilty of epistemic negligence. This would be for him to fail to do certain things. Can such a failure have positive causal bearing on the belief formed, so as to affect the reliability of that process?1

  1  Process reliabilism is discussed more fully in Sosa (forthcoming).

82 Ernest Sosa

4.2  Animal Competence and Reflective Justification: The Importance of Synchronism Plenty of beliefs are of course clearly based on conscious reasoning from given evidence, and stored through retentive memory. Suppose initial uptake of the given that is perfectly competent and apt, through introspection or perception. Suppose the resulting belief to be competent, and aptly stored. This is all compatible with the believer’s forgetting how he initially acquired that belief. One might later be able to say little more than “I just remember.” What then is the later epistemic standing of that belief? About this we now face a puzzle. We are focused on a time late in the life of the belief. No one can now detail how well that belief was first acquired or retained over time. Concerning its specific subject matter, moreover, the evidence now available is deplorably slight. If the belief must now rest on the evidence presently available, it may no longer count as epistemically justified. How competently can the believer now retain his belief? The belief ’s present epistemic standing must be assessed in the light of how good the subject’s memory is for that sort of belief. Unfortunately, the competence involved in such second-order assessment might provide far less by way of reliability or justification than the competence that yields the belief on the first order. Such first-order competence will often combine excellent initial acquisition with excellent later retention, excellent perception with excellent memory. What then is the believer to do as time passes? Should confidence dwindle in tandem with reduced qualification to endorse first-order belief? Consider the steady decay of the information required for endorsement. Compatibly with such second-order weakening, retentive memory can remain strong indeed, so as to ensure that the retained belief is very probably true, given the excellent perceptual competence that produced it, and the excellent retentive memory that kept it securely stored. That belief may thus constitute first-order, animal knowledge of the highest quality. In many cases what decays over time is only the reflective, second-order perspective. Here is an example. At noon on a certain date you are mistreated as a child. You know extremely well that it is noon on that date. You store that belief for months, even years. Maybe you retain it through excellent memory. In general people would not remember so well. Maybe in general your own memory does not work so well. But it does in this case, on this sort of subject matter. That event stands out in your mind, and your memory of it is outstanding. The perception-plus-memory manifest in your continuing belief is of the highest quality. When perception and memory work as they do in this case, they are extremely reliable. Compatibly with that, your second-order competence can decay. Just based on common sense, you may come to doubt that your memory of that event is as good as in fact it is. You may even learn of the accumulating evidence that memory of such mistreatment is much less reliable than common sense had supposed. Human beings in general do not recollect as reliably as had been thought, not even on such important matters, perhaps especially not

Knowledge and Time  83 on such subject matter. By hypothesis, however, your memory is in this case extremely reliable. That is one example of the phenomenon I wish to highlight. But we need not suppose our protagonist to have abnormal powers. Another sort of example would involve just normal human perception and memory. In combination these might lead reliably to a correct present belief. Nevertheless, the believer may now be unable to detail how he acquired and retained his belief through the operation of his excellent faculties. Moreover, the puzzle does not arise merely from a clash between externalist reliabilism and internalist evidentialism. What is crucially important is rather the clash between two distinguishable epistemic statuses that a belief can have. First, there is the status that a present belief gains diachronically through the subject’s thinking and memory over time, no matter how internal such thinking and memory may be over time, and no matter how internal the subject’s initial data may be. Second, there is the status that a present belief gains at a given moment synchronically through the support of reasons present to the thinker’s consciousness at that very moment. So, the problem transcends the divides between externalism and internalism, and between evidentialism and reliabilism. Now a further question must be faced. Suppose we have an instance of such disparity between animal quality and reflective quality. This is a divergence between first, the high status a belief derives diachronically from a retention-involving first-order competence, and second, the much lower status that the same belief might have synchronically at the given time, due to the diminished epistemic quality of the believer’s relevant second-order competence, either because this competence is much less reliable or because in any case it provides less by way of epistemic justification. Having been mistreated as a child, later in life you may doubt your competence to remember despite remembering perfectly well. For another example, consider arithmetical calculation. You may doubt your ability to perform a complex addition without flaw despite having performed it flawlessly. Suppose your reflective capacity to endorse a given first-order belief to be much diminished. Your endorsement from the second order depends for its epistemic quality on the second-order competence exercised. What about your judgment on the first order? Here’s the question on the first order: What should you affirm within the privacy of your own thought, and what should you assert to others? That is the sort of judgment required for conscious reasoning, practical or theoretical, and for the proper conveyance of information when one speaks in one’s own person. Should your first-order judgment be assessed by the quality of your first-order retention-involving animal competence? Or should it be assessed rather through synchronic exercise of your second-order competence? These two ways of assessing a first-order judgment might differ dramatically, since the two competences can differ greatly in quality. Moreover, it is not just the assessment of a first-order judgment that depends on whether we invoke the first-order animal competence or, alternatively, the second-order reflective competence. There is also this question:  Which perspective

84 Ernest Sosa should have priority in your own continued believing and judging? Should you trust your excellent first-order competence, or should you trump that competence once your belief is under present scrutiny, with the inevitable bearing of the second-order considerations? Should you now decide whether to trump based on all the reasons presently available to you for conscious consideration? Suppose we give priority to the reasons presently available. This in effect recognizes the importance of a kind of reflective knowledge, involving a second-order judgment (or disposition to so judge). Such judgment itself depends for its standing on the quality of the competence that it manifests. Reflective knowledge will thus have a particularly important role. It will enable your conscious reasoning, practical and theoretical, and will sustain your place in the community as bearer of information transmissible through testimony. Such reflective knowledge is constituted by an act of judgment (or disposition to judge), indeed by two: one on the second order, and one on the first order. And these are the judgments that figure in our conscious reasoning, and in our sincere assertion when we inform others. We have been considering this question: What should determine one’s judgment and its quality at the present moment? Is it diachronic competence, even if its initial inputs have long receded from view, or is it rather the reasons available and operative in the present time slice? If we opt for the present-time-slice, this upgrades the second-order perspective, simply because synchronic reasons for stored beliefs are now to be found mainly within that perspective. Justified judgment at t will normally involve your second-order competence to assess your first-order competence to form and sustain beliefs. Your original basis recedes with time, which forces you to rely more and more on your epistemic self-trust. Suppose that your first-order belief is put in doubt, either through overt disagreement, or through a challenge to your relevant competence. In responding to such doubt, you will need to defend your competence. Your first-order belief must now be defended from that second-order perspective, and it may turn out to be more or less defensible from that perspective. It is hard to deny, in any case, that how properly one proceeds epistemically in endorsing (or not) a first-order belief is determined by the reasons consciously accessible at the time of reflection. It would seem deplorably stubborn to sustain and endorse a belief in the teeth of total available evidence strongly against it. And this is so even if that belief is correct through diachronic first-order competence that is supremely reliable.2 A belief that is thus apt through diachronic competence can still fall short. It falls short if it turns out not to be endorsable synchronically through the balance of available reasons. And this failure must bear on how the subject proceeds on the first order. He cannot properly judge by relying blindly on his stored belief and on the diachronic   2  In keeping with our irenic humor, we can make room for competences that are justification-reliable, as well as those that are truth-reliable. Our argument can thus be as relevant to the diachronic internalist evidentialist as to the diachronic externalist reliabilist.

Knowledge and Time  85 competence that sustains it. As our subject consciously considers his options, he must now prefer suspending to affirming. He may even need to contradict in synchronic judgment his diachronically sustained belief. Only rationally competent synchronic endorsement can now lift one’s belief to the higher level of reflective knowledge. Once a belief is under scrutiny, only such reflective knowledge qualifies as a proper basis for conscious reasoning, practical or theoretical. And, finally, only such knowledge then sustains proper assertion, enabling the subject to fulfill his communal role as purveyor of information.

4.3  Knowledge, Justification, and the Problems of Dogmatism If you already know the answer to a given question, how weighty is the prima facie reason that this gives you to close your mind? Section 4.1 considered the importance of avoiding negligence for retaining your knowledge. In order to keep one’s knowledge, often one must not ignore available evidence. Suppose that someone you trust tells you he will deposit what he owes you in your bank account by noon. Suppose his testimony and reliability to be so trustworthy that in the afternoon you know the money to be in the account. Might you now write an important check on the basis of that knowledge? Might you save some time by not verifying online? Suppose next you acquire some evidence that your friend is not so trustworthy. Someone tells you that he has early Alzheimer’s, and someone else reports that he has once let her down. Suppose also that these reports do not remove your extremely well-based knowledge immediately upon receipt. What if you do write the important check, while still not bothering to verify? Would it be right to think that even now, later in the afternoon, when you write the check, you do so in the knowledge that the money is in the account? I am not relying here on a contextualist point that our subject would speak incorrectly in saying to his protesting spouse “But I knew the money was there,” that he would speak incorrectly because the spouse’s protest raises the standards for the truth of “S knows,” by highlighting how important it is that the check not bounce. Nor am I  supposing that knowledge is pragmatically subject-sensitive. I  am not supposing that, given how much turns on whether he is right, the subject no longer really knows if he fails to confirm. Nor am I supposing, finally, that the subject’s merely knowing that the money is in the account is not enough to justify his writing that check, that the high stakes require him to make sure, or to make more sure. Of course, any or all of those factors might be operative, and advocates of the corresponding views of knowledge and of how knowledge bears on action might invoke their favorite views in accounting for the incorrectness of the subject’s speech to his spouse. But someone unpersuaded by any of those views might still reject the subject’s claim that he knew the money was in the account.

86 Ernest Sosa Let us put aside here all of those factors and the corresponding epistemological views. There is a further reason to think that the subject speaks incorrectly, namely that his negligence makes it false that he knew. One might accuse him of a kind of negligence in believing that the money is in the account, and in voicing that belief sincerely to his spouse. In the circumstances, he should have confirmed online before sustaining his belief and acting on it. He should have confirmed of course given how easily he could have done so, and how much was at stake. But he “should” have confirmed in any case, even if he had not been writing an important check. Arguably, he was not “epistemically justified” in continuing to believe absent confirmation. How would one argue for this? The argument is not just that the subject is obliged all things considered to stop his normal life and go online. Nor need one argue that he is required to confirm, for the proper conduct of his intellectual life. We might focus rather on the normative status that is constitutively involved in knowledge: that is to say, on the belief ’s “epistemic justification.” In order to remain thus justified, we might contend, our subject needs to confirm. He should confirm in order just simply to keep on believing knowledgeably. He is hence negligent in not doing so, not just practically negligent, but epistemically negligent. We thus recognize a distinctive reason why at that later time our subject “should” verify whether the money was in the account. This distinctive “should” is not a matter of what he should do absolutely all things considered, nor is it even a matter of what he should do all epistemic things considered. Rather it is a matter of what he should do if he is to retain his knowledge that the money is in the account, and even his epistemic justification for so believing.3 Compatible with this, of course, he may have more important things to do than just to retain his knowledge, and he may even have more epistemically or intellectually important things to do. True, he is obliged to confirm if he is to remain sufficiently epistemically justified, so as to continue to know. But from this it does not follow that he is obliged to verify all things considered, nor even that he is obliged to verify all epistemic or intellectual things considered. Here is an analogy that may help to clarify my intended point. A pilot may land his plane at the intended destination after difficult flying through very bad weather. Is the success of that flight creditable to the pilot’s competence? His piloting competence was certainly manifest in how he maneuvered so as to reach the destination and land safely. In some ways then, his competence was manifest. He did have the proper skills seated in his brain and nervous system. He did have the constitutional competence required of a good pilot. He also was in good shape as he piloted that plane to safety. He was

  3  In Sosa (1964), and reprinted in part as ­chapter 1 of Sosa (1991), I suggest that the analysis offered brings to light a “striking fact . . .: besides justification for “believing that . . .” . . . justification for “not believing that . . .” is also involved in the concept of knowledge.” This is then supported by examples of knowledge-precluding epistemic negligence. Intuitive confidence that negligence matters long preceded my advocacy of virtue in preference to process reliabilism.

Knowledge and Time  87 fully alert and sober, for example, and he was in good condition in every other relevant respect. Finally, he was also well situated, which includes that the plane was under his operational control, was in good shape, had enough fuel, and so on. And the storm was not entirely unmanageable. It had to be managed, but that was not beyond the capability of a good pilot. Given all of that, did the flight’s success manifest the competence of the pilot? Well, it did manifest a relevant competence—of constitution, inner condition, and situation— that would crown his attempts with success. The things he did as he flew that plane and landed it safely derived from his complete first-order competence, which included his constitutional skill, his good condition, and his manageable situation. The success of the flight did manifest that threefold complete competence on the first-order. Is the success of the flight not therefore fully creditable to the pilot? This turns out to be unclear. Suppose the plane was small, and the pilot responsible for ensuring that it would be in good condition for the flight. In particular, he needed to make sure that the tank had enough fuel. But he neglected to do so. He took off without checking at all. Now is he still creditable? That is no longer so plausible. Now the flight’s reaching its destination seems not after all fully creditable to that pilot. The pilot did reach his destination, but his doing so is not a fully apt performance on his part. It is not a performance that manifests all of the relevant competence required for it to be fully creditable. In particular, the pilot’s negligence in failing to check the condition of the tank puts a stain on his performance, and makes its success too much a matter of luck rather than competence. And something similar would seem to apply in epistemic domains. Epistemic negligence detracts from epistemic performance, as when we believe in pursuit of truth. Even if it does happen to succeed on the first order by attaining truth, the success of our performance might still fall short simply because of our negligence. If knowledge is a matter of apt intellectual performance in pursuit of truth, therefore, we get the result that negligence can deny us knowledge, or at least knowledge of a certain epistemically desirable level. We are denied fully apt attainment of truth when we attain truth despite intellectual negligence. We are negligent when we should be open to verifying evidence, but close our minds instead. When this happens our success is lucky. And the luck involved is no less luck when it is good luck, as when the tank happens to be full anyway, or the further evidence would all be confirming evidence. Finally, our results in section 4.2 provide additional reasons for retaining an open mind even when one already knows. These are again reasons independent of contextualism, subject-sensitive invariantism, and pragmatic encroachment. Recall the emphasis of section 4.2 on your diminished awareness of how a belief is initially formed. Recall the gradual reduction of your ability reliably to endorse beliefs. Recall finally what you will need to rely on as your initial acquisition of the belief recedes into the past. Your later rational judgment will need to draw on the evidence available to you at that later time. And this will inevitably be new evidence that bears on the truth

88 Ernest Sosa of the proposition believed. In blocking later evidence, as soon as you know that p, you will hence preclude or impair your later rational judgment that p. That reveals a cost of closing your mind even when you have a prima facie reason for doing so: namely, the reason that you might otherwise be misled by contrary evidence. The cost is that you would now too often, too seriously preclude or impair your ability to judge rationally that p. You would negligently, willfully overlook evidence available synchronically that might bear importantly on your judgment. This would entail a massive epistemic loss in perpetuity.

References Harman, G. (1973). Thought. Princeton: Princeton University Press. Kripke, S. (2011). Philosophical Troubles: Collected Papers, Vol. 1. Oxford: Oxford University Press. Sosa, E. (1964). “The Analysis of ‘Knowledge That P’,” Analysis 25(1): 1–8. Sosa, E. (1991). Knowledge in Perspective. Cambridge: Cambridge University Press. Sosa, E. (forthcoming). “Process Reliabilism and Virtue Epistemology,” in H. Kornblith and B. McLaughlin (Eds.), Alvin Goldman and His Critics. Malden, MA: Wiley-Blackwell.

5 Can there be a Knowledge-First Ethics of Belief?* Dennis Whitcomb

5.1  Stage Setting What should we believe, and which of our beliefs are justified? These questions, both important, are distinct. You might look at your hands and believe that you have them—but believe this simply on a whim, and not because you see them. Even though you believe what you should, your belief is unjustified. Thus we should distinguish propositions one should believe from justified beliefs. Additionally, we should distinguish beliefs from withholdings of belief. We should sometimes withhold, just as we should sometimes believe; and withholdings are sometimes justified, just as beliefs are sometimes justified. The job of the ethics of belief is to delineate the cases where we should believe and where we should withhold, as well as the cases of justified belief and justified withholding. In this chapter, I’ll be exploring the first part of this task—the theory of what we should believe and what we should withhold belief on. I’ll explore several knowledge-first theories about these issues, that is to say, several theories that use the notion of knowledge to delineate the cases in which we should believe and in which we should withhold belief. It is an open question to what extent we should take knowledge first in our epistemological theorizing. If a knowledge-first theory of what we should believe (and what we should withhold belief on) can be made to work, then that is some reason to adopt the knowledge-first approach to epistemology as a whole. If no such theory can be made to work, then that is some reason to reject the knowledge-first approach to epistemology as a whole.   *  For invaluable comments on this chapter, I thank to Esa Diaz-Leon, Dan Howard-Snyder, Frances Howard-Snyder, Hud Hudson, Clayton Littlejohn, Gerald Marsh, Jonathan Matheson, Aidan McGlynn, Michelle Saint, Steve Steward, Nick Treanor, Ryan Wasserman, audiences at the University of Victoria and the University of Manitoba, and the students in my Theory of Knowledge course at Western Washington University.

90  Dennis Whitcomb I won’t converge on a view about which one of these two options is correct. Instead, I will explore the problems and prospects of the approach. The exploration will proceed via closely engaging numerous attempts to build the knowledge-first theories in question. Most of those attempts don’t work, for reasons I will bring out. Others might work, depending on how their details end up; this too I will bring out. When I inquire into what we should believe, I mean to ask that question in a specifically epistemic sense. To get a fix on this sense, put yourself in the position of an undergraduate who has just learned about Pascal’s wager. Imagine you have the following common reaction to the wager: “OK, sure, that shows that believing in God serves my interests. But when I asked whether I should believe in God, I didn’t want to know whether that belief would serve my interests. I wanted to know something else . . . .” Undergraduates regularly have this reaction to Pascal’s wager. And they are regularly relieved to be told that there is a particular thing they are looking for: epistemic (as opposed to practical) reasons to believe in God, and in particular epistemic reasons strong enough that they should, from an epistemic point of view, believe. It is the particular “should” delineated here, that I focus on in the current chapter. It is the epistemic should. If you are inclined to deny the existence of this “epistemic” should, then think about it this way instead. There is such a thing as propositional justification. To have propositional justification to believe a proposition is to have good enough epistemic reason to believe it. Similarly, to have good enough propositional justification to withhold belief on a proposition, is to have good enough epistemic reason to withhold belief on it. What we should believe, in my sense—what I am calling the epistemic sense—is just what we have propositional justification to believe. Similarly, what we should withhold belief on—in the epistemic sense—is what we have propositional justification to withhold belief on. In the terminology of “propositional justification,” then, the current chapter is searching after a knowledge-first theory of what we are propositionally justified to believe and withhold belief on. To my ear, “should believe” talk expresses the same issues more naturally than “propositional justification” talk, and so I will stick to the former.

5.2  Identity Theories Our task is to build a knowledge-first theory of what we should believe and what we should withhold belief on. The most straightforward way to discharge this task is to simply identify what we should believe with what we know. Timothy Williamson comes close to advocating such an identification in the following passage: . . . the fundamental rule of assertion is that one should assert p only if one knows p. .. more speculatively, we may project the account of assertion back on to its mental counterpart, judgment (or belief). What results is the rule that one should judge (or believe p) only if one knows p. (Williamson 2000: 11)

Can There Be a Knowledge-First Ethics of Belief?  91 Similarly, Jonathan Sutton argues at length that One must: believe p only if one knows p. (Sutton 2007: 44)

Notice the colon after the “must.” It’s important; it specifies that the “must” has wide scope ranging over the entire rest of the sentence. Contrast this wide-scope view with: One must believe p only if: one knows p.

This view gives the “must” narrow scope ranging only over “believe p.” To start to see the differences between the two views, consider the moral principle “you should, if you kill an animal, kill it humanely.” On at least one reading, this principle is uncontroversial. But we should be clear on the what the relevant reading is. Suppose that we read the principle as giving its “should” narrow scope. Then the principle amounts to: Narrow-Scope Animal Killing Theory If you kill an animal, then you should kill it humanely. This principle is very implausible. Suppose that, in a fit of rage about my constant bragging about my pet dog Frank, you sneak into my house and kill him. It would then be true that you kill Frank. From this and the narrow-scope animal killing theory (and some simple logic), we get the result that you should kill Frank humanely. But surely that is a mistake; surely you should not kill Frank at all, humanely or otherwise (see Figure 5.1). So the narrow-scope animal killing theory is far from uncontroversial. Yet there is a sense in which “you should, if you kill an animal, kill it humanely” is uncontroversial. Plausibly, that sense is what we get when the “should” has wide scope: Wide-Scope Animal Killing Theory You should (kill an animal only if you kill it humanely). Some theorists think this principle is meaningless as stated, on the grounds that “should” does not meaningfully take sentences as its objects.1 If you are one of those theorists, then read the wide-scope animal killing theory as follows: You should be such that the following is true of you: (you kill an animal only if you kill it humanely). Either reading is fine for our current purposes. Maybe they amount to the same thing anyway. In any case, we can now ask: does the wide-scope animal killing theory yield the implausible result that you should kill Frank humanely, in the case described above? No, it does not. To see why not, recall that (p → q) is equivalent to (¬p ∨ q). One way to make it true that “if you kill Frank then you kill him humanely,” then, is to make it false that you kill him. In such a case you might, consistently with the truth of the wide-scope animal killing theory, also be conforming to another true principle, a true principle to the effect that “you should not kill Frank, humanely or otherwise.”   See Schroeder (2004) for illuminating discussion of these issues.

1

92  Dennis Whitcomb

Figure 5.1: Frank

In other words, there are possible worlds in which it is true both that you should not kill Frank, not even humanely, and that you should (kill Frank only if you kill him humanely). In some of these worlds you might even kill him—humanely. But it would still be true that you shouldn’t kill him, humanely or otherwise. By killing him, even humanely, you would do something you shouldn’t do, something that violates the principle “you shouldn’t kill Frank humanely or otherwise” even though it does not violate the (also true) principle “You should: kill Frank only if you kill him humanely.” This means that, even along with the assumption that you do kill Frank, and even along with the assumption that you do kill Frank humanely, it does not follow from the widescope animal killing theory that you should kill him humanely. So there is a separation between what is said by the wide-scope animal killing theory, and what you should do with respect to killing animals. As it happens, this separation is quite thorough. Effectively, the wide-scope animal killing theory only tells us about the conditions under which you should do the following conjunctive act: killing-an-animal-and-not-killing-it-humanely. It does not tell us anything else. In particular, it does not tell us anything about the conditions under which you should kill an animal, or the conditions under which you should kill an animal humanely. If you don’t kill one, then the theory tells us nothing about whether you should kill it, or whether you should kill it humanely. If you do kill one, then once again

Can There Be a Knowledge-First Ethics of Belief?  93 the theory tells us nothing about whether you should kill it, or whether you should kill it humanely. It just tells us you shouldn’t kill it inhumanely, which is something quite different from killing it, and also quite different from killing it humanely. To generalize the point, it is this: wide-scope theories, which take the form “you should be such that (P only if Q),” do not tell us anything about whether you should be such that P, or about whether you should be such that Q. They only tell us this: you should not be such that (P and not-Q). Putting this point to work, we can now see why it turns out that, even in conjunction with the assumption that you kill Frank in the case described above, it does not follow from the wide-scope animal killing theory that in this case you should kill him humanely. The reason is this: the wide-scope animal killing theory does not say anything about whether you should kill Frank humanely in any case. Ipso facto, it does not say that you should kill him humanely in the case described above. With the differences between wide-scope and narrow-scope should-statements now broached, we can return to the question of what we should believe. Recall the theories on offer from Williamson and Sutton. The basic idea behind those theories is that we should believe only the things we know. If we should go along with this idea, then it is hard to see why we shouldn’t also go along with the idea that we should believe all the things we know. Combining these two ideas, we arrive at the view that we should believe all and only what we know. It is important to distinguish between narrowscope and wide-scope versions of this view. If the “should” has narrow scope, the idea amounts to: Narrow-Scope Identity Theory (You should believe p) iff (you know p). To be completely explicit about this, the “iff ” here is intended to express the material biconditional, and the whole view is intended to be implicitly preceded by universal quantifiers ranging over people, worlds, and times. Stated completely explicitly, then, the narrow-scope identity theory amounts to: For any person s, world w and time t: (s should believe p at t in w) iff (s knows p at t in w). I will consider numerous views in this chapter, and it would be cumbersome to state each of them so explicitly. Thus I will often leave implicit the quantifiers over people, worlds, and times, as well as leaving implicit the point that the biconditionals are material. As for the narrow-scope identity theory, it is clearly mistaken. Knowledge entails belief; hence it follows from the narrow-scope identity theory that whenever we should believe we do believe. This is tantamount to the claim that we cannot fail to believe what we should, which is absurd. It will help to have a name for this particular absurd result. We’ll say that a theory entails “compulsory completeness with respect to what we should believe,” or in short, “compulsory completeness,” when it entails that we always believe everything we should.

94  Dennis Whitcomb This terminology is borrowed in the obvious way from logic. Just as a logical system is complete with respect to a given property x iff every formula having x is a theorem, one’s belief corpus is complete with respect to a given property x iff every proposition having x is something one believes. From an epistemic point of view, it is good for one’s belief corpus to be complete with respect to what one should believe. Of course, it is possible for our belief corpuses to not be complete with respect to what we should believe. Theories denying this possibility have the problem of compulsory completeness. The narrow-scope identity theory is false because it has this very problem. Of course, there is an easy fix. We can just widen the scope of the “should” (as Sutton does by putting the colon after his “must,” and as Williamson means to do with his own principle). This gives us: Wide-Scope Identity Theory You should (believe p iff you know p). Or, to put it completely explicitly, For any person s, world w and time t: s should not be such that he believes p at t in w without knowing it, or such that he knows p at t in w without believing it.(The last clause is trivially satisfied, since knowledge entails belief.) To see why this theory avoids the problem of compulsory completeness, suppose that you should believe p. From the narrow-scope identity theory, it follows that you know p (and hence that you believe it). From the wide-scope identity theory, it does not follow that you know p. For the wide-scope identity theory doesn’t tell us anything about the conditions under which you should believe p, just as the wide-scope animal killing theory doesn’t tell us anything about the conditions under which you should kill animals. Just as the wide-scope animal killing theory speaks only of the conditions under which you should kill animals inhumanely (and says they never obtain), the wide-scope identity theory speaks only of the conditions under which you should believe-and-not-know (and says they never obtain). Beyond that, the wide-scope identity theory says nothing. Ipso facto, it says nothing about the conditions under which you should believe things. From the point of view of ethicists of belief, who are in the business of trying to figure out what we should believe, this is not a helpful way out of the problem of compulsory completeness. What about cases where we don’t both believe and fail to know? For that matter, what about cases where we do both believe and fail to know? The theory tells us that, in any given case, we should not (believe and not know). Even for the cases where we both believe and fail to know, this doesn’t entail that we shouldn’t believe—it only entails that we shouldn’t believe and not know, which is something quite different. (Compare: even for the cases where you both kill Frank and kill him inhumanely, the wide-scope animal killing theory doesn’t entail that you shouldn’t kill him—it only entails that you shouldn’t kill him inhumanely, which is something quite different.)

Can There Be a Knowledge-First Ethics of Belief?  95 In sum, there is a dilemma for theories identifying what we should believe with what we know. On the one hand, those theories can give their “should” narrow scope. So understood, the theories are well enough informative; but the information they give us is egregiously mistaken. On the other hand, they can give their “should” wide scope. Understood like this, the theories do not seem to make egregious mistakes. However, it is only by being objectionably uninformative about what we should believe, in fact completely silent about the matter, that they avoid such mistakes.

5.3  Counterfactual Theories The wide-scope identity theory doesn’t tell us anything about what we should believe. Can its informational deficiencies be remedied? We might try to remedy them with counterfactuals, for instance by adopting: Narrow-Scope Counterfactual Theory (You should believe p) iff (if you were to believe p, you would know p). But this theory has the standard problem for theories involving counterfactuals— the “conditional fallacy” problem.2 To see this, suppose that you have more or less the same evidence all of us currently have about whether Barack Obama is the president of the United States, but due to a pure hatred of Obama you can’t quite yet bring yourself to believe it—you think he must be ineligible due to cheating in the vote count, or perhaps to being a Kenyan, and so not really president at all. Of course, you are wrong about that. Obama really is president. Nor are there any Gettier hijinks on the scene; nor would you believe on an improper basis if you were to believe—you’d simply give in to the very same evidence you actually have, and get over your irrational hang-ups by believing, on the basis of your evidence, that Obama is president. So far so good: the theory tells us, correctly, that you should believe Obama is president. But now let us add a twist. Suppose that you are being closely watched by a moderately malevolent demon with a taste for irony. If you come to believe that Obama is president, the demon will immediately kill Obama, instantaneously replacing him with a macrovisually indistinguishable Kenyan. This twist does not make a difference to whether you should believe that Obama is president. But notice that without the twist it is true that you would know if you were to believe, whereas with the twist it is false that you would know if you were to believe. This means that if the narrow-scope counterfactual theory is true, then the twist does make a difference to whether you should believe. And, again, the twist does not in fact make a different to whether you should believe. The theory is therefore mistaken. Is there a way to reformulate the theory so as to avoid the problem? We might try to do that by changing the relevant counterfactual, somehow building into it a stipulation  

2

  See Shope (1978).

96  Dennis Whitcomb that no such things as ironic demons are present in the worlds where the theory requires us to know things if in the actual world we should believe those things. Let us say that a given batch of conditions are “propitious for your knowing p” just in case those conditions don’t block your belief that p from being knowledge.3 Putting this notion of propitious conditions to work, we can consider: Revised Narrow-Scope Counterfactual Theory (You should believe p) iff (if you were to believe p, and the conditions were propitious for your knowing p, you would know p). This new view avoids its predecessor’s problems with the ironic demon. However, the cure is worse than the disease. Every proposition is always such that, if you were to believe it and the conditions for your knowing were propitious, you would know it. The revised narrow-scope counterfactual theory therefore entails that we should always believe everything. Is there is a more charitable way to revise the narrow-scope counterfactual theory? Perhaps we should take it that, in the worlds where the theory requires you to know p if in the actual world you should believe p, no non-mental conditions block your belief from being knowledge. Let us say that “the non-mental conditions are propitious for knowing p” just in case none of those conditions block your belief that p from being knowledge. Using this notion of non-mental conditions that are propitious for knowing p, we can rebuild the view as follows: Re-revised Narrow-Scope Counterfactual Theory (You should believe p) iff (if you were to believe p, and the non-mental conditions were propitious for your knowing p, you would know p). This view, unlike its predecessor, avoids the result that we should always believe everything. However, a new problem arises. Suppose that you don’t believe some proposition p, but that if you were to believe it, you would believe it for bad reasons. To fill in the details of the case, suppose that you have more or less all the same evidence most of us have about Obama, plus a small smattering of evidence from conspiracy-theoretic films about the matter. The large preponderance of your evidence indicates that Obama really is president, of course. But due to your hatred of Obama, you tend to pay much more attention to evidence gleaned from conspiracy theory literature, than to the rest of your evidence. You are on the fence between believing that Obama is not president (because he is a Kenyan—as many conspiracy theories say) and believing that he is president (because he is the son of two American communists—as is maintained by a flashy new documentary film you just watched).4 In such a case, the following counterfactual is false: if you were to believe that Obama is president, and the non-mental conditions for knowing were propitious, you would   3  The notion of conditions that are “propitious for knowing” is appropriated from Bird (2007: 85–6). Bird does not define that notion, but he seems to mean something like the above by it.   4  See Gilbert (2012).

Can There Be a Knowledge-First Ethics of Belief?  97 know that Obama is president. If you were to believe that Obama is president, and the non-mental conditions were propitious, you would not know that Obama is president. Instead of knowing as much, you would believe as much for a bad reason (namely, that Obama had as parents two American communists). As a result, the re-revised narrowscope counterfactual theory entails that you should not believe that Obama is president. But in fact, you should believe that Obama is president. After all, you possess more or less all the same evidence as do the rest of us with respect to the issue of whether Obama is president, aside from a bit of conspiracy literature to which you are irrationally attached. The re-revised narrow-scope counterfactual theory is therefore mistaken. Is there another more charitable way to reinterpret the narrow-scope counterfactual theory? No—at least, not as far as I can tell. But there is still one more way to try to save the counterfactual approach: we might try to get around the problems by widening the scope of the “should.” This gives us: Wide-Scope Counterfactual Theory You should (believe p iff (if you were to believe p, you would know p)).5 However, it should be clear that the move from narrow to wide scope here won’t help, any more than the move from narrow- to wide-scope identity theories will help. The narrow-scope counterfactual theory does not tell us anything about the conditions under which you should believe things. It speaks only of the conditions under which you should be such that (you believe and the embedded counterfactual doesn’t hold), or such that (the embedded counterfactual holds and you don’t believe). To say something about only these conditions, is to say nothing at all about the conditions under which you should believe things.

5.4  Knowledge-Minus-Belief Theories We’ve seen in two cases now that when narrow-scope theories yield bad results, it doesn’t help to try to avoid those results by widening the scope of their “shoulds.” The pattern generalizes. In general, wide-scope theories change the subject matter. For the remainder of our exploration of knowledge-first theories of what we should believe (and withhold on), then, we can leave wide-scope theories aside. Focusing on narrow-scope theories, we’ve already seen that identity approaches and counterfactual approaches come up short. In the wake of these two approaches it is natural to try a different approach appealing to the conditions on knowledge. On this approach, we should believe a given proposition if and only if we meet all the conditions required for knowing that proposition, other than the belief condition. On this way of thinking about things, we build up the theory of what we should believe by

  5  Sutton (2005: 373–4) seems to suggest this view; also see Sutton (2007: 56). For useful discussion, see Coffman (2010) and Littlejohn (2012).

98  Dennis Whitcomb taking knowledge and removing belief; whatever thing is left, that thing is necessary and sufficient for being such that we should believe. The most obvious way of developing this approach takes it that you should believe when, for every condition knowledge requires other than the belief condition, you meet it. Using the box for metaphysical necessity, this amounts to: Non-belief Conditions Theory (You should believe p) iff (for every condition x such that □(Kp → x) and x ≠ Bp, x obtains). But this won’t quite do. Consider the conditions true belief and knowledge. Each of these is necessary for knowledge, and not identical to belief. Thus, the non-belief conditions theory tells us that every case where you should believe is a case where these conditions hold. But these conditions both entail belief. So the theory tells us that every case where you should believe is a case where you do believe. Compulsory completeness returns. Trying again, we might consider: Non-Belief-Requiring Conditions Theory (You should believe p) iff (for every condition x such that □(Kp → x) and ¬ο(x → Bp), x obtains). According to this view, you should believe when, for every condition that is required by knowledge but which does not itself require belief, you meet it. By developing the “knowledge minus belief ” approach in this way we nicely avoid the problem of compulsory completeness, because in building up the conditions under which you should believe, we factor out whatever requires you to believe. But there is a different problem. Consider the condition having a propositional attitude whose content is p. This condition is required by knowledge, and does not require belief. Thus it is a condition we must meet if we should believe a given proposition, according to the non-belief-requiring conditions theory. The theory tells us that we should believe only those propositions such that we (already) have propositional attitudes with those propositions as their contents. And this is a mistake. To see why it is a mistake, consider Watson, that well-intentioned but dim investigator working for Sherlock Holmes. Watson may have never even considered the proposition that Smith is the murderer; but this may only be because he did not properly follow his evidence. We can even imagine Holmes explaining it all to him after the fact; to this explanation, Watson might respond “Now I see—I should have believed it was Smith.” In this sort of case Watson speaks correctly; he should have believed it was Smith, even though he had no propositional attitude with that particular content. (If you don’t find this case convincing as it stands, alter it so that Watson is not dim, but just intellectually lazy.)6

 

  Thanks to Esa Diaz-Leon for alerting me to this problem. It may also be a problem for a principle

6

Can There Be a Knowledge-First Ethics of Belief?  99 In light of the problems with the non-belief-conditions and non-belief-requiring conditions theories, we might try to move from the notion of necessary conditions on knowledge, to notion of conditions at least partly in virtue of which we know things.7 There are many necessary conditions on knowing p, which are not also conditions at-least-partly in virtue of which we know p whenever we do know p. For instance, that 1+1=2 must be true whenever anyone knows that he has hands; hence it is a necessary condition on knowing that one has hands. But it is not one of the conditions at-least-partly in virtue of which we know that we have hands whenever we do know as much. To be sure, it is a condition at-least-partly in virtue of which we know that 1+1=2, whenever we know that proposition. But still, it is not a condition at-least-partly in virtue of which we know that we have hands, whenever we know we have hands. Let us say that a condition x is “knowledge-constitutive for p” if and only if: whenever anyone knows p, it is at-least-partly in virtue of x that he knows p. Putting the notion of knowledge-constitutive conditions to work, we can consider some new versions of the knowledge-minus-belief approach to what we should believe. For instance, we can consider a reformulated version of the non-belief conditions theory, to wit: Non-belief knowledge-constitutive conditions theory (You should believe p) iff (for every condition x such that x is knowledgeconstitutive for p and x ≠ Bp, x obtains). This particular approach does not help, though, because it brings back compulsory completeness. Plausibly, whenever anyone knows any proposition p, it is partly in virtue of this condition—properly forming his belief that p—that he knows p. What is “properly forming” a belief? Perhaps, forming it in a reliable manner. Or perhaps, forming it on the basis of one’s evidence, or on the basis of epistemic virtues, or in some other special way. In any case, it is plausible that some sort of proper formation is required, whatever the details of “proper formation” turn out to be. But this means that the non-belief knowledge-constitutive conditions theory entails that whenever we should believe p, we properly form the belief that p. And properly forming the belief that p entails having the belief that p. Thus the theory tells us that whenever we should believe p, we do believe p. We can’t fail to believe what we should.

advocated by Smithies (2012: 268, 284) depending on how that principle is best interpreted. On Smithies’ view, it is “correct” for you to believe p iff you meet all the “epistemic, as opposed to psychological” conditions on knowing p. He calls this principle “the K rule for correct belief.” It is not fully clear whether what one should believe is identical to what it is “correct” for one to believe, in Smithies’ sense of correctness. Nor is it fully clear whether, on Smithies’ use of these terms, the “epistemic, as opposed to psychological” conditions on knowing amount to the conditions which are required by knowledge but which do not themselves require belief. If these identities both hold, though, then the problems for the non-belief-requiring conditions theory are also problems for Smithies’ “K rule for correct belief.”  

  Thanks to Ryan Wasserman for making this insightful suggestion.

7

100  Dennis Whitcomb Trying again, we might reformulate the non-belief-requiring conditions theory via the notion of knowledge-constitutive conditions. This gives us: Non-belief-requiring knowledge-constitutive conditions theory (You should believe p) iff (for every condition x such that x is knowledgeconstitutive for p and ¬□(x → Bp), x obtains). According to this theory, you should believe p when (and only when) you meet all the conditions which both (a) are conditions at-least-partly in virtue of which anyone who knows p, knows p, and (b) don’t require believing p. This view nicely sails around the problem of compulsory completeness. For, much like its predecessor that involved mere necessary conditions instead of knowledge-constitutive conditions, this theory factors out whatever requires you to believe p. This theory also sails around the problem of conditions like having a propositional attitude with p as its content. Such conditions are plausibly not among the conditions in virtue of which anyone who knows p, knows p. Additionally, this theory sails around the problem of conditions like forming the belief that p in the right way. Although that condition is knowledge-constitutive for p, it requires believing p; and for that reason, the theory does not say it must obtain, in order for it to be true that we should believe p. One might worry that similar conditions still bring up the same problems. For instance, one might worry that the condition properly forming a propositional attitude with p as its content is required for us to be such that we should believe p, given the theory we are now considering. However, that worry is off-target. Although properly forming a propositional attitude with p as its content is plausibly necessary for knowing p, it is not plausibly knowledge-constitutive for p. Although it is at-least-partly in virtue of properly forming the belief that p that we know p, it is not at-least-partly in virtue of properly forming a propositional attitude with p as its content that we know p. The non-belief-requiring knowledge-constitutive conditions theory finds its way around all of the problems so far discussed for the preceding theories. It is probably the best attempt we’ve seen so far, among knowledge-first attempts to say what we should believe. But it has its own laundry list of difficulties. Most obviously, it tells us that in every Gettier case, and in every case where p is false, it is false that you should believe p. For instance, it is false that you should believe there is a barn in front of you, when you are looking at one that (unbeknownst to you) is surrounded by barn facades. Similarly, it is false that you should believe that the job-getter has ten coins in his pocket, when the boss says that you will get the job, and you count ten coins in your own pocket, but (unbeknownst to you) someone else who also has ten coins in his pocket will get the job. Whenever something happens to be false—for instance, whenever it is false that you have hands because (unbeknownst to you) you are a handless brain in a vat being fed perceptions as of the actual world—it is also false that you should believe that thing.

Can There Be a Knowledge-First Ethics of Belief?  101 These results seem clearly mistaken, even absurd.8 Can they be explained away? Perhaps. Numerous theorists argue that one should assert p only if one knows p, or that one should use p as a reason for acting only if one knows p. These views tell us that we should never assert (or act on) anything false, or anything we are Gettiered about. Such results seem clearly mistaken, and even absurd, in the same ways in which the corresponding results about what we should believe seem mistaken and even absurd. However, there is a standard response to them. The response consists in distinguishing, on the one hand, what we should assert (or act on), and on the other hand, what is reasonable or excusable for us to assert (or act on). This distinction provides some sugar for the bitter pill of saying that Gettier victims and brains in vats regularly assert and act on what they shouldn’t. Those characters turn out to be still making reasonable or excusable assertions, and engaging in reasonable or excusable actions, even though they are doing and asserting things they shouldn’t. There is something positive to be said of those assertions and actions, then, even though they are not assertions people should make or actions people should do. When theorists find it clearly mistaken or absurd to say that we shouldn’t assert (or act on) what we don’t know, they are just failing to distinguish what is reasonable or excusable to assert or act on, from what we should assert or act on. The seemingly absurd results about assertion and action are thus explained away.9 Could a similar explanation be applied to the similarly absurd-seeming results about what we should believe? Could such an explanation adequately defend the non-belief-requiring knowledge-constitutive conditions theory from its problems involving falsehoods and Gettier cases? Starting with the non-belief-requiring knowledge-constitutive conditions theory and then adding such an explanation, we end up with a package of views claiming that (a) we believe what we shouldn’t in Gettier cases, and when we are brains in a vats, but that (b) these beliefs are nonetheless reasonable or excusable. Such package views—views packaging fairly demanding knowledge-theoretic claims about what we should believe with fairly undemanding claims about what is reasonable or excusable for us to believe— have quite a bit to recommend them.10 In the end, they may turn out to work. However, if they do work, then they work by passing the theoretical buck from the notion of what we should believe to the notion of what is reasonable or excusable for us to believe. And if there is not something significant also said about the latter notion, then it is hard to shake the idea that the package views relocate our original question instead of answering it. We start out by asking what we should believe. Package knowledge-first theories answer this question, but their answers lean hard on a new notion, the notion of what is reasonable or excusable for us to believe. In order to fully adjudicate these package   8  Of course, these problems aren’t unique to the non-belief-requiring knowledge-constitutive conditions theory. They apply to many other theories as well, for instance the narrow-scope identity theory. For useful discussion, see Littlejohn (2012).   9  See Williamson (2000), DeRose (2002); Sutton (2007), and Hawthorne and Stanley (2008).   10  I once argued for versions of them concerning the satisfaction of curiosity (see Whitcomb 2010), but I’m now unsure about those arguments.

102  Dennis Whitcomb knowledge-first theories, we need to ask a second question:  what is reasonable or excusable for us to believe? Perhaps there is a good answer. But until we have one, it is hard to shake the idea that what mattered about our first question, or at least a significant part of what mattered about our first question, has been relocated (under the notion of what is reasonable or excusable to believe) instead of being resolved.11 Can the package views adequately fill in the details about what is reasonable or excusable for us to believe? Here is a reason to doubt as much. Knowledge-first theories of what we should believe, if they are combined with theories of what is reasonable or excusable for us to believe, ought to be combined with knowledge-first theories of the latter issue. And it is hard to see how there could be an adequate knowledge-first theory of the latter issue. Each so-far-discussed theory of what we should believe— the identity theories, the counterfactual theories, the knowledge-minus-belief theories—could be reinterpreted as a theory about what is reasonable or excusable for us to believe. But the reinterpreted theories seem just as problematic, and for just the same reasons, as do the originals. So, if we are going to find an adequate knowledge-first theory of what is reasonable or excusable for us to believe, then we will have to find it in some place other than reinterpreted versions of the views we’ve already discussed. Where else is there to look? One prima facie promising strategy is to look at the additional theories (about what we should believe) that I will discuss in the rest of the current chapter. However, the same point applies with those additional theories. As I will argue, each of those additional theories has some sort of problem. As for what those problems are, we’ll get to them soon enough. For now, the relevant point is this: each of those problems applies with equal force whether its target theory is about what we should believe, or about what is reasonable or excusable for us to believe. This means that we are not going to find—not in the sorts of ideas I explore here anyway—an adequately developed knowledge-first theory of what is reasonable or excusable to believe. Perhaps we can find one somewhere else. But the prospects do not appear promising. As a result, what we might call “the package strategy” for knowledge-first views about what we should believe—that is, the strategy of saving those views from objections by packaging them with further views about what is reasonable or excusable to believe—does not appear to be a promising strategy. We should therefore stop the attempt to use that strategy to save the non-belief-requiring knowledge-constitutive conditions theory. And without that strategy, it is not clear how to save that theory. Nor is it clear how we might come up with another better knowledge-minus-belief theory. Those theories do not appear to work out. We need to look elsewhere.

5.5  Duplication Theories Trying again to formulate a knowledge-first theory of what we should believe and what we should withhold belief on, we turn to the notion of duplication. A given (possible)   For a similar point about the similar package views about assertion and action, see Gerken (2011).

11

Can There Be a Knowledge-First Ethics of Belief?  103 person is your x-duplicate at a given time just in case they have all the same x-properties you have at that time. For instance, someone is your physical duplicate at t just in case they have all the same physical properties you have at t. Similarly, someone is your mental duplicate at t just in case they have all the same mental properties you have at t. These notions can help us formulate several prima facie plausible knowledge-first approaches to the ethics of belief. For instance, consider: Mental Duplication Theory (You should believe p) iff (p is known by at least one of your mental duplicates). According to this view, you should believe a given proposition p at a given time just in case it is metaphysically possible that there is some person, who has all the same mental properties as you at that time, and who knows p at that time. At first blush at least, this view comports with widespread intuition about what we should believe in numerous cases— for instance Gettier cases and fake barn cases.12 But whatever its virtues, this view does not give us what we are after. For suppose that you do not believe p. Then, none of your mental duplicates know p, because none of them even believe it. Hence the view entails that whenever you do not believe p, it isn’t the case that you should believe p. In other words, it entails that you cannot fail to believe what you should. This is the problem of compulsory completeness, once again leaving a prima facie promising knowledge-first theory in its wake. We could try to save the mental duplication approach by focusing on past mental duplicates. Let us say that a (possible) person is your past mental duplicate just in case, at every time in your past, that person was your mental duplicate. Putting this notion of past mental duplication to work, we can consider: Past Mental Duplication Theory (You should believe p) iff (p is known by at least one of your past mental duplicates).13   12  Actually, on reflection it is not so clear that this theory adequately deals with all of these cases. Suppose that you, while looking at a particular barn that nobody else is looking at, believe that that thing is a barn. On some views of content (and modality and transworld identity), it turns out that the content of your belief is not identical to the content of any other possible person’s belief. Any other possible person would be looking at a different barn (in a different possible world—which is why it would be a different barn), thinking of it that it is a barn. And that person’s belief, being the belief that that thing is a barn, would have a different content than your belief, the belief that the thing you are looking at is a barn. Supposing that differences in content are sufficient for differences in belief, the mental duplication theory mistakenly entails that when you are in fake barn country, you should not hold the belief you would express by uttering “that is a barn in front of me.” This is a problem for the mental duplication theory. Similar problems arise for theories that involve modal conditions on knowledge—safety conditions, for instance. Perhaps we should try to solve these problems by advocating some sort of internalistic theory of content, or by denying the sufficiency of differences in content for differences in belief, or by building up some sort of counterpart-theoretic treatment of belief identity, or by building up some non-world-theoretic treatment of modality. Or perhaps we shouldn’t try to solve them at all, but instead take them to refute the theories they target. I will leave these issues—that arise for most any duplication theory—aside. For useful discussion, see Manley (2007).   13  Bird (2007) develops a somewhat similar approach to the notion of doxastically justified belief—as opposed to my own target notion of what we should believe (that is, the notion of what we have propositional justification to believe). For discussion of Bird’s views, see McGlynn (2012). For another similar knowledge-first approach to doxastically justified belief, see Reynolds (2013).

104  Dennis Whitcomb Interestingly, this theory avoids the problem of compulsory completeness. To see why, suppose you don’t believe p. What follows from this supposition? It does not follow that none of your past mental duplicates believe p, or that none of them know p. They may believe or know p, even if you don’t, owing to the ways in which they currently differ from you despite sharing your mental past. In such cases you might fail to believe p even though you should; thus, completeness is not compulsory. By focusing on the past instead of the present, the past mental duplication theory avoids compulsory completeness. But there is a new problem. Suppose that S is your past mental duplicate right now, but that he differs mentally from you now in that he now sees that he lacks hands. He knows that he lacks hands; hence the narrow-scope past mental duplication theory entails that you, right now, should believe that you lack hands. And of course that is a mistake. The general point of this example is that beings with the same mental histories should sometimes believe different things owing to their current differences, even if one of them knows those things. The past mental duplication theory is inconsistent with this point. Its focus on the past instead of the present extricates us from the problem of compulsory completeness, but it does so by denying the fact that beings with the same mental histories should sometimes believe different things owing to their current differences, even if one of them knows those things. Since the notions of mental duplication and past mental duplication both turn out to be unhelpful when taken on their own, we might try combining them. Let us say that someone is your up-to-now mental duplicate just in case they are your mental duplicate and were your mental duplicate at every time in the past. With this notion we can now formulate another theory, to wit: Up-to-Now Mental Duplication Theory (You should believe p) iff (p is known by at least one of your up-to-now mental duplicates). But this brings back compulsory completeness. For suppose that you don’t believe p. Then, none of your up-to-now mental duplicates believe p either. So none of them know p. So it’s false that you should believe p. From the narrow-scope up-to-now mental duplication theory and the assumption that you don’t believe p, then, we get the result that it is false that you should believe p. The theory entails that you can’t fail to believe what you should. To summarize the discussion so far:  several initially appealing knowledge-first approaches to the ethics of belief fall short. For various reasons, we end up falling short with identity approaches, counterfactual approaches, various knowledge-minus-belief approaches, and various approaches based on mental duplication. It might be tempting, given all this, to conclude that the knowledge-first ethics of belief amounts to a degenerating research program, and indeed that the ethics of belief shows to be degenerate the whole knowledge-first approach to epistemology. Such conclusions would be very satisfying to many traditional epistemologists, tired of being accused themselves by knowledge-firsters of being the degenerates on the scene. But I don’t think these conclusions are warranted, at least not yet.

Can There Be a Knowledge-First Ethics of Belief?  105 Let us help ourselves to the notion of “one’s evidence,” that is to say the notion of the evidence you have—taking that notion as well enough understood to be usefully employable in the ethics of belief.14 On the basis of this notion, we can start again the project of building a knowledge-first ethics of belief. Let us say that someone is your evidential duplicate just in case they have all the same evidence you have. Putting this notion to work, we can consider: Evidential Duplication Theory (You should believe p) iff (p is known by at least one of your evidential duplicates). This theory won’t quite do. Suppose you have no evidence whatsoever that Dean Martin is a good singer, even though you find yourself believing it. Suppose also that, although you’ve completely forgotten this, you formed that belief by trusting your mother’s testimony, which you knew to be unreliable about such things.15 Finally, suppose that one of your evidential duplicates formed the same belief on good grounds (which he has forgotten) having to do with listening to Dean Martin, comparing him to other singers, and so on, and without any Gettier hijinks happening. Given these suppositions, your evidential duplicate knows that Dean Martin is a good singer. Thus it follows from the evidential duplication theory that you should believe that Dean Martin is a good singer. But that is a mistake. Trying again, we can conjecture: Past Evidential Duplication Theory (You should believe p) iff (p is known by at least one of your past evidential duplicates). This theory, however, fails to properly deal with the relevance of your current evidence to what you should believe. Suppose that one of your past mental duplicates now has, for the very first time, an enormous amount of evidence that he lacks hands. Since you know you have hands, it follows—from the past evidential duplication theory—that he should believe that he has hands. This is a mistake. So neither the evidential duplication approach, nor the past evidential duplication approach, turn out to work. Trying yet again, we might conjecture: Up-to-now Evidential Duplication Theory (You should believe p) iff (p is known by at least one of your up-to-now evidential duplicates). According to this theory, you should believe exactly those propositions that are known by someone (that is, some possible person) who shares (and always has shared) your evidence. I think that this theory nicely avoids all the problems I’ve raised so far for

  14  Of course, there is a great deal of recent controversy over the notion of “one’s evidence.” See Feldman (1986), Williamson (2000), Kelly (2006), Neta (2008), and Goldman (2009).   15  This is a well-known sort of scenario; for discussion, see Conee and Feldman (2001), Goldman (2001), Feldman (2005), and Greco (2005).

106  Dennis Whitcomb the other knowledge-first approaches. For one thing, it avoids the problem of being too uninformative that the wide-scope views face. Because its “should” takes narrow scope, it tells us whether we should believe any given proposition in any given case. And in combination with a plausible “uniqueness” principle—a principle to the effect that in any given case and for any given proposition p, either you should believe p, or you should believe not-p, or you should withhold on p—it also tells us whether we should withhold on any given proposition in any given case.16 It therefore gives us a fully specified account of what we should believe and what we should withhold belief on. Furthermore, it does not have the problem of compulsory completeness, because even if you do not believe something, one of your up-to-now evidential duplicates might still know that thing.17 In such cases it turns out that you should believe even though you do not believe. Nor does the theory face problems with slow-thinking Watsons or forgotten nefarious histories of belief formation. Our slow-thinking Watson should believe that Smith is the murderer, even though he has never even considered that proposition, because some of his up-to-now evidential duplicates (namely the faster-thinking ones) know that Smith is the murderer. And we get an correct account of, and an explanation of why, you should not believe that Dean Martin is a good singer, when you find yourself believing as much but, unbeknownst to you because you’ve forgotten, you originally formed that belief by trusting your known-tobe-unreliable mother about the matter. Here you should not believe that Dean Martin is a good singer, because none of your up-to-now evidential duplicates know as much. The up-to-now evidential duplication theory, in combination with the uniqueness principle, amounts to our best option so far for a knowledge-first theory of what we should believe and what we should withhold belief on. But it has an obvious problem: it entails that we should never believe necessary falsehoods. Being necessarily false, such claims are necessarily unknown; in which case nobody’s up-to-now evidential duplicates ever know them; in which case the theory entails that nobody should ever believe them. And that is a mistake, because sometimes we should believe necessary falsehoods. For instance, if the entire mathematical community tells us, for several hundred years, that a certain very complicated mathematical claim is true, then we should believe that claim—even if that claim, for some subtle reason, turns out to be false (and hence necessarily false).18   16  The uniqueness principle, though plausible, is not uncontroversial. See White (2005), Ballantyne and Coffman (2011), and Matheson (2011).   17  This assumes, pace Williamson (2000), that the following conjunction is false: knowledge is identical to evidence and knowledge entails belief. Should that conjunction hold, the up-to-now evidential duplication theory would have the problem of compulsory completeness.   18  Following Bird’s (2007) approach to doxastic justification, we could invoke the notion of “corresponding propositions” to get around this problem of necessary falsehoods. The strategy here would be drop the idea that we should believe p iff p is known by some of our up-to-now evidential duplicates, and replace it with the idea that we should believe p iff some proposition corresponding to p is known by some of our up-tonow evidential duplicates. In cases where p is necessarily false, then, we should nonetheless sometimes believe p, because propositions corresponding to p are known by our up-to-now evidential duplicates. Perhaps

Can There Be a Knowledge-First Ethics of Belief?  107 Once again, we need to look elsewhere if we are going to find an adequate knowledge-first ethics of belief. Since the appeal to evidence didn’t work out, it is natural to turn to the other notion most popular in standard contemporary theorizing about the ethics of belief, namely the notion of a belief-forming process.19 Reliabilists theorize about the ethics of belief in terms of reliable belief-forming processes. From a knowledge-first perspective, it makes sense to replace appeals to the notion of a belief-forming process with appeals to the notion of a knowledge-producing process. A theory built on this notion could say, perhaps, that we should believe whatever it is that knowledge-producing processes would have us believe. Since “trusting what mathematicians have been saying for hundreds of years” is a knowledge-producing process, this approach may be able to circumvent the problems with necessary falsehoods that refute even the best of our evidence-centered knowledge-first theories. Of course this theory won’t work as stated; it has numerous problems. For one thing, it has the standard “conditional fallacy” problems for theories involving counterfactuals. For another thing, it gives us very few details about the notion of a knowledge-producing process. It would be nice to get some details about (for instance) whether knowledge-producing processes generate knowledge whenever they are used, or some of the times they are used, or most of the times they are used, or most of the times they are used in the actual world, or what. And even if these first two problems this line of thought can be made to work, but it leans hard on the new notion of a “corresponding proposition.” Until we can say something informative about that new notion, the approach seems underspecified.   19  Here is one more attempt to make the appeal to evidence work out. Combining some standard evidentialist ideas from Conee and Feldman (1985) with some knowledge-first ideas from Williamson (2000), we might (a) identify one’s evidence with one’s knowledge, and (b) conjecture that one should believe (or withhold on) p iff believing (or withholding on) p fits one’s evidence. However, there are problems with this view. Evidence may not be identical to knowledge. And even if it is, there are still other problems. Without an account of “fit,” the view is very underspecified. And it is hard to see what an adequate account of “fit” would look like. The most obvious accounts of fit define that notion in terms of probability, so that believing p “fits” one’s evidence just in case p is rendered probable enough by one’s evidence. But then how probable is “probable enough”? Presumably (but see Whitcomb (2013) for some ways to relax this assumption), there is some value x such that for any proposition p, p is rendered “probable enough” by your evidence iff its probability given your evidence is at least x. What then is the value of x? If x = 1, the view becomes too skeptical, entailing (for instance) that we should believe relatively few things on the basis of testimony. For example, suppose that you are irrationally unwilling to accept the testimony of others, perhaps because your father constantly lied to you during your childhood. A wholly trustworthy person might tell you that the store has pumpkins on sale. You might decline to believe this person, coming to believe only that he said pumpkins are on sale, and not that pumpkins are on sale. If your evidence is identical to your knowledge, and you should believe only those propositions having probability 1 given your evidence, then you are making no mistake in this case, because in this case you should not believe that the store has pumpkins on sale. But in fact, in this case you are making a mistake; you should believe that the store has pumpkins on sale. So the view is too skeptical if x = 1. But if x < 1 the view is too credulous, entailing that we should believe that our lottery tickets are losers, even before the drawing takes place. Thus we are left with either an underspecified theory or a false theory—a theory underspecified due to the underspecified notion of fit, or a theory false due to being either too skeptical or too credulous. This is the standard theoretical pattern: we get either a counterexample-free theory that is uninformative, or an informative theory that has counterexamples. We could try to get out of the pattern by replacing the notion of fit with some sort of shiny new Bayesian machinery. But this strategy is unpromising: see Williamson (2000: 184–237) for some steps in its direction, and Whitcomb (2008) for a case that these steps cannot get us out of the pattern.

108  Dennis Whitcomb can somehow be solved, a third problem remains. Consider the process “exercising an infallible ability to know things.” Presumably, this is a knowledge-producing process on any adequate characterization of that notion. But then, we are threatened by the result that we should believe every truth, no matter how much (misleading) evidence we might possess against that truth. For if we should believe whatever it is that knowledge-producing processes would have us believe, and “exercising an infallible ability to know things” is a knowledge-producing process, then we should always believe every truth. Perhaps we can avoid this third problem by drawing on the well-known reliabilist notion of processes that are “available” to a given believer—processes that are in some sense readily at the believer’s disposal.20 Since normal human beings do not have processes like “exercising an infallible ability to know things” available to them, our process-focused approach is still a live option. Putting that approach into a precise theory now, and using the notion of an “available” knowledge-producing process, we might conjecture: Process Availability Duplication Theory (You should believe p) iff (p is believed by at least one of your up-to-now knowledge-producing process availability duplicates). This evades the problems involving counterfactuals, and the problems involving processes like “exercising an infallible ability to know things.” But it still isn’t a very good theory. It allows that two people might have dramatically different evidence—one might have all the evidence in the world that he has hands, and the other all the evidence in the world that he doesn’t—and still they should believe all the same things if they are knowledge-producing process availability duplicates. This point motivates the return of the notion of evidential duplication. Adding that notion to our current approach, we might conjecture: Up-to-now Evidential and Knowledge-Producing Process Availability Duplication Theory (You should believe p) iff (p is believed by at least one possible person who is your knowledge-producing process availability duplicate and your up-to-now evidential duplicate). We still aren’t out of the woods, though, because one of these duplicates of yours might believe things for terrible reasons. For instance, one of these duplicates of yours might believe, for terrible reasons, that he lacks hands. But then, our theory entails that you, right now, should believe that you lack hands. Maybe what is required for something to be what you should believe, then, is that it could be believed on the basis of a knowledge-producing process, by someone who

20

  See Goldman (1979).

Can There Be a Knowledge-First Ethics of Belief?  109 shares, and has always shared, your evidence. Putting this idea into our standard format, we get: Evidential Duplication and Knowledge Production Theory (You should believe p) iff (p is believed, via a knowledge-producing process, by at least one possible person who is your up-to-now evidential duplicate and your process availability duplicate). This is the last conjecture I’ll explore. How good is it? It certainly lacks various details. For instance, it lacks various details about the notion of a knowledge-producing process. Even supposing that we have decided how often a process must produce knowledge to count as a “knowledge-producing” process, it still remains to be settled how those processes are individuated. That is to say, it still remains to be settled what the conditions are under which one use of a knowledge-producing process counts as a use of the same knowledge-producing process as does another use of a knowledge-producing process. Suppose that you and I both listen to a mathematician tell us something of the form “the mathematical community is absolutely certain that p,” where “p” is replaced by one sentence for me, and a different sentence (expressing a different proposition) for you. Suppose we both trust her, so that I come to form the belief expressed by the sentence replacing “p” in my case, and you come to form the belief expressed by the sentence replacing “p” in your case. Did we use the same knowledge-producing process in coming to form our beliefs? If we didn’t, or if it is in principle possible that we didn’t given the setup of the case, then what makes for (or could in principle make for) the difference? Such questions surely deserve answers. They are the analogues of the questions standard reliabilists must answer in order to solve the “generality problem”—the problem of individuating belief-producing processes (as opposed to knowledge-producing processes). It is not clear how we ought to answer these questions. Additionally, it is not clear whether our final conjecture can deal with all the problems of the other conjectures. In particular, it may not be able to deal with slow-thinking Watsons. When Watson does not realize that Smith is the murderer (or even consider that proposition), and this is only because Watson is too dim to appreciate the fact that his evidence inculpates Smith, does Watson have “available” to him any knowledge-producing processes that, compatibly with possessing the evidence he actually possesses, can lead to knowledge that Smith is the murderer? If so, then our final conjecture deals with the Watson cases; if not, not. So, does Watson have available to him the relevant sorts of knowledge-producing processes? I suppose that, in order to make our final conjecture work, we need “available” knowledge-producing processes to include (in Watson’s case) something that would or could get him to know that Smith is the murderer, and to exclude (in the case of everyday people) processes like “exercising an infallible ability to know things.” I am not sure how to define the notion of availability so as to yield these results.

110  Dennis Whitcomb

5.6 Conclusions We’ve found, in the evidential duplication and knowledge production theory, a conjecture that serves as a knowledge-first theory of what we should believe and (given the uniqueness principle) what we should withhold on, and that does not make any obvious mistakes. But it avoids those mistakes by leaving numerous important details for another day: details about how often a process must produce knowledge to count as “knowledge-producing,” about how it is that knowledge-producing processes are individuated, and about exactly which knowledge-producing processes are “available” to us, in the sense of availability conjured up by the theory. Perhaps these details can be filled in; perhaps not. Zooming out, and looking at the foregoing dialectic as a whole, what we see is that in a series of attempts to build up a knowledge-first ethics of belief, the detailed attempts have clear problems, and the attempts without clear problems lack details. This should not be at all surprising. It is the pattern we see over and over again in attempts to give necessary and sufficient conditions for things. It is the standard pattern. Although the discovery of this pattern here is not surprising, it is theoretically interesting—both for the ethics of belief, and for contemporary epistemology more generally. It is theoretically interesting for the ethics of belief, because it shows where (or at least, some of the places where) the knowledge-first approach needs more detail in order to be viable. And it is theoretically interesting for contemporary epistemology more generally, because it brings a measure of disconfirmation to a certain idea that is popular in certain circles.21 In particular, it brings a measure of disconfirmation to the idea that by taking the notion of knowledge-first, we can escape the frustrating patterns—that is, the patterns of informative theories facing counterexamples and counterexample-free theories that are uninformative—that we find in other more traditional approaches.

References Ballantyne, N. and Coffman, E. J. (2011). “Uniqueness, Evidence, and Rationality,” Philosophers’ Imprint 11(18): 1–13. Bird, A. (2007). “Justified Judging,” Philosophy and Phenomenological Research 74(1): 81–110. Coffman, E. J. (2010). “Is Justified Belief Knowledge? Critical Notice of Jonathan Sutton, Without Justification,” Philosophical Books 51(1): 1–21. Conee, E. and Feldman, R. (1985). “Evidentialism,” Philosophical Studies 41(8): 15–34. Conee, E. and Feldman, R. (2001). “Internalism Defended,” in Hilary Kornblith (Ed.), Epistemology: Internalism and Externalism. Cambridge: MIT Press, 231–60. DeRose, Keith. (2002). “Assertion, Knowledge, and Context,” Philosophical Review 111: 167–203.

 

  See Williamson (2000), Hossack (2007), and Sutton (2007).

21

Can There Be a Knowledge-First Ethics of Belief?  111 Feldman, Richard. (1986). “Having Evidence,” in David Austin (Ed.), Philosophical Analysis. Dordrecht: Kluwer Academic Publishers, 83–104. Feldman, Richard. (2005). “Justification is Internal,” in Matthias Steup and Ernest Sosa (Eds.), Contemporary Debates in Epistemology. Oxford: Blackwell, . Gerken, Mikkel. (2011). “Warrant and Action,” Synthese 178: 529–47. Gilbert, Joel. (2012). “Dreams of My Real Father:  A  Story of Reds and Deception,” Documentary Film. Goldman, Alvin. (1979). “What is Justified Belief?” in George Pappas (Ed.), Justification and Knowledge. Dordrecht: Reidel, 1–24. Goldman, Alvin. (2001). “Internalism Exposed,” in Hilary Kornblith (Ed.), Epistemology: Internalism and Externalism. Cambridge: MIT Press, 207–30. Goldman, Alvin. (2009). “Williamson on Knowledge and Evidence,” in Patrick Greenough and Duncan Pritchard (Eds.), Williamson on Knowledge. Oxford: Oxford University Press, 73–91. Greco, John. (2005). “Justification Is Not Internal,” in Matthias Steup and Ernest Sosa (Eds.), Contemporary Debates in Epistemology. Oxford: Blackwell, 257–69. Hawthorne, John and Stanley, Jason. (2008). “Knowledge and Action,” Journal of Philosophy 105(10): 571–90. Hossack, Keith. (2007). The Metaphysics of Knowledge. Oxford: Oxford University Press. Kelly, Thomas. (2006). “Evidence,” Stanford Encyclopedia of Philosophy, Edward N. Zalta (Ed.), URL = . Littlejohn, Clayton. (2012). Justification and the Truth Connection. Cambridge:  Cambridge University Press. Manley, David. (2007). “Safety, Content, Apriority, Self-Knowledge,” Journal of Philosophy 104(8): 403–23. Matheson, Jonathan. (2011). “The Case for Rational Uniqueness,” Logic and Episteme 2(3): 359–73. McGlynn, Aidan. (2012). “Justification as ‘Would-Be’ Knowledge,” Episteme 9(4): 359–74. Neta, Ram. (2008). “What Evidence Do You Have?” British Journal for the Philosophy of Science 59: 89–119. Reynolds, Steven. (2013). “Justification as the Appearance of Knowledge,” Philosophical Studies 163(2): 367–83. Schroeder, Mark. (2004). “The Scope of Instrumental Reason,” Philosophical Perspectives 18: 337–64. Shope, Robert. (1978). “The Conditional Fallacy in Contemporary Philosophy,” Journal of Philosophy 75(8): 397–413. Smithies, Declan. (2012). “The Normative Role of Knowledge,” Noûs 46(2): 265–88. Sutton, Jonathan. (2005). “Stick to What You Know,” Noûs 39(3): 359–96. Sutton, Jonathan. (2007). Without Justification. Cambridge, MA: MIT Press. Whitcomb, Dennis. (2008). “Williamson on Justification,” Philosophical Studies 138(2): 161–8. Whitcomb, Dennis. (2010). “Curiosity was Framed,” Philosophy and Phenomenological Research 83(1): 664–87. Whitcomb, Dennis. (2013). “One Wage of Unknowability,” Synthese 190: 339–52. White, Roger. (2005). “Epistemic Permisssiveness,” Philosophical Perspectives 19: 445–59. Williamson, Timothy. (2000). Knowledge and its Limits. Oxford: Oxford University Press.

6 Truth as the Fundamental Epistemic Good Duncan Pritchard

6.1 Introduction That truth is the fundamental epistemic good used to be orthodoxy within epistemology. Indeed, isn’t it simply characteristic of the epistemic that it is directed at truth? Relatedly, isn’t epistemology concerned with the good to which a properly conducted intellectual inquiry is directed, and isn’t the goal of such an inquiry truth?1 In the recent literature in epistemology, however, the idea that truth is the fundamental epistemic good, along with the related thesis that truth is the goal of inquiry, has increasingly been treated as dubious. Indeed, I think it is now fair to say that the new orthodoxy in epistemology is that truth is not the fundamental epistemic good.2 What is less clear, however, is just why there has been this radical shift in our epistemological thinking. In what follows I will be examining three considerations that one might offer as grounds for rejecting the truth thesis, and I will argue that they are all unpersuasive. The dialectical strategy of this chapter is thus explicitly negative and piecemeal. The former, because I will not be offering a positive defence of the claim that truth is the fundamental epistemic good, but rather arguing that the reasons proffered for rejecting this claim are unconvincing. The latter, because I will only be examining a selection of considerations which are held to count against this claim. Nonetheless, this should be sufficient to demonstrate that we should look again at this “old fashioned” conception of the epistemic good.   1  See David (2001: 151–2) for a long list of quotations illustrating how widely held this view once was in epistemology. The list of contemporary epistemologists that he quotes includes William Alston, Laurence Bonjour, Roderick Chisholm, Richard Foley, Alvin Goldman, Keith Lehrer, Paul Moser, Alvin Plantinga, and Ernest Sosa.   2  For two recent high-profile attacks on the idea of truth being the fundamental epistemic good, see DePaul (2001; cf. David 2001) and Kvanvig (2005; cf. David 2005).

Truth as the Fundamental Epistemic Good  113 In section 6.2, I spell out what I have in mind when I talk of truth as the fundamental epistemic good. In sections 6.3–6.5, I then proceed to consider the three considerations that I take to have led commentators to abandon this thesis and I argue that they are all unconvincing. In section 6.6, I offer some concluding remarks.

6.2  Truth as the Fundamental Epistemic Good When I say that truth is the fundamental epistemic good, I mean that from a purely epistemic point of view it is ultimately only truth that we should care about. Call this the truth thesis. The easiest way to further unpack what I take to be involved in the truth thesis is to see what consequences I draw from it. One type of consequence that I draw concerns epistemic value. To say that truth is the fundamental epistemic good is to be committed to the idea that there is only one finally epistemically valuable epistemic good and that this is truth, with all other epistemic goods being merely instrumentally epistemically valuable relative to truth. So, for example, on this view insofar as epistemic justification is an epistemic good, then it’s epistemic goodness is derivative on the epistemic goodness of truth, in the sense that we instrumentally epistemically value justification because we finally epistemically value truth. Elsewhere, I have characterized this view as epistemic value T-monism, in that: (i) it is a view about epistemic value specifically (that’s the “epistemic value” part); (ii) it says that there is just one finally epistemically valuable epistemic good (that’s the “monism” part); and (iii) it says that this finally epistemically valuable epistemic good is truth (that’s the “T” part).3 A key point to note about treating truth as a finally epistemically valuable good is that it does not follow that truth is finally valuable simpliciter. For example, it does not follow that the truth always generates some pro tanto goodness. The point is just that from a purely epistemic point of view truth is rightly to be valued non-instrumentally. But this is entirely consistent with arguing that the general value of truth is merely instrumental (or indeed, that it has no value at all). Perhaps, for example, the only value of truth from a non-epistemic point of view is practical value, which is presumably a form of instrumental value. I don’t think we should be surprised by this, nor is this point specific to the epistemic domain, since one can draw analogous conclusions about other domains. Suppose one argued that the fundamental aesthetic good is beauty, for example, such that the beautiful is finally aesthetically valuable. One is not thereby committed to claiming that the beautiful is finally valuable simpliciter. That is, one might draw this conclusion, but it would be a further step. In particular, it is open to one to argue that the beautiful, when assessed outwith a specifically aesthetic axis of evaluation, is only instrumentally

  3  See, for example, Pritchard, Millar, and Haddock (2010: ch. 1) and Pritchard (2011). Following Goldman (e.g., 1999, 2002; Goldman and Olsson 2009), this proposal is also known in the literature as veritism.

114  Duncan Pritchard valuable, or perhaps not valuable at all. Indeed, the point being made here is just an instance of a more general point that one cannot infer, without further argument anyway, that because something is good when assessed along one axis of evaluation then it is also good when assessed along another axis of evaluation, much less that it is good simpliciter.4 A second consequence that I draw from the truth thesis is that truth is the telos of a properly conducted intellectual inquiry. By this I mean that seeking truth is the constitutive goal of a properly conducted intellectual inquiry, in that if one is undertaking an activity which is not directed at seeking the truth, then one is not undertaking a properly conducted intellectual inquiry at all. Of course, it is consistent with this claim that properly conducted intellectual inquiries can have all sorts of other goals too, some of them non-epistemic. It is an interesting question how the claim that truth is the fundamental epistemic good and the claim that truth is the constitutive goal of a properly conducted intellectual inquiry are related. Clearly they are deeply entwined proposals, to the extent that it is hard to imagine someone defending the one without the other. I just characterized the latter thesis as being a consequence of the former thesis, but one could just as well argue that the former is a consequence of the latter (or indeed that they both spring from a common source). For our purposes, we can set this issue to one side, since all that matters is that these theses go hand-in-hand and that in defending the truth thesis I will be defending both of them. One final point about the truth thesis is in order. I have characterized this thesis in terms of the claim that truth is the fundamental epistemic good, but proponents of the truth thesis often characterize the view not in terms of truth simpliciter but rather specifically in terms of true belief.5 Read one way, this substitution of truth with true belief is harmless. In order to see this, consider again the idea that beauty is the fundamental aesthetic good, such that from a purely aesthetic point of view it is ultimately just the beautiful that one cares about. Now, the thought might be that just as in valuing beauty we thereby value, and thus seek out, those things (beautiful artworks, the aesthetic experience of beauty, etc.) which instantiate that property, so in valuing truth we thereby value, and thus seek out, items which have that property, such as true beliefs (which, from a first-person point of view at any rate, are the most obvious bearers of this property). On this reading, it is still truth that is the fundamental epistemic good.

  4  It is worth bearing in mind here Geach’s (1956) distinction regarding predicative and attributive adjectives. In particular, from “x is a big flea” it does not follow that “x is a flea” and “x is big”, since the claim being made is only the attributive claim that x is big for a flea. (Compare: “x is a red flea.”) Similarly, in claiming that something is, say, epistemically valuable, one is on the face of it just making the attributive claim that the item in question is valuable from a specifically epistemic point of view, and not the predicative claim that the item is both epistemic and valuable. I discuss the topic of epistemic value in a number of places, but see especially Pritchard (2007, 2011), Pritchard, Millar, and Haddock (2010: ch. 1), and Pritchard and Turri (2011).   5  See, for example, David (2001, 2005).

Truth as the Fundamental Epistemic Good  115 That is, on this reading saying that true belief is the fundamental epistemic good is basically a useful way of roughly paraphrasing, rather than redefining, the truth thesis. There could be something more radical going on here, however, in which case it’s worthy of note. Consider again the aesthetic analogy. One might argue that what is of fundamental aesthetic value is not beauty itself but rather that which we seek out as instantiating beauty. So the fundamental aesthetic good is not beauty but rather some common item which has that property, such as the aesthetic experience of beauty. Perhaps the thought is that we somehow unduly “reify” aesthetic value to think otherwise. Similarly, one might be tempted to advance an analogous claim in epistemology— viz., that, at risk of an undue reification, it is not truth that is of fundamental epistemic value, but rather some common item which has that property, such as true belief. If that is what is going on when truth is replaced by true belief as the fundamental epistemic good, then this is not a mere paraphrasing of the truth thesis, but a redefinition. My view is that the reasoning behind this replacement is mistaken. The aesthetic case is illustrative in this regard. From a purely aesthetic point of view, we surely value the aesthetic experience of beauty because we value beauty, which indicates that beauty plays a more fundamental axiological role here. Similarly, I would maintain that from a purely epistemic point of view we value true belief because we value truth.6 This is not the place to get into the merits or demerits of the truth versus true belief conceptions of the fundamental epistemic good. What is salient for present purposes is just that I am construing the truth thesis specifically in terms of truth and not true belief. (That said, I will comment further below on the issue of whether we should substitute truth with true belief on this score).7

6.3  Epistemic Value and the Value of the Epistemic One reason why contemporary epistemologists are inclined to reject the truth thesis concerns the apparent inability of proponents of this thesis to accommodate the widespread intuition that knowledge is more valuable than mere true belief. In particular, the   6  Here’s another reason to prefer true belief as the fundamental epistemic good to mere truth, pressed to me by John Turri. If one goes for the latter thesis, then what is to stop someone merely desiring the truth or hoping that certain propositions are true, while never actually seeking out true beliefs? But that would be akin to someone regarding beauty as the fundamental aesthetic good and yet eschewing the opportunity of actually experiencing beauty for no other reason than that she prefers to merely desire such beauty from afar, or hope that there is beauty in the world. That doesn’t look like a coherent state of affairs. Moreover, I take it as an advantage of the view that it is truth that is the fundamental epistemic good that it can account for the epistemic goodness of propositional attitudes other than belief. That is, the epistemically good agent will surely be someone who not only seeks out true beliefs and avoids false ones, but who also has various other propositional attitudes too, such as desiring that others should form their beliefs with similar epistemic care.   7  One might also try to motivate the truth thesis by appealing to the relationship between belief and truth. In particular, many commentators have suggested that truth is the telos of belief. For reasons of space, I will not be considering this issue here. I briefly discuss its relevance to questions of epistemic value in Pritchard (2011: §1). For more on the idea that belief is in some sense “aimed” at truth, see Wedgwood (2002), Shah (2003), and Shah and Velleman (2005).

116  Duncan Pritchard contention is that epistemic value T-monism is inconsistent with the claim that knowledge is more valuable than mere true belief, and thus that, since the latter is undeniable, the former, and with it the truth thesis, must go. This problem is known as the swamping problem. There are two premises to the swamping problem. The first is epistemic value T-monism, which we have just discussed in the last section. The second is a general claim about value. This holds that instrumental value gets “swamped” when the relevant non-instrumental good in question obtains. To take an oft-cited example, let us grant that coffee is ultimately valued for its sensory qualities—its taste, smell, appearance, quantity, and so on. Someone who values coffee in this way will also value those implements which are good means to successful coffee production, such as reliable coffee-making machines (that is, coffee-making machines which regularly produce good coffee). Here is the problem. For while we might well instrumentally value reliable coffee-making machines, on account of their propensity to deliver us great coffee, we don’t care about how a cup of coffee was produced once we have a great cup of coffee before us. To illustrate this point, consider two identical cups of coffee, identical in every sensory respect—same taste, same smell, same quantity, same appearance, and so on—but where only one of the cups of coffee was produced by a reliable coffee-making machine. Would it matter to us which of these two cups of coffee we were given? Put another way, would we be willing to pay more to have one of these cups of coffee over the other? I think it is pretty clear that we would not. The upshot of this is that we only value reliable coffee-making machines as a means to good coffee, but that once we’ve got the great coffee this “swamps” the value of having coffee produced by a reliable coffee-making machine.8 The thought is that what goes for coffee applies to true belief, at least if the conception of epistemic value offered by epistemic value T-monism is right. For take an epistemic standing like reliability. According to epistemic value T-monism, the epistemic value of a belief being formed via a reliable process is instrumental epistemic value relative to the good of truth, but if the truth is already gained—i.e., if our agent already has a true belief in the target proposition—then that it was formed via a reliable process cannot confer any additional value. In particular, as with the coffee cup case, a true belief formed via an unreliable process (or in an unjustified way, etc.,) will be no less epistemically valuable than a true belief formed via a reliable process (or in a justified way, etc.). But of course, if that’s right, then whatever epistemic standing one adds to true belief in order to get knowledge can confer no additional epistemic value either, and hence it follows on this view that knowledge cannot be epistemically more valuable than mere true belief. One way of thinking about the swamping problem is that it is a reductio of epistemic value T-monism, at least if one grants that it is undeniable that knowledge is more valuable than mere true belief. It’s certainly clear that lots of epistemologists have

 

  The coffee example is due to Zagzebski (2003).

8

Truth as the Fundamental Epistemic Good  117 conceived of the swamping problem in this way.9 But a closer inspection of what is going on here reveals that this problem is in fact entirely illusory. In order to see this, take another look at the swamping argument, which we can formulate as follows: The Swamping Problem (SP1) If the value of a property possessed by an item is only instrumental value relative to a further good and that good is already present in that item, then this property can confer no additional value on that item. (SP2) Epistemic properties are only instrumentally valuable relative to the good of truth. (SC)  Knowledge that p can be no more valuable than mere true belief that p. (SP1) is the general claim about value that we discussed earlier, and (SP2) is essentially a statement of epistemic value T-monism. And yet (SP1) and (SP2) seem to clearly entail the conclusion, (SC), which is in direct conflict with the widely held claim that knowledge is more valuable than mere true belief. So construed, one can see why this argument looks like a reductio of epistemic value T-monism. But there is a subtle sleight of hand in play here. In particular, we should recall the point made in the previous section that we cannot (without further argument anyway) infer from the fact that something is valuable (or disvaluable) when assessed along a certain axis of evaluation that it is valuable (or disvaluable) along another axis of evaluation, much less that it is valuable (or disvaluable) simpliciter. In particular, for our purposes, it does not follow from the fact that something is lacking in epistemic value that it is lacking in value simpliciter. In short, we should recognize the distinction between the value (or disvalue) of the epistemic and epistemic value (or disvalue). This point is important because the premises of the swamping argument do not in fact entail the conclusion as formulated as above. Instead, what they entail is this weaker conclusion: (SC*) Knowledge that p can be no more epistemically valuable than mere true belief that p. This conclusion is entirely compatible with the claim that knowledge is more valuable than mere true belief. Moreover, notice that this is not simply a matter of splitting   9  For some of the key statements of the swamping problem, see Jones (1997), Swinburne (1999), Riggs (2002a, 2002b), Kvanvig (2003, 2010), and Zagzebski (2003). Note, however, that, at least in early work on this issue, the swamping problem was usually thought of as impacting not on a particular conception of epistemic value but rather on specific theories of knowledge (usually reliabilism). I think it is clear in the contemporary literature that the problem is now recognized as being properly understood as being directed against a certain theory of epistemic value. I should note that in earlier work I understood the swamping problem as posing a reductio for epistemic value T-monism too. See Pritchard, Millar, and Haddock (2010: ch. 1) and Pritchard (2011). As the present article suggests, my view has shifted on this point. See Pritchard (forthcomingb) for an overview of how I respond to the swamping problem now.

118  Duncan Pritchard hairs. There are, after all, many different ways in which knowledge could be more valuable than mere true belief over and above the specifically epistemic axis of evaluation. Indeed, many of the leading proposals for why knowledge is more valuable than mere true belief do not appeal to anything specifically epistemic at all, but rather employ such considerations as the greater practical utility of knowledge (i.e., its greater practical value), the special role that knowledge plays in a life of flourishing (i.e., its greater ethical value), and so on.10 The defender of epistemic value T-monism can appeal to any one of these accounts of the special value of knowledge in order to explain why knowledge is more valuable than mere true belief without in any way contravening her commitment to epistemic value T-monism. Given that the truth of (SC*) is entirely compatible with the falsity of (SC), and thus compatible with the claim that knowledge is in general of greater value than mere true belief, is there anything stopping us from assenting to (SC*)? In particular, is the idea that knowledge is no better than mere true belief from a purely epistemic point of view worrisome, once we realize that it is entirely compatible with the idea that knowledge is more valuable simpliciter than mere true belief? I don’t see why it should be. Or, at least, I don’t see why it would be worrisome unless one had already rejected epistemic value T-monism, in which case the putative counter-intuitiveness of (C*) could hardly constitute an argument against epistemic value T-monism. There is a related issue in play when it comes to epistemic value T-monism, which is that opponents of this proposal often characterize the view as being opposed to there being a range of ways in which epistemic standings can be good. The problem is the “monism” part of the thesis, since it has connotations of offering an unduly narrow and restrictive conception of the value of a domain.11 But this is yet another sleight of hand. In one sense, of course, epistemic value T-monism is opposed to a kind of value pluralism about the epistemic domain, since it is opposed to pluralism about the fundamental epistemic goods of that domain. But as we just saw in our discussion of the swamping problem, there is nothing at all to prevent the proponent of epistemic value T-monism from arguing that epistemic standings are valuable in lots of different and interesting ways, such as by having a practical or ethical value. This would explain why knowledge has a special importance to epistemologists, but it wouldn’t be at all in conflict with epistemic value T-monism. Indeed, as I  have argued elsewhere, we severely impoverish the debate about the value of epistemic standings like knowledge if we confine our attentions to the swamping problem, for the important issue is not whether knowledge has greater epistemic value than mere true belief, but rather the more general question of

  10  For a recent example of a defence of the special value of knowledge that is of the first sort, see Millar (2011); for a recent example of a defence of the special value of knowledge that is of the second sort, see Greco (2009).   11  This is a guiding theme of Kvanvig’s work on this topic, for example, which has been extremely influential in contemporary epistemology. See, for example, Kvanvig (2003, 2005). See also DePaul (2001).

Truth as the Fundamental Epistemic Good  119 whether knowledge (along with other factive epistemic standings, such as understanding) is more valuable than mere true belief.12 In fact, rather than offering a narrow and restrictive conception of the goodness of the epistemic, epistemic value T-monism is in fact a liberating thesis. If the goodness of the epistemic is to be entirely understood in terms of epistemic goodness, then that blurs out the many and varied ways in which the epistemic can be good. For example, as just noted, epistemic value T-monism is consistent with an account of the ethical value of knowledge. On the epistemic pluralist picture, in contrast, there is the danger that the special value of knowledge in play here would just be treated as another kind of fundamental epistemic value, and the specifically ethical dimension to this value would be lost. Rather than trying to characterize all the ways that the epistemic can be valuable as types of epistemic value, we should instead recognize the distinction between the value of the epistemic and epistemic value. Only in doing so—and in not being so quick to assume that when something epistemic is good, then it is good qua epistemic—can we properly account for the myriad ways in which the epistemic can be good.

6.4 Trivial Truths A second way in which the truth thesis has been attacked involves the problem that is putatively posed for this thesis by trivial truths. The reasoning goes something like this. If truth is the fundamental epistemic good, then it follows that we should value all truths, even the trivial ones, such as how many grains of sand there are on a certain section of a beach. But clearly we don’t value trivial truths, and hence it follows that truth cannot be the fundamental epistemic good.13 We can express this argument as follows: The Trivial Truths Problem (TP1) If truth is the fundamental epistemic good, then we should value all truths, even the trivial ones. (TP2)  We rightly do not value trivial truths. (TC)  Truth is not the fundamental epistemic good. Given the discussion of the previous section we should be immediately suspicious of this train of reasoning. After all, even if it is right that the truth thesis entails that we should value all truths, even the trivial ones, the value in play will be specifically epistemic value. As such, it is entirely compatible with the truth thesis that trivial truths are disvaluable in lots of ways (e.g., are of no practical utility, play no role in a life of

  12  See Pritchard, Millar, and Haddock (2010: ch. 1) for a detailed defence of why we should disentangle the general question of the value of the epistemic from the specific problem posed by the swamping problem.   13  Versions of this general line of argument abound in the contemporary epistemological literature. For a sample of high-profile endorsements of this reasoning, see DePaul (2001), Sosa (2001), and Goldman (2004). (The “sand” example is due to Sosa.)

120  Duncan Pritchard flourishing, and so on), and hence are of no value at all (or of vanishingly small value, or of disvalue), all things considered. That is, one could argue that we should modify our formulation of the trivial truths problem as follows: The Trivial Truths Problem* (TP1*) If truth is the fundamental epistemic good, then we should epistemically value all truths, even the trivial ones. (TP2*)  We rightly do not epistemically value trivial truths. (TC*)  Truth is not the fundamental epistemic good. Furthermore, one could contend that so construed (TP2*) is false, in that from a purely epistemic point of view we should value all truths, even trivial ones. More precisely, one could argue that all truths, even the trivial ones, have epistemic value (perhaps even pro tanto value) but claim that this is entirely compatible with the obvious truth of (TP2). For when it comes to trivial truths the epistemic value in play is outweighed by other considerations, such that the all things considered value of trivial truths is at most negligible, and typically either non-existent or negative.14 That said, one might be puzzled even by the more modest claim that the most trivial of truths is of epistemic value. I think the source of this puzzlement relates to the relationship between our conception of the fundamental epistemic good and the goal of inquiry that I discussed earlier. Recall that I claimed that the fundamental epistemic good is that to which properly conducted intellectual inquiry is directed, to the extent that an inquiry that was not directed at this end simply wouldn’t count as a properly conducted intellectual inquiry. Here is the worry. If truth is the goal of properly conducted intellectual inquiry, then doesn’t it follow that such an inquiry should be directed at all truths, even the trivial ones? But properly conducted intellectual inquiry is not aimed at all truths but only the significant (i.e., non-trivial) ones. Hence it follows that truth cannot be the goal of properly conducted intellectual inquiry. We can formulate this concern as follows: The Trivial Truths Problem** (TP1**) If truth is the constitutive goal of properly conducted intellectual inquiry, then properly conducted intellectual inquiry aims at all truths equally, even the trivial ones. (TP2**) Properly conducted intellectual inquiry does not aim at all truths equally, since it does not aim at trivial truths. (TC**)  Truth is not the constitutive goal of properly conducted intellectual inquiry. The idea is that the truth thesis, construed as entailing the claim that truth is the constitutive goal of inquiry, entails a commitment to (TP1**), and that it hence flounders on  

  For a defence of this general line of argument, see Kvanvig (2008).

14

Truth as the Fundamental Epistemic Good  121 (TP2**). The upshot is that insofar as the truth thesis entails that truth is the constitutive goal of inquiry, then it must be rejected. Whereas arguing that even trivial truths have epistemic value is at least a viable way of responding to the previous formulation of the trivial truths problem, it is hard to see how an analogous claim would help here. For note that the analogous claim would presumably be that a properly conducted intellectual inquiry should be indifferent as to whether the truths it uncovers are trivial or profound, and that hardly seems plausible. Indeed, the exact opposite seems to be the case, in that a properly conducted intellectual inquiry would surely focus on the weighty truths at the expense of the trivial ones. The problem ramifies. For example, insofar as the trivial truths are easier to uncover, as presumably they are, then shouldn’t the efficient intellectual inquirer focus her attentions on them to the exclusion of the profound truths (e.g., shouldn’t the efficient inquiry seek out the information contained in obscure phone books)? Furthermore, presumably some of the trivial truths are the kinds of truths that could be uncovered by pretty much anyone at any time. For instance, consider those trivial truths that are easily gleaned by anyone with an internet connection. Does it follow that insofar as one has an internet connection, then one is under some sort of prima facie cognitive obligation to uncover these truths (i.e., if one has nothing better to do), and thereby promote the epistemic good? Fortunately for the proponent of the truth thesis, however, there is a subtle confusion in play here, one that has been skillfully exposed in recent work by Nick Treanor (2012; 2013). For as he points out, it doesn’t follow from the fact that all truths are epistemically good that all truths are equally epistemically good, such that even trivial and significant truths need to be of equal weight from the perspective of epistemic value. The claim that inquiry should aim at “weighty” truths rather than trivial truths is thus entirely compatible with the idea that truth is the constitutive goal of inquiry. As Treanor points out, the style of argument in play in the trivial truths problem as just formulated is essentially as follows: If inquiry aims at truth, then it aims equally at every truth (even the trivial ones); but inquiry does not aim equally at every truth, and so inquiry does not aim at truth. In order to demonstrate that this reasoning is faulty, Treanor offers us the following example of reasoning which is analogous, and yet which is clearly defective: If gold mining aimed for gold, it would aim at every piece of gold equally—every piece of gold is equally gold, after all. But gold mining aims for flakes more than for dust, for nuggets more than for flakes, and for great veins more than anything else. So it does not aim for gold. (Treanor 2013)

In the same way, that inquiry aims at truth does not entail that it aims at all truths equally; instead, it is entirely in keeping with inquiry aiming at truth that it focuses on the important truths at the expense of the trivial ones. The moral is that (TP1**) is false.

122  Duncan Pritchard Now one might initially baulk at this line of reasoning, on the grounds that what differentiates the profound and the trivial truths is not their truth, in that they are both equally true. Accordingly, so the response goes, the analogy with gold breaks down, since we are not being offered a “flake” of truth and a “rich vein” of truth and being asked to choose between them, but simply two individual propositions both of which are true. But as Treanor shows, this way of thinking about the choice that inquiry must make between profound and trivial truths, while superficially compelling, is in fact deeply flawed. Suppose for the sake of argument that one is faced with a straight choice between two propositions, one of which is trivial—regarding the number of grains of sand on a section of a beach, say—and one of which is profound—a fundamental scientific truth, say. While it is undoubtedly correct that there are just two competing true propositions on offer, it does not follow that both propositions offer us an equal amount of truth. A single proposition, after all, can contain compressed within it a whole body of information, and in this sense present us with more truth than a competing proposition that is not so informative, even if both propositions are equally true. This is just what we would expect to be the case when it comes to choosing between a profound truth and a trivial one. A true statement of fundamental science offers us more truth than a true statement regarding grains of sand on the beach. We know more by knowing the true statement of fundamental science than we do by knowing the true statement regarding grains of sand on the beach. Accordingly, if one’s goal is truth, then one will opt for the profound truth every time. Truth is thus more like flakes and nuggets of gold than we realize. To further illustrate this point, consider the following sentences: (1) The cat sat on the mat. (2) The black cat sat on the mat. (3) The black cat sat on the red mat. (4) The fat black cat sat on the red mat. (5) The fat black cat sat on the square red mat. (6) The fat black cat sat on the heavily-worn square red mat. (7) The fat black cat sat on the large heavily-worn square red mat. It should be clear that one could continue this series pretty much indefinitely (although the sentences would become rather cumbersome very quickly). Let us stipulate that all these sentences are true. If we “weigh” truth exclusively in terms of whether the proposition expressed by a sentence is true, then we would be committed to saying that all of these sentences contain an equal amount of truth. And yet it is clearly the case that (7) offers us more truth than (1), and that a responsible truth seeker would prefer (7) to (1) for just that reason. As before, we can put this point in terms of knowledge: we know more by knowing (7) than we do by knowing (1). Indeed, a responsible truth seeker would surely prefer (7) even to these two true sentences: (*)  The cat sat on the mat. The cat was black.

Truth as the Fundamental Epistemic Good  123 But if we are to weigh truth in terms of counting true propositions, then the reverse should be the case, in that (*) will give us two “units” of truth versus (7)’s one. And yet it should be manifest that in knowing (*) we actually know less than by knowing (7). So much the worse for the idea that we should weigh truth by counting true propositions.15 As a final point on this topic, it is worth noting that this claim about weighing truths has a bearing on the issue of whether we should construe the truth thesis in terms of truth or in terms of true beliefs. Opting for the latter has a drawback, in that it counsels one to maximize truth in one’s beliefs, which is naturally rendered as advising one to get as many true beliefs as possible. But this is a mistake. We want to maximize truth, but the upshot of the foregoing is that we cannot measure truth in terms of true propositions, still less by counting how many true beliefs we have in those propositions. Note that the point here is not that proponents of the “true belief ” rendering of the truth thesis cannot accommodate any alternative conception of how to weigh truth. The point is rather that the focus on true belief as opposed to truth serves to obscure the fact that it is a mistake to weigh truth by counting true propositions. This is yet another reason to conceive of the fundamental epistemic good as truth rather than true belief.

6.5 Inquiry-Stoppers This brings us to the third and final consideration that we will be looking at which is often offered against the truth thesis. Here is the general line of thinking. From a purely epistemic point of view, it does not suffice to legitimately close inquiry that one has the truth; rather one needs to in addition have properly established that one has the truth.16 Hence, it follows that it is not truth that is the goal of inquiry, but rather whatever epistemic standing that is required in order for one to have properly established that one has the truth. We can express this argument as follows: The Inquiry-Stopper Problem (IP1) If truth is the constitutive goal of properly conducted intellectual inquiry, then truth is what legitimately closes such an inquiry. (IP2)  Truth is not what legitimately closes properly conducted intellectual inquiry. (IC)  Truth is not the constitutive goal of properly conducted intellectual inquiry. Let’s start by examining the case for (IP2). Typically—these days anyway—the epistemic standing that is held to be required to legitimately close inquiry is knowledge. Mere true belief is not enough to legitimately close inquiry, so the argument goes, since one can gain a true belief in all manner of inappropriate ways, and one

  15  For a subtle discussion of how we should weigh truth once we move away from the (clearly false) model on which we simply count true propositions, see Treanor (2013).   16 The caveat that we are talking about what closes inquiry from a purely epistemic point of view is of course vital, since there might be all kinds of non-epistemic reasons to legitimately close an inquiry (lack of time, opportunity cost, and so on). In what follows, I will take this caveat as implicit.

124  Duncan Pritchard should continue to inquire until the true belief in question is properly epistemically grounded.17 Only then will one have established that one has the truth which one seeks. And an epistemic standing that falls short of knowledge won’t suffice either, and for the same reason, since it wasn’t yet established that one has the truth that one seeks. The obvious case in point here is a Gettierized justified true belief. Of course, when one’s belief is Gettierized one takes oneself to have knowledge (and with good reason). That one would think it appropriate to close one’s inquiry at this point is thus evidence in favor of the idea that it is knowledge that legitimately closes inquiry. Furthermore, that one would reopen the inquiry upon discovering the Gettierization, and so discovering that one did not know after all, is also evidence in favor of the claim that knowledge is the “inquiry-stopper.” It is only with knowledge, then, that we have an epistemic purchase on the truth that suffices to legitimately close inquiry. Accordingly, it is concluded, knowledge is the goal of properly conducted intellectual inquiry.18 I do not dispute that what legitimately closes inquiry is an epistemic standing, though for what it’s worth I don’t think that knowledge is a very plausible candidate. More precisely, while I grant that inquiry can sometimes be legitimately closed by knowledge, I would maintain that often what is required to legitimately close inquiry is a more elevated epistemic standing—understanding—where this too is conceived of as a factive notion, but one that epistemically demands more of the inquirer than mere knowledge.19 The reason for this is that knowledge can be gained in a fashion such that one lacks understanding of what one knows, and in such cases I think a properly conducted intellectual inquiry will not close but rather continue until understanding   17  Of course, since we’re here talking specifically about a “properly conducted” intellectual inquiry, it’s actually not obvious that such an inquiry could lead to mere true belief, but let’s set that issue to one side.   18  For a particularly clear and recent statement of this line of objection to the truth thesis, one that explicitly substitutes knowledge as the goal of inquiry for truth, see Millar (2011), to which Kvanvig (2011) is a response. Here is Millar: Suppose that you are inquiring into whether something is so. Your aim is to find out whether it is so. Since finding out is nothing less than coming to know, what you aim for is knowledge. (Millar 2011: 63) Millar goes on to say that this conception of the goal of inquiry is compatible with inquiry aiming at truth, but it is clear from how he clarifies what he means by this remark that he views the goal of inquiry as being knowledge of the truth rather than truth itself. For example, he remarks that “the aim [of inquiry] is to grasp the truth, and knowledge is the form that such grasping takes if the aim is achieved” (Millar 2011: 63). Note that Millar draws heavily on the so-called “knowledge-first” programme in epistemology as defended by Williamson (2000), though it’s unclear whether Williamson himself would endorse the idea that knowledge rather than true belief is the goal of inquiry. For some recent defences of the claim that knowledge legitimately closes inquiry, see Kappel (2010), Gelfert (2011), and Kelp (2011).   19  Actually, this isn’t quite right, since I also hold that understanding can sometimes be epistemically less demanding than knowledge, in that it is compatible with a form of epistemic luck that knowledge isn’t compatible with (though even then understanding is still a factive notion). For more on my conception of understanding, and how it relates to knowledge, see Pritchard (2009, forthcominga) and Pritchard, Millar and Haddock (2010: ch. 4). For some key alternative views on the epistemology of understanding, see Elgin (1996, 2004, 2009), Zagzebski (2001), Kvanvig (2003, 2009, 2012), and Grimm (2006, forthcoming).

Truth as the Fundamental Epistemic Good  125 is gained. It is, I claim, typically only understanding which sates the curiosity which stimulates inquiry. Consider, for example, someone who wonders what causes the movements of the tides, and who is told, by someone authoritative in this regard, that the movements of the tides are caused by the gravitational pull of the moon on the earth. Our agent thus comes to know that the movements of the tides are caused by the gravitational pull of the moon on the earth. Suppose further, however, that our hero has no real conception of how it could be that the moon’s gravitational pull could have this effect on the movements of the tides. While knowing the answer to the question she has posed, she doesn’t understand it. Should mere knowledge of the answer to her question suffice to close her inquiry in this regard? It seems not. Rather, she should continue to inquire until she understands why cause and effect are related in this way, and so understands the answer she has been given. In this way, I think that it is often the case that it is understanding, rather than knowledge, which legitimately closes inquiry. In any case, let us return to the objection that since what legitimately closes properly conducted intellectual inquiry is not truth, but rather something epistemic like knowledge, it follows that truth is not the goal of inquiry. Given that I grant that what legitimately closes inquiry is not truth but something epistemic (typically understanding), it should be clear that my objection to this line of argument concerns not the antecedent claim in this inference, but rather the inference itself. That is, it is not (IP2) which I reject in the inquiry-stopper problem, but rather (IP1). In particular, I do not think that it follows from the fact that what legitimately closes inquiry is something epistemic, like knowledge or understanding, that it is this epistemic standing, and not truth, which is the goal of inquiry. An analogy will be helpful here. Suppose our goal is to produce a great cup of coffee, and that we engage in certain practices to achieve that end, such as selecting beans, choosing a suitable coffee-making machine, reading up on the art of making a great cup of coffee, and so on. At the point at which our activities produce a great cup of coffee, then those activities will have been successful, in that they will have achieved their goal. But it would not be appropriate to conclude those activities at this point, since in order to do that one must first establish that this goal has been achieved, and that requires tasting (and thereby coming to know) that the coffee is great.20 Crucially, however, that knowing that great coffee has been produced is required to legitimately close a great-coffee-directed activity does not entail that this activity was really aimed at the production of knowledge of great coffee after all (or, for that matter, the tasting of great coffee etc.), rather than just great coffee. The tasting, after all, is just a way of establishing that this activity has successfully reached its conclusion. In the same way, that what legitimately closes inquiry might be knowledge, or some other (factive) epistemic standing like understanding, does not entail that what inquiry   20  It needn’t be you who tastes it of course; indeed, it might be better for a coffee aficionado to taste it and confirm that it’s great coffee.

126  Duncan Pritchard is really aimed at is this epistemic standing rather than truth. If one’s inquiry leads one to truth, then at that point one’s inquiry has been successful. But it is only in grasping that the truth has been attained—where this might involve knowledge, understanding, or some other factive epistemic standing, depending on one’s point of view—that one’s inquiry is legitimately closed. There is thus no tension between the idea that truth is the goal of inquiry and that what legitimately closes inquiry is something epistemic like knowledge or understanding. A final note is in order here. I mentioned in section 6.2 that the truth thesis is often formulated in such a way that it is not truth which is the fundamental epistemic good but rather true belief, and at various junctures I have offered considerations against this alternative construal of the truth thesis. Now one reason one might have for opting for a formulation of the truth thesis in terms of true belief is that properly conducted intellectual inquiry is directed at true belief rather than truth. I noted in section 6.2 that I think this is a mistake. For sure, we want inquiry to lead to true belief, but we only care about true belief because we care about truth (just as the aesthete considered above only cares about the aesthetic experience because she cares about beauty). The discussion of this section should have reinforced this idea. First, because even if there is a direct link between what is the goal of inquiry and what legitimately closes inquiry, this won’t lend support to the idea that true belief (as opposed to knowledge, or understanding) is the goal of inquiry. As we have seen, what legitimately closes inquiry, while not truth, is not true belief either. Second, and more importantly, because the discussion of this section reminds us that what is the goal of inquiry and what legitimately closes inquiry is not the same thing anyway. So even if it were right that it is true belief that legitimately closes inquiry, it wouldn’t follow that it is true belief and not truth which is the goal of inquiry.

6.6  Concluding Remarks I have considered three prominent reasons for rejecting the truth thesis and found them all wanting. Of course, this in itself doesn’t show that the truth thesis is correct, not least because there may be further arguments out there against the truth thesis that are more compelling than the ones I have considered. Moreover, I haven’t offered any positive defense of truth thesis. But my goal here has not been to demonstrate that the truth thesis is correct. Instead, I have been concerned merely to show that this is a thesis that is far more plausible than it is given credit in contemporary epistemology. In particular: (i) it can accommodate the widespread contention that knowledge is more valuable than mere true belief; (ii) it can accommodate, perhaps even better than alternative views, a pluralism about the good of the epistemic (even though it is incompatible with pluralism about the epistemic good); (iii) it is not affected by the problem putatively posed by trivial truths, since that problem trades on a dubious conception of how one should “weigh” truth; and (iv) it is entirely compatible with the idea that what

Truth as the Fundamental Epistemic Good  127 legitimately closes inquiry is not truth but rather a particular kind of factive epistemic relationship to the truth, such as knowledge and understanding.21

References Brady, M. S., and Pritchard, D. H. (Eds.). (2003). Moral and Epistemic Virtues. Oxford: Blackwell. David, M. (2001). “Truth as the Epistemic Goal,” in M. Steup (Ed.), Knowledge, Truth, and Duty: Essays on Epistemic Justification, Virtue, and Responsibility. Oxford: Oxford University Press, 151–69. David, M. (2005). “Truth as the Primary Epistemic Goal: A Working Hypothesis,” in E. Sosa and M. Steup (Eds.), Contemporary Debates in Epistemology. Oxford: Blackwell, 296–312. DePaul, M. (2001). “Value Monism in Epistemology,” in M. Steup (Ed.), Knowledge, Truth, and Duty: Essays on Epistemic Justification, Virtue, and Responsibility. Oxford: Oxford University Press, 170–86. Elgin, C. (1996). Considered Judgement. Princeton, NJ: Princeton University Press. Elgin, C. (2004). “True Enough,” Philosophical Issues 14: 113–31. Elgin, C. (2009). “Is Understanding Factive?” in A. Haddock, A. Millar, and D. H. Pritchard (Eds.), Epistemic Value. Oxford: Oxford University Press, 322–30. Geach, P. T. (1956). “Good and Evil,” Analysis 17: 32–42. Gelfert, A. (2011). “Expertise, Argumentation, and the End of Inquiry,” Argumentation 25: 297–312. Goldman, A. (1999). Knowledge in a Social World. Oxford: Oxford University Press. Goldman, A. (2004). Pathways to Knowledge:  Private and Public. Oxford:  Oxford University Press. Goldman, A. and Olsson, E. J. (2009). “Reliabilism and the Value of Knowledge,” in A. Haddock, A. Millar, and D. H. Pritchard (Eds.), Epistemic Value. Oxford: Oxford University Press, 19–41. Greco, J. (2009). “The Value Problem,” in A. Haddock, A. Millar, and D. H. Pritchard (Eds.), Epistemic Value. Oxford: Oxford University Press, 313–21. Grimm, S. (2006). “Is Understanding a Species of Knowledge?” British Journal for the Philosophy of Science 57: 515–35. Grimm, S. (forthcoming). “Understanding as Knowledge of Causes,” in A. Fairweather (Ed.), Virtue Scientia: Virtue Epistemology and Philosophy of Science. Dordrecht, Holland: Springer. Jones, W. (1997). “Why Do We Value Knowledge?” American Philosophical Quarterly 34: 423–40. Kappel, K. (2010). “On Saying that Someone Knows:  Themes from Craig,” in A. Haddock, A. Millar, and D. H. Pritchard (Eds.), Social Epistemology. Oxford:  Oxford University Press, 69–88. Kelp, C. (2011). “What’s the Point of ‘Knowledge’ Anyway?” Episteme 8: 53–66. Kvanvig, J. (2003). The Value of Knowledge and the Pursuit of Understanding. Cambridge: Cambridge University Press.   21  Thanks to Allan Hazlett and Nick Treanor for helpful discussions of topics related to this chapter. Special thanks to Jon Matheson and John Turri who each read and commented on an earlier incarnation. An earlier version of this chapter was presented in June 2012 at the “Dimensions of Normativity” conference at Goethe University, Frankfurt. I am grateful to the audience on this occasion, and especially Marcus Willaschek, who commented on my chapter at length.

128  Duncan Pritchard Kvanvig, J. (2005). “Truth is not the Primary Epistemic Goal,” in E. Sosa and M. Steup (Eds.), Contemporary Debates in Epistemology. Oxford: Blackwell, 285–96. Kvanvig, J. (2008). “Pointless Truth,” Midwest Studies in Philosophy 32: 199–212. Kvanvig, J. (2009). “The Value of Understanding,” in A. Haddock, A. Millar, and D. H. Pritchard (Eds.), Epistemic Value. Oxford: Oxford University Press, 95–112. Kvanvig, J. (2010). “The Swamping Problem Redux: Pith and Gist,” in A. Haddock, A. Millar, and D. H. Pritchard (Eds.), Social Epistemology. Oxford: Oxford University Press, 89–111. Kvanvig, J. (2011). “Millar on the Value of Knowledge,” Proceedings of the Aristotelian Society, suppl. vol. 85: 83–99. Kvanvig, J. (2012). “Curiosity and a Response-Dependent Account of the Value of Understanding,” in T. Henning and D. Schweikard (Eds.), Knowledge, Virtue, and Action. London: Routledge, 151–74. Millar, A. (2011). “Why Knowledge Matters,” Proceedings of the Aristotelian Society, suppl. vol. 85: 63–81. Pritchard, D. H. (2007). “Recent Work on Epistemic Value,” American Philosophical Quarterly 44: 85–110. Pritchard, D. H. (2009). “Knowledge, Understanding and Epistemic Value,” in A. O’Hear (Ed.), Epistemology (Royal Institute of Philosophy Lectures). Cambridge:  Cambridge University Press, 19–43. Pritchard, D. H. (2011). “What is the Swamping Problem?” in A. Reisner and A. Steglich-Peterson (Eds.), Reasons for Belief. Cambridge: Cambridge University Press, 244–59. Pritchard, D. H. (2012). “Anti-Luck Virtue Epistemology,” Journal of Philosophy 109: 247–79. Pritchard, D. H. (forthcominga). “Knowledge and Understanding,” in A. Fairweather (Ed.), Virtue Scientia:  Virtue Epistemology and Philosophy of Science. Dordrecht, Holland: Springer. Pritchard, D. H. (forthcomingb). “Veritism and Epistemic Value,” in H. Korblith and B. McLaughlin (Eds.), Alvin Goldman and His Critics. Oxford: Blackwell. Pritchard, D. H., Millar, A., and Haddock, A. (2010). The Nature and Value of Knowledge: Three Investigations, Oxford: Oxford University Press. Pritchard, D. H., and Turri, J. (2011). “Knowledge, the Value of,” Stanford Encyclopaedia of Philosophy, Edward N. Zalta (Ed.), URL  =  . Riggs, W. (2002a). “Beyond Truth and Falsehood:  The Real Value of Knowing that P,” Philosophical Studies 107: 87–108. Riggs, W. (2002b). “Reliability and the Value of Knowledge,” Philosophy and Phenomenological Research 64: 79–96. Shah, N. (2003). “How Truth Governs Belief,” Philosophical Review 112: 447–83. Shah, N., and Velleman, D. (2005). “Doxastic Deliberation,” Philosophical Review 114: 497–534. Sosa, E. (2001). “For the Love of Truth?” in A. Fairweather and L. Zagzebski (Eds.), Virtue Epistemology:  Essays on Epistemic Virtue and Responsibility. Oxford:  Oxford University Press, 49–62. Treanor, N. (2012). “Trivial Truths and the Aim of Inquiry,” Philosophy and Phenomenological Research. Treanor, N. (2013). “The Measure of Knowledge,” Noûs 47(3): 577–601. Wedgwood, R. (2002). “The Aim of Belief,” Philosophical Perspectives 16: 268–97.

Truth as the Fundamental Epistemic Good  129 Williamson, T. (2000). Knowledge and Its Limits. Oxford: Oxford University Press. Zagzebski, L. (2001). “Recovering Understanding,” in M. Steup (Ed.), Knowledge, Truth, and Obligation:  Essays on Epistemic Justification, Virtue, and Responsibility. Oxford:  Oxford University Press, 235–52. Zagzebski, L. (2003). “The Search for the Source of the Epistemic Good,” Metaphilosophy 34: 12– 28; and reprinted in Brady and Pritchard (2003), 13–28.

7 Wide-Scope Requirements and the Ethics of Belief Berit Brogaard

William Kingdon Clifford proposed a vigorous ethics of belief, according to which you are morally prohibited from believing something on insufficient evidence. Though Clifford offers numerous considerations in favor of his ethical theory, the conclusion he wants to draw turns out not to follow from any reasonable assumptions. In fact, I will argue, regardless of how you propose to understand the notion of evidence, it is implausible that we could have a moral obligation to refrain from believing something whenever we lack sufficient evidence. I will argue, however, that there are wide-scope conditional requirements on beliefs but the beliefs in question need not be beliefs for which we have sufficient evidence. I then argue that we are epistemically, but not morally, required to form epistemically valuable beliefs. However, these beliefs, too, need not be beliefs for which we have sufficient evidence.

7.1  Moral Requirement and Beliefs When Clifford proposed his ethics of belief, what he had in mind was the conditions under which we are morally required to form or withhold belief. His stern view of the ethics of belief is that “It is wrong always, everywhere, and for anyone to believe anything on insufficient evidence” (1877: 295). One of his most famous arguments for this view originates in the case of the ship owner: Ship Owner A cruise owner selling cruise tickets discovers that something is wrong with one of his cruise ships. Being desperately in need of money, he manages to suppress his concerns and form the “sincere and comfortable conviction that his vessel was thoroughly safe and seaworthy” (1877: 70). He continues selling tickets. The ship goes down in mid-ocean, and he collects the insurance money.

Wide-Scope Requirements and the Ethics of Belief  131 Clifford argues that the ship owner is responsible for the death of the passengers on that ship because he believed the ship was safe on insufficient evidence. Clifford argues that even if we changed the ending of the story, and the ship arrived safely at its destination, the ship owner would still be to blame for believing on insufficient evidence. We might articulate Clifford’s argument more formally as follows: (1) When you believe on insufficient evidence, your belief is likely to be false. (2) False beliefs could have morally harmful consequences. (3) So, when you believe on insufficient evidence, you may be the author of morally harmful consequences. (4) You should attempt to avoid being the author of morally harmful consequences (5) So, you ought to believe only on sufficient evidence. Clifford himself does not articulate the argument underlying his case and his views. But it seems plausible that this sort of argument was what he had in mind. This argument, however, is unsound. While it is true that the ship owner in Clifford’s example was to blame for not repairing the ship before departure, it is not clear that examples like these show that one should never believe something on insufficient evidence. Consider the following counter-case:1 Evil Doctor A very evil doctor, who has brain cancer, invents a medication that can cure cancer and has no side effects. Being extremely evil the doctor intends to cure his own cancer but also intends not to share the results with the rest of the world. He cures his own cancer but the small brain lesions that remain from the tumor cause him to believe the medication is a poison that will make people undergo a slow and painful death. He gains access to the water supplies for the New York area and releases large quantities of the drug into the drinking water. As a consequence of his actions, he saves the lives of thousands of people who were otherwise going to die from cancer. In this case believing on insufficient evidence leads to a very good outcome. Based on a case like this, we could formulate an argument analogous to the one above as follows: (1) When you believe on insufficient evidence, your belief is likely to be false. (2) False beliefs could have morally good consequences. (3) So, when you believe on insufficient evidence, you may be the author of morally good consequences. (4) You should attempt to be the author of morally good consequences. (5) So, you ought to believe only on insufficient evidence. The argument’s conclusion, of course, is absurd. But it helps us identify what is wrong with Clifford’s argument. The problem with the argument, I think, lies with the move   1 If the analogy needs strengthening, we can envisage that the scientist suppresses his belief to make money.

132  Berit Brogaard from (3) and (4) to (5). While (3) and (4) may be true, the move itself is invalid. You could be the author of morally good or bad consequences, when you believe on insufficient evidence, but it doesn’t follow from (3) and (4) that you ought to believe only on sufficient (or insufficient) evidence. Clifford’s argument, as I have reconstructed it, is unsound. But the question remains whether there is a different route to the conclusion that it is always wrong to believe on insufficient evidence. One proposal of how to get to something like Clifford’s conclusion is to start off with the assumption that we are epistemically required to believe anything only on sufficient evidence. While it is still an open question whether we are morally obliged to believe only on sufficient evidence, it is prima facie plausible that we are epistemically required to believe only on sufficient evidence. This claim is implied by Richard Feldman’s (2000) version of evidentialism: For any proposition p, time t, and person S, if S has any doxastic attitude at all toward p at t, then S epistemically ought to have at t the attitude towards p supported by S’s evidence at t. I will question this position below. But I will grant it here for argument’s sake. Given this assumption it may be thought that we can get to Clifford’s conclusion in the following way (Chignell 2010): (1) We have an epistemic obligation to possess sufficient evidence for all of our beliefs. (2) We have a moral obligation to uphold our epistemic obligations. (3) Thus, we have a moral obligation to possess sufficient evidence for all of our beliefs. While initially compelling, the argument is unsound. As I will argue below, the first premise is false. But let it be granted, for argument’s sake at least, that premise 1 is true. The question then becomes whether there are grounds for believing premise 2. As it turns out, premise 2, given the truth of premise 1, is false. It is not hard to find a counterexample to premise 2. Consider the following case: Belief Detector Mary, a successful journalist for a major newspaper, has strong evidence for believing that the tea party movement, were it to gain too many supporters, would seriously damage the welfare of the country. One of her former friends, now a supporter of the tea party movement and also a successful inventor, has just invented a belief detector that can measure what people believe and whether people have successfully suppressed their beliefs. One fateful day he shows up at Mary’s office and threatens to kill her 3-year-old daughter and make it look like an accident, unless she suppresses all of her beliefs about the tea party movement.

Wide-Scope Requirements and the Ethics of Belief  133 In this case Mary has sufficient evidence for believing that the tea party movement, were it to gain too many supporters, would seriously damage the welfare of the country. So, if premise 1 is true, she is epistemically required to hold this belief. However, if she continues to hold the belief, then her daughter will likely die, and she and her daughter’s father will be miserable for the rest of their lives. If she suppresses her belief, on the other hand, nothing of consequence will happen (let’s suppose). So, morally speaking, Mary ought to suppress her belief, which is to say, she does not have a moral obligation to uphold her epistemic obligations. If the first premise of the argument is true, as we have assumed, then the second premise is false. So the argument is unsound. It may be thought that we can still maintain a version of Clifford’s thesis if we draw a distinction between prima facie and ultima facie moral requirements.2 For example, one might hold that you have a prima facie moral requirement not to lie but that this prima facie moral requirement can be overridden, for example, in circumstances in which lying saves someone’s life. Given the distinction between prima facie and ultima facie moral requirements, one might offer the following variation on Clifford’s thesis: It is prima facie, but not necessarily ultima facie, morally wrong to believe something on insufficient evidence.

I have two worries about this weakened principle. First, it plainly is not a way to read Clifford’s thesis. He clearly specifies that believing on insufficient evidence is wrong always, everywhere and for anyone. To say that it is prima facie wrong to believe anything on insufficient evidence is to say that there potentially are circumstances in which it is not wrong to believe something on insufficient evidence. So the weakened principle entails the negation of the thesis that we have a categorical moral requirement to believe only on sufficient evidence. It is, for this very reason, inconsistent with Clifford’s evidentialism. Second, I doubt that prima facie moral requirements are very informative. We have a prima facie moral requirement not to lie. This means, roughly, that in the absence of overruling moral considerations, we should not lie. But it is also true then that it is prima facie morally permissible to lie. It is certainly true that in the absence of overruling moral considerations, lying is permissible. So prima facie lying is both permissible and impermissible. Any act that is merely prima facie morally impermissible is also prima facie morally permissible. So prima facie moral requirements by themselves cannot be action guiding. They are, therefore, uninformative.

7.2  Wide-Scope Moral Requirements There is a potential worry about the very idea of there being moral requirements pertaining to belief. Moral requirements are supposed to guide action, not belief

 

2

  For more on this distinction, see Shafer-Landau (1997).

134  Berit Brogaard formation. Believing by itself as opposed to expressing what you believe or acting on what you believe does not seem to be the kind of thing that could be morally impermissible.3 Clifford’s claim that ‘no one man’s belief is in any case a private matter which concerns himself alone’ appears to be false (Clifford 1877: 292). Belief alone does not lead to action and hence does not lead to, or constitute, moral wrongdoing. But suppose now that you hold a desire that together with your belief will motivate you to act on your belief.4 It is exceedingly plausible that if you are morally required not to do A and you believe B leads to A, then you are morally required not to do B. So if you act on your belief-desire pair, and your action is morally wrong, then you ought to either give up your desire to do B or you belief that B leads to A. Consider the following case: Envious Dean You are the dean of the College of Arts and Sciences. You don’t want your successful younger colleague to get tenure because you are envious of her achievements, so you come to believe mistakenly that you ought to deny her tenure owing to flaws in her factual record. As a dean you can seriously affect your colleague’s chance of getting tenure. Your belief and your desire together motivate you to deny your colleague tenure. So you deny your colleague tenure. Because of the bad job market, your colleague never finds a new position and must live from adjunct job to adjunct job. Her kids are starving and soon enough a bankruptcy trustee takes possession of her non-exempt property. In this case, it seems right to say that you were morally required to give up either your desire or your belief. The most natural reaction to this case probably is to morally blame you for your lack of evidence for your belief. But notice that the only reason we would morally blame you for your lack of evidence for your belief is that you were motivated to act on your belief. If you had believed you ought to deny your colleague tenure because of flaws in her record but had had an overruling desire not to deny her tenure, you would be a strange bird, perhaps even subject to epistemic blame, but you wouldn’t have been motivated to act on your belief, so you wouldn’t have been morally blameworthy. We can formulate these kinds of conditional moral requirements as wide-scope conditional requirements of the following form (where “belief ” refers to beliefs for which you have insufficient evidence and which are likely to lead to morally bad consequences if you act on them): You ought to (if you dominantly desire that p, then you do not believe that q).   3  Anti-Humeans hold that motivating reasons explain action (see, e.g., Korsgaard 1986). But I suppose no one would hold that belief by itself is sufficient for an action to occur (Smith 2010).   4  It’s very plausible that more than belief and desire is needed to motivate action (e.g., intention). Here I shall assume that belief and desire suffice for motivation. Nothing of consequence hinges on that assumption.

Wide-Scope Requirements and the Ethics of Belief  135 A dominant or overruling desire should here be understood as the strongest desire among all your desires, the desire that wins out when all is said and done. Consider a simple case: the waitress tells you that your desert options are Dark Chocolate Torte with vanilla ice cream, Chocolate Hazelnut Tartufo, or old-fashioned apple crisp with whipped cream. You have a sweet tooth and are having trouble deciding among the three delicious options. After much deliberation you decide on the Dark Chocolate Torte. In this case, you had a desire for any one of three desert options, but you had an overruling desire for the Dark Chocolate Torte. An overruling desire need not be one whose satisfaction is associated with any feelings of satisfaction (Smith 2010). An agent may “reluctantly” do something, which is to say that the satisfaction of the overruling desire lacks pleasure and perhaps causes the agent great pain (for example, leaving someone you love on the grounds that love just isn’t enough to make the relationship work can be immensely painful. But if you decide to leave a person, you nonetheless have an overruling desire to do just that). Wide-scope requirements are not reducible to narrow-scope requirements.5 For example, it seems plausible that while you shouldn’t commit burglary, if you do commit burglary, you should at least leave the sentimental items behind. However, suppose we express the conditional requirement as a narrow-scope requirement: If you commit burglary, then you ought to steal only the non-sentimental items Now, suppose you do commit burglary. By modus ponens it follows that you are morally required to steal only non-sentimental items from people’s houses. But you are not (unconditionally) morally required to steal anything. Formulating the requirement as a wide-scope requirement gets us out of this kind of bootstrapping: You ought to (if you commit burglary, then you steal only the non-sentimental items). When the antecedent is satisfied, narrow-scope requirements trivially entail wide-scope requirements. For example, the narrow-scope requirement, if you commit burglary, then you ought to steal only the non-sentimental items, entails the embedded wide-scope requirement: If you commit burglary, then you ought to (if you drink beer, then you steal only the non-sentimental items). But wide-scope requirements do not entail narrow-scope requirements. We cannot derive a moral requirement to steal from this wide-scope requirement. The wide-scope requirement can be satisfied in two ways: Either by not doing the action specified in the antecedent or by doing the action specified in the antecedent and the action specified in the consequence. The same applies to the desire-belief wide-scope conditionals. Wide-scope conditional requirements can be satisfied by giving up the desire or keeping the desire and giving up the belief. It is plausible, then, that belief is subject to wide-scope conditional   5  For recent defenses of this claim, see Broome (2000, 2001, 2002), Dancy (2000), and Wallace (2001). For earlier defenses, see Hill (1973), Greenspan (1975), and Darwall (1983).

136  Berit Brogaard moral requirements even if it is not in general governed by any unconditional moral requirements. Some examples (requiring the right kind of context): (1) You ought to (if you believe that women are less intelligent than men, you do not desire to act on this belief when serving on hiring committees). (2) You ought to (if you believe your ex-wife ought to die a slow and painful death because she found a new man, you do not desire to be an instigator of any action that could be a cause of her death). (3) You ought to (if you believe that your colleague’s new summer coat looks like a lab coat, you do not desire to tell her this). We cannot derive unconditional requirements from the conditional ones. For example, we cannot infer from the last of these conditional requirements that you have an unconditional moral obligation not to desire to tell your colleague that her new summer coat looks like a lab coat. The upshot is that while we do not have unconditional moral obligations with respect to beliefs, we have a lot of conditional ones. It may be objected that my argument presupposes a deontological or consequentialist ethical framework as opposed to a virtue-theoretical approach. A virtue ethicist may insist that your beliefs can be morally wrong even if you don’t act on them for the reason that they reflect your moral or intellectual character. This, however, is not quite right. A virtue ethics that does not provide action-guidance does not have anything to say about which actions are morally right or wrong. Such a theory would be a normative theory about people’s psychological states rather than their actions and hence wouldn’t be an ethical theory. For a virtue theory to be an ethical theory it must provide action-guidance, which is to say, it must give us a way of deciding which actions are right and wrong. Virtue-theoretical approaches do exactly that. Though virtue theorists may disagree about which character traits are virtuous, they all morally prohibit actions that are not grounded in virtue. Another objection to my argument turns on the tight connection between belief and action. It may be that we are psychologically unable to believe something without it eventually tainting our actions. This claim is a rather strong one but also one the truth of which is not completely unimaginable. For example, if I believe that men are better philosophers than women, this belief may eventually show up in my actions on hiring committees, editorial boards, or when making decisions about who to invite to the conferences I organize. If, however, all our beliefs eventually end up staining our actions, then perhaps there is a class of beliefs we ought to steer clear of. Beliefs in this class would be those that are likely to be accompanied by hidden desires, for example, racist or sexist beliefs. Tests like the Harvard Implicit Association Tests show that the vast majority of us have racist and sexist beliefs and that we are motivated to act on them. However, I do not think this objection succeeds. It is no doubt true that most of us have racist and sexist beliefs and that we are motivated to act on them but for the motivation to act on these beliefs to have any effect on action, it must derive from an

Wide-Scope Requirements and the Ethics of Belief  137 overruling desire or drive. Suppose a white woman chooses to take the stairs once she sees that a black man is waiting for the elevator. It’s not her racist beliefs alone that are motivating her actions but her racist beliefs together with an overruling desire (at that moment) not to be in close proximity of the black man. The point here is that beliefs alone are not enough to motivate us to act. When beliefs motivate us to, or make us, act, they are associated with overruling desires. So, it doesn’t follow that we ought to steer clear of certain types of belief. What does follow, though, is that we ought to either avoid holding these beliefs or forsake any overruling desires to act on them. There has been a long-standing concern about using deontological language in relation to belief.6 The concern is that there is no interesting sense in which we can control our beliefs. For example, if you see that I am holding a loaded gun in my hand, it would be exceedingly difficult, if not psychologically impossible, for you to give up the belief that I am holding a loaded gun in my hand. If “ought” implies “can,” then it makes no sense to ever say that we ought to reject or revise our beliefs.7 Or so the argument goes. One way of replying to this concern is to deny that “ought” implies “can.” Richard Feldman (2000), for example, argues that we can have obligations in virtue of taking on a certain job or playing a certain role. For example, we have obligations as parents, regardless of whether we can fulfill them. He thinks that we have epistemic obligations vis-à-vis belief in virtue of our role as believers. I will return to our epistemic objections below. What matters here is that if Feldman is right about role obligations, then it is plausible that we can have obligations qua intelligent beings that act in the world, and that those obligations are obligations not to both desire and believe certain things. Ultimately I don’t want to deny that “ought” implies “can.” Judging that a person ought to have fulfilled her role obligations seems to imply that she could have done it (perhaps by taking some really difficult steps). The “ought” and the “can” need not apply simultaneously. You can have an obligation to do A and be unable to do A as long as there was a time at which you could have ensured that you could now do A. For example, Kurt may have an obligation to be a provider for his children even after he has gambled away home and money and has no way of providing for his children. Though he no longer can provide for his children, he could have taken steps to ensure that he could now provide for his children. Granting this much does not affect my argument, however. I agree that there are many cases in which we cannot control our beliefs. But I think that most of us can control our desires over time. So if we have an obligation to give up a belief or not desire something, we can fulfill, or could have fulfilled, our obligation. In many cases we can also control belief, for instance, beliefs we dogmatically have as a result of our upbringing or culture. In this regard, holding a belief is different from being a certain height,   6  For instance, see Alston (1988) and Plantinga (1993) for arguments against; and Montmarquet (1986), Steup (2000), Ryan (2003), and Nottelmann (2006) for arguments for.   7  The “can” here is the “can” of practical ability.

138  Berit Brogaard to use Thomas Kelly’s example (Kelly 2002).8 Because our height is almost entirely outside of our control, it makes no sense to say that we ought to be a certain height.9 Had it been something we could change over time by, say, taking certain vitamins, it might have made sense to say things like “If you want to be a supermodel, you really ought to be a few inches taller.” Holding a belief may, in some cases, be on a par with having a certain weight. It makes perfect sense to say to an obese person that they ought not be as heavy as they are. They cannot fulfill this prudential obligation instantaneously but they may be able to do it over time. It may now be objected, however, that if there are cases in which there is only one way to satisfy our wide-scope conditional requirements, then these requirements are lacking in some way.10 As we have just seen, there are many cases in which we cannot control our beliefs. In those cases, we can satisfy a wide-scope conditional requirement only by putting an end to our desire. In other cases our ultimate moral obligations will require us to choose one way of proceeding rather than another. Consider the wide-scope conditional requirement mentioned above: You ought to (if you believe that women are less intelligent than men, you do not desire to act on this belief when serving on hiring committees). Suppose you do believe that women are less intelligent than men and that you desire to act on this belief when serving on hiring committees. In this case, you can, in principle, satisfy your wide-scope moral obligation in two different ways: You can give up your belief that women are less intelligent than men or you can refrain from desiring to act on this belief when serving on hiring committees. But suppose someone threatens to torture your friend mercilessly unless you stop desiring to act on your belief. In this case it would seem that you should do whatever you can to rid your mind of the desire to act on your belief. Wide-scope conditional requirements do not specify these types of asymmetries; they merely specify that you cannot hold two specific attitudes without violating a moral requirement. Some people find that problematic.11 However, as John Brunero (2012) has pointed out in a slightly different context, this observation does nothing to undermine the truth of the wide-scope requirements. The wide-scope requirements tell you that there are two ways of proceeding as far as your moral requirements go. They do not specify what you ought to do all things considered.   8  Here is Kelly: “It’s plausible to suppose that the reason why the expected consequences of my being a certain height make no difference to whether it is rational for me to be that height derives from my utter lack of control over my height. (Perhaps if I could control my height, then it would be more rational for me to be some heights rather than others)” (2002: 168).   9  I say “almost entirely” because we could change our height by getting surgery. Likewise, I could give you the book I borrowed from you on the day I promised even if you are on an island that is only accessible via helicopter. In both cases completing the action is extremely difficult and would have serious consequences. So, where “can” is the “can” of practical ability, we cannot do it.   10  This objection is a variation on objections raised against Broome by Schoeder (2004) and Bedke (2009).   11  For example, see Schoeder (2004) and Bedke (2009).

Wide-Scope Requirements and the Ethics of Belief  139 So it may be that one of the two ways of proceeding is necessary in light of your total moral requirements or abilities but it is nonetheless still true that you are required to take one of the two routes. To recap: We are not morally required to hold or not hold any beliefs (regardless of the evidence we have or lack). However, there are wide-scope moral requirements governing bad-belief/desire pairs. We cannot always decide what to believe but even when we cannot make these kinds of decisions, we may still be able to satisfy the wide-scope requirements by deciding what to desire.

7.3  Epistemic Requirements and Beliefs It may be thought that even if we are not morally obliged to believe only on sufficient evidence, at least we are epistemically required to believe only on sufficient evidence. The fundamental epistemic norm is often taken to be some variation of the following norm: The Truth Norm You ought to maximize your true beliefs and minimize your false beliefs.12 In the majority of cases we cannot determine whether a belief is true or false without evidence. So we can satisfy the Truth Norm only by believing on sufficient evidence. It follows that if the Truth Norm, or some close variant, is the fundamental epistemic norm, then we are epistemically required to believe only on sufficient evidence. In previous work I argued that the Truth Norm, construed as a fundamental epistemic norm, fails because it does not always aim at what is intellectually valuable. Consider the following case (Brogaard forthcoming):13 Brain Damage A has a brain condition that causes him to intend to keep track of truths about leaves. He believes that he can achieve this only if he intends to count the leaves on the trees in his garden every day. If A does what he believes is necessary for him to intend to keep track of truths about leaves, and he is a good counter, his intention is likely to maximize true belief and minimize false ones. If he didn’t intend to count leaves, he would go about his everyday business forming a lot more false beliefs than he does if he is just counting leaves all day. But intuitively, he should not intend to stay in his garden counting leaves all day. Intellectual flourishing requires having a wide range of different kinds of true beliefs. So, despite the fact that A’s intention maximizes true beliefs and minimizes false ones,

  12  See David (2001), Wedgwood (2002), Burge (2003), Peacocke (2004:  15), and Shah and Velleman (2005).   13  Jason Rogers and Jonathan Matheson (2011) offer another compelling case (though in a different context) in which an intelligent believer adds double negation to mathematical truths over and over.

140  Berit Brogaard A ought to not have that intention. As the Truth Norm can be overruled, it is not the fundamental epistemic norm. A more plausible candidate to be the fundamental epistemic norm is what I have called “intellectual flourishing” (Brogaard forthcoming). Intellectual flourishing is a continuous process of living a good intellectual life. It is the epistemic equivalent of Aristotle’s eudaimonia (well-being, flourishing, happiness). For Aristotle, the ultimate aim of our lives is virtuous activity. But your ability to engage in virtuous activity will be diminished if you lack in certain ways, for instance, if you are not loved or you have no close friends. So eudaimonia requires not only having a virtuous character, and acting on it, it also requires possessing certain goods and being in certain relationships with other people. Although “eudaimonia” is sometimes translated as happiness, eudaimonia is not a purely subjective state. Nor is it a disposition to feel a certain way. You can feel a certain way or be disposed to feel a certain way without ever achieving eudaimonia (Kraut 1979; Haybron 2008). For example, you may feel extremely happy alone and avoid forming friendships for this reason but without close friends you cannot act virtuously in the fullest sense (Kraut 2012). So you are not flourishing. If we extend this idea of eudaimonia to the intellectual realm, then intellectual flourishing might involve such things as avoiding intellectual bigotry, seeking to expand on one’s knowledge, making wise intellectual choices, being respected and admired intellectually, and having good intellectual cohorts. Just as we cannot flourish, in Aristotle’s sense, in solitude, so we cannot flourish intellectually outside of an intellectual community. Intellectual flourishing differs in this respect from knowledge acquisition. While a brain in a vat that is not properly connected to an intellectual community could, in principle, acquire knowledge as well as you and me, it cannot flourish intellectually. While virtuous character traits and well-functioning cognitive faculties and abilities can lead to a good intellectual life, there are many cases in which true belief flows from virtuous character traits or well-functioning cognitive faculties and abilities but in which the agent is not on the right track intellectually speaking.14 Each individual is unique and thus possesses a particular set of personality traits and mental abilities and is situated in her own social and historical context. Needs, mental acumen and circumstances affect an individual’s eudaimonia. An activity that can contribute to one individual’s eudaimonia may not be relevant to another’s. For example, it’s intellectually admirable if a person with a spinal cord injury that leaves her paralyzed from the waist down decides to undergo extensive locomotor training because this kind of training involves a kind of motivation, tenacity, and patience that outruns our expectations for people with spinal cord injury. But walking on a treadmill in the gym every day does not in normal circumstances contribute to the intellectual flourishing of an able man or woman, regardless of how intellectually virtuous he or she is and regardless of how well his or her cognitive faculties and abilities function.   14  For defenses of virtue epistemology in the more traditional sense, see, e.g., Montmarquet (1987), Axtell (1997), Greco (2000, 2003), Zagzebski (2003), Sosa (2007), and Greco and Turri (2011).

Wide-Scope Requirements and the Ethics of Belief  141 Being intellectually virtuous may also be insufficient for intellectual flourishing if intellectual achievements that flow from the virtues are not admirable by public measures. Consider the following case:15 Rich Uncle You are an ambitious philosopher with generally well-functioning cognitive faculties and abilities and many good personality traits. You invest a great deal in writing articles and books. You have what seems to be a perfectly successful career. Your articles and books regularly win prizes and public praise. In fact, however, unbeknownst to you, all your papers and books are published and assessed by people hired by your rich uncle who took pity on your because you are such a bad philosopher. Needless to say, in this case you do not flourish intellectually despite believing that you do, as your intellectual achievements fail to meet public measures of greatness. Being intellectually virtuous in the conventional sense may preclude flourishing intellectually. Suppose A hears of a new proof that God does not exist. A knows that if he sees the proof and the proof is correct, he will become terribly depressed and will spend the rest of his life in isolation from intellectual cohorts. To ensure that he flourishes intellectually, A must refrain from looking at the proof, even if this move does not involve the exercise of intellectual virtue. We can, of course, correctly say of an agent who flourishes intellectually that he or she is “intellectually virtuous,” he or she just isn’t virtuous in the classical sense. There is no one set of character traits that an agent who flourishes intellectually must have. What can be a positive trait in one situation or for one person may be a bad trait in a different situation or for a different person. For example, you should not be intellectually honest while carrying out an experiment that involves deceit (e.g., Milgram’s experiment). Truth-telling in this situation would ruin the experiment. Even intellectual justice can counteract eudaimonia. To be unjust in the intellectual domain is to do something that could potentially hinder the intellectual flourishing of others. Destroying other people’s intellectual property, preventing others from developing their mental abilities, rewarding unworthy rather than worthy intellectual achievements, obstructing intellectual amity and camaraderie, disrespecting the intellectual work of others on irrelevant grounds, such as gender or skin color, are all prima facie intellectually unjust activities. The very possibility of eudaimonia presupposes justice. However, what counts as unjust in one situation may count as just in another. In general, it is unjust to prevent people from posting their thoughts on their personal website. However, it may be just to prevent people from posting bigoted content on their personal website. At the end of his essay Clifford states that “it is wrong in all cases to believe on insufficient evidence; and where it is presumption to doubt and to investigate, there it is

15

  This example is adapted from Brogaard and Smith (2005).

142  Berit Brogaard worse than presumption to believe” (1877: 309). This is problematic even if “wrong” is read as “epistemically wrong.” If intellectual flourishing is the fundamental epistemic norm, which seems very plausible, then we are not epistemically required to believe on sufficient evidence in every single case. Consider the following case: Good Intellectual Cohort You have evidence for believing that one of your intellectual cohorts is cheating on his wife. You are a strong opponent of any form of adultery. You realize that your belief that your intellectual cohort is cheating on his wife is negatively affecting your collaborations on a book project. So you decide to use whatever techniques are available to suppress the belief. You succeed in suppressing the belief. In the process you form the belief that your intellectual cohort is a morally good person by your standards. The two of you finish the book, which turns out to become extremely influential in your area. In this case, you believe that your intellectual cohort is a morally good person by your standards on insufficient evidence. If not having this belief would seriously damage your book project, you are epistemically required to believe something on insufficient evidence. So not only are you not epistemically required to refrain from believing something on insufficient evidence, you are epistemically required to believe something on insufficient evidence. So, even if we change Clifford’s thesis to be about our epistemic obligations rather than our moral obligations, the thesis fails. There are many cases in which we are not epistemically required to refrain from believing something despite lacking evidence. As this is a denial of an implication of evidentialism, evidentialism is false. Some have argued that practical considerations can never rationalize belief (see, e.g., Kelly 2002). The idea is that the only thing that can rationalize belief is a reason upon which the belief is based. The basing relation is supposed to be one that makes a difference to whether or not you continue to hold the belief in the presence of counterevidence. Suppose you believe that it is raining based on the reason that everyone is entering the hallway dripping wet. If you are told that the botanical garden has installed a big water hose outside that makes everyone wet, you will no longer believe that it’s raining. If, on the other hand, you believe that your colleague is a good man on the basis of successful attempts to forget the counterevidence, and you are told convincingly that you have never made any attempts to forget any counterevidence, you are still going to believe your colleague is a good man. The problem with this sort of argument, if taken to be a refutation of my view, is that it presupposes that rational, or “epistemically good,” belief is belief based on evidence. But that is exactly the position I have offered reasons against. You can, in special circumstances, have a rational or “epistemically good” belief that is not based on evidence. The reason for this is simple: Belief based on evidence isn’t always intellectually valuable. Belief based on evidence is not intellectually valuable when holding the belief can have harmful intellectual consequences.

Wide-Scope Requirements and the Ethics of Belief  143 The mistake committed here, I believe, is that of assuming that practical considerations can only contribute to prudential and moral rationality. The problem is presumably that of thinking in terms of “practical considerations.” “Practical considerations,” as the phrase is normally used, is an umbrella term for moral and prudential considerations. But moral and prudential considerations aim at very different goals. The aim of thinking about the consequences of pulling the trigger of a gun while pointing it at someone is that of avoiding moral harm, whereas the aim of thinking about the consequences of procrastinating before an upcoming exam is that of avoiding prudential harm. But analogously, the aim of weighing the epistemic consequences of holding certain beliefs is that of avoiding epistemic harm. All of these considerations are “practical.” Yet their aims are very different. This suggests that epistemic goodness (of belief) can be related to practical considerations and hence can be unrelated to evidence. So (pace Kelly) practical considerations can make beliefs rational. It should be granted, of course, that while we do not have a moral or an epistemic requirement to believe on sufficient evidence in every single case, there are many cases in which we are epistemically required to withhold belief if we lack evidence. But those are the cases in which not believing on sufficient evidence hinders intellectual flourishing. In sum: Clifford was wrong in thinking that we are unconditionally morally required to believe on sufficient evidence. Belief by itself is not subject to moral assessment. However, while we do not have an unconditional moral requirement to believe only on sufficient evidence, we do have wide-scope conditional requirements to either stop holding “bad” beliefs or stop desiring to act on them. Turning to our epistemic requirements: The epistemic consequences of belief can affect the belief ’s epistemic standing. So we don’t have an epistemic requirement to believe only on sufficient evidence. However, we do have an epistemic requirement not to prevent intellectual flourishing.16

References Alston, W. (1988). “The Deontological Conception of Epistemic Justification,” Philosophical Perspectives 2: 257–99. Axtell, G. (1997). “Recent Work in Virtue Epistemology,” American Philosophical Quarterly 34: 410–30. Bedke, M.S. (2009). “The Iffiest Oughts:  A  Guise of Reasons Account of End-Given Conditionals,” Ethics 119: 672–98. Brogaard, B. (2014). “Intellectual Flourishing as the Fundamental Norm,” in John Turri and Clayton Littlejohn (Eds.), Epistemic Norms: New Essays on Action, Belief, and Assertion. Oxford: Oxford University Press.

  16  I  am grateful to John Hawthorne, Anna-Sara Malmgren, Kristian C.  Marlow, Jon Matheson, Axel Mueller, Baron Reed, the students in my 2012 epistemology seminar, and an audience at Northwestern University for comments on an earlier version of this chapter.

144  Berit Brogaard Brogaard, B. and Smith, B. (2005). “On Luck, Responsibility, and the Meaning of Life,” Philosophical Papers 34: 443–58. Broome, J. (2000). “Normative Requirements,” in J. Dancy (Ed.), Normativity. Oxford: Blackwell, 78–99. Broome, J. (2001). “Normative Practical Reasoning,” Proceedings of the Aristotelian Society, suppl. vol. 75: 175–93. Broome, J. (2002). “Practical Reasoning,” in J. L. Bermudez and A. Millar (Eds.), Reason and Nature. Oxford: Oxford University Press, 85–111. Brunero, J. (2012). “Instrumental Rationality, Symmetry and Scope,” Philosophical Studies 157: 125–40. Burge, T. (2003). “Perceptual Entitlement,” Philosophy and Phenomenological Research 67(3): 505–48. Chignell, A. (2010). “The Ethics of Belief,” The Stanford Encyclopedia of Philosophy, Edward N. Zalta (Ed.), URL = . Clifford, W. K. (1999), “The Ethics of Belief,” in T. Madigan (Ed.), The Ethics of Belief and Other Essays. Amherst, MA: Prometheus, 70–96. (Original edition, 1877.) Dancy, J. (2000). Practical Reality. Oxford: Oxford University Press. Darwall, S. (1983). Impartial Reason. Ithaca, NY: Cornell University Press. David, M. (2001). “Truth as the Epistemic Goal,” in M. Steup (Ed.), Knowledge, Truth, and Duty. New York: Oxford University Press, 151–70. Feldman, R. (2000). “The Ethics of Belief,” Philosophy and Phenomenological Research 60: 667–95. Greco, J. (2000). Putting Skeptics in Their Place. New York: Cambridge University Press. Greco, J. (2003). “Knowledge as Credit for True Belief,” in M. DePaul and L. Zagzebski (Eds.), Intellectual Virtue:  Perspectives from Ethics and Epistemology. Oxford:  Oxford University Press, 111–34. Greco, J. and Turri, J. (forthcoming). “Virtue Epistemology,” The Stanford Encyclopedia of Philosophy, Edward N.  Zalta (Ed.), URL  =  . Greenspan, P. S. (1975) “Conditional Oughts and Hypothetical Imperatives,” Journal of Philosophy 72: 259–76. Haybron, D. (2008). The Pursuit of Unhappiness: The Elusive Psychology of Well-Being. New York: Oxford University Press. Hill, T. (1973). “The Hypothetical Imperative,” Philosophical Review 82: 429–50. Kelly, T. (2002). “The Rationality of Belief And Some Other Propositional Attitudes,” Philosophical Studies 110: 163–96. Korsgaard, C. (1986). “Scepticism about Practical Reason,” Journal of Philosophy 83: 5–25 Kraut, R. (1979). “Two Conceptions of Happiness,” Philosophical Review 88: 167–97. Kraut, R. (2012). “Aristotle’s Ethics,” The Stanford Encyclopedia of Philosophy, Edward N. Zalta (Ed.), URL = . Rogers, J. and Matheson, J. (2011). “Bergmann’s Dilemma:  Exit Strategies for Internalists,” Philosophical Studies 152: 55–80. Montmarquet, J. (1986). “The Voluntariness of Belief,” Analysis 46: 49–53. Montmarquet, J. (1987). “Epistemic Virtue,” Mind 96(384): 482–97.

Wide-Scope Requirements and the Ethics of Belief  145 Montmarquet. J. (1993). Epistemic Virtue and Doxastic Responsibility. Lanham, MD: Rowman & Littlefield. Nottelmann, N. (2006). “The Analogy Argument for Doxastic Voluntarism,” Philosophical Studies 131: 559–82. Peacocke, C. (2004). The Realm of Reason. Oxford: Oxford University Press. Plantinga, A. (1993). Warrant: The Current Debate. New York: Oxford University Press. Ryan, S. (2003). “Doxastic Compatibilism and the Ethics of Belief,” Philosophical Studies 114: 47–79. Schoeder, M. (2004). “The Scope of Instrumental Reason,” Philosophical Perspectives 18: 337–64. Shafer-Landau, R. (1997). “Moral Rules,” Ethics 107: 584–611. Shah, N. and Velleman, J. D. (2005). “Doxastic Deliberation,” Philosophical Review 114: 497–534. Smith, M. (2010). “Humeanism about Motivation,” in T. O’Connor and C. Sandis (Eds.), A Companion to the Philosophy of Action. Oxford: Blackwell, 153–8. Sosa, E. (2007). A Virtue Epistemology: Apt Belief and Reflective Knowledge, Vol. 1. Oxford: Oxford University Press. Steup, M. (2000). “Doxastic Voluntarism and Epistemic Deontology,” Acta Analytica 15: 25–56. Wallace, R. (2001). “Normativity, Commitment and Instrumental Reason,” Philosophers’ Imprint 1: 1–26. Wedgwood, R. (2002). “The Aim of Belief,” Philosophical Perspectives 16: 267–97. Zagzebski, L. (2003). “The Search for the Source of Epistemic Good,” Metaphilosophy 34: 12–28. Zagzebski, L. (1996). Virtues of the Mind: An Inquiry into the Nature of Virtue and the Ethical Foundations of Knowledge. Cambridge: Cambridge University Press.

8 The ‘Ethics of Belief ’ is Ethics (Period) Reassigning Responsibilism Trent Dougherty

The central thesis that I sketch in this essay is that an area of philosophical discussion that has traditionally gone under the name “the ethics of belief ” and plays a certain kind of blame game is, when properly considered, applied ethics and not epistemology. In particular, it bears no logical relation to the theory of epistemic justification or knowledge or understanding,1 though, of course, there might be ample contingent connection. This taxonomic point has, I argue, significance for how we think about this traditional area of discourse and the problems it investigates. In section 8.1, I give an example of the phenomenon under investigation. The target is to be understood in terms of that example rather than via abstract definition. Anything with sufficient resemblance to that example will fall under what I say. In section 8.2, I lay out my strategy for reducing this traditional “ethics of belief ” and illustrate it with two examples. In section 8.3, I set out in some detail the reductive thesis about epistemic justification. In section 8.4, I briefly address a considerable number of objections that both defend and clarify the central thesis. The plenitude of objections cannot be addressed here in any detail, for my central purpose to recommend a new way to proceed in an old debate. However, each of the objections I address has been raised in various presentations of various forms of this chapter or in discussion with numerous interlocutors.2   1  What about Wisdom? I doubt that wisdom is a proper object of epistemology, but that depends on what wisdom is. As Ryan (2013) makes painfully clear, conceptions of wisdom vary radically, making knowledge look simple by comparison.   2  What I will not do is engage further with theorists on the other side whose work has been discussed such that I have nothing further to add. This includes the fine work of Jason Baehr (e.g., 2011a and 2011b) and of Guy Axtell (2011a, 2011b, and 2012). For replies to Baehr, see Dougherty (2011a and MS). For my engagement with Axtell, see Dougherty (2011c and 2012). I simply don’t have anything to add on their work. My goal here is to succinctly present a description of another way of thinking about the issue with fewer commitments.

The ‘Ethics of Belief’ is Ethics (Period)  147

8.1  An Ethics of Belief Exemplar The human impulse to blame runs deep. We blame each other for all kinds of things. Among them are having beliefs we think one shouldn’t have. Here is a prime example3 of a prophet of epistemic righteousness. In the fourteenth century, religious belief was quite plainly neither irrational nor unwarranted— and thus not culpably false belief—but after the scientific revolution and the Enlightenment, it is less clear . . . I am going to assume—uncontroversially among most philosophers but controversially among reformed epistemologists—that `reformed epistemology’ is nothing more than an effort to insulate religious faith from ordinary standards of reasons and evidence in common sense and the sciences, and thus religious belief is a culpable form of unwarranted belief given those ordinary epistemic standards. (Leiter 2010: 954–5)

Our prophet does not say clearly what he means by “culpably false belief,” “culpable form of unwarranted belief,” or “blameworthy epistemic irresponsibility” (954). However, he repeats the claim that “some religious beliefs are insulated from evidence, as evidence is understood in common sense and the sciences” (955). I think the claim here is not that the religious believers’ beliefs do not fit the evidence they have. Rather, the problem is that due to failures in inquiry—e.g., turning a blind eye to relevant sources of evidence—the evidence they have is not what it should be. That is, all the ethical language here is not issuing from a commitment to mainstream epistemological evidentialism—such as I will describe and affirm in section 8.3 below—which is a thesis about fit between one’s belief states and one’s actual evidence at that time. Instead, “the ethics of belief ” here pertains to what one’s evidence ought to be or perhaps would have been, if one were a responsible inquirer. This kind of criticism, it seems to me, is the standard referent of the phrase “the ethics of belief,” especially in its locus classicus, Clifford’s “The Ethics of Belief.”4 Evidence that the ethics of belief is thought to go well beyond the scope of mainline epistemological evidentialism is the persistent suggestion that evidentialism is deficient for reasons pertaining to inquiry.5 This inquiry-based usage of “ought” pertaining to belief is on display most frequently in the negative in statements that one should not believe something or implicit in such laments as “How can you believe that?!” The charge is an accusation of vice. Indeed, there are two main forms of “virtue   3  I ask you please to note that this example serves as an initial paradigm for the kind of ethics of belief I think has been predominantly referred to as “the ethics of belief.” I do not think this notion or much else of interest in philosophy can be given simple, explanatory, counter-example free abstract definitions. What kind of cases am I centered on? This kind.   4  Objection:  I  think you have read Clifford wrong! Reply:  I  have little interest in Clifford exegesis— though I will engage in some below. But it seems clear that this is at least one standard reading of Clifford, and even if it were a terrible reading of Clifford—which is facilitated by his inconsistencies—it is still an interesting line of thought and clearly evinced in much doxastically judgmental speech.   5  This suggestion can seen from Kornblith (1983), to which Conee and Feldman reply in their classic (1985) paper “Evidentialism,” as well as to Baehr (2011), to which Conee and Feldman reply in Dougherty (2011a).

148  Trent Dougherty ethics”: virtue reliabilism and virtue responsibilism. Reliabilists focus on sub-personal systems and their probabilistic properties. Responsibilists focus on the agent, sometimes particularly on her character.6 The ethics of belief as on display above is a firmly responsibilist notion. To turn to the most well known account which practically defines the phrase, the very first section of Clifford’s “Ethics of Belief ” is entitled “The Duty of Inquiry.” Like the First Amendment to the Bill of Rights, I don’t think its status as the arche point is an accident. On the contrary, I think it signals that the essence of the ethics of belief under scrutiny is essentially about duties regarding evidence gathering. So when I say, “The ethics of belief is ethics (period)”—which is clearly meant to imply “and not epistemology”—I mean that the notion of epistemic responsibility is not part of epistemology, but, rather, simply a sub-discipline of standard ethics, or, more precisely, applied ethics. In fact, I take it to be applied ethics exactly on par with, say, medical ethics or business ethics. It is standard kind of ethical theory applied to a particular domain—e.g., medicine, business, belief, etc. What is at stake here is not merely taxonomy—though I think that is worth getting right. Rather, this taxonomic thesis is a lens through which to see some important truths about the nature of the normativity involved in the ethics of belief.

8.2  Eliminating “Epistemic” I have recently on several occasions defended the following identity thesis by way of the reduction of responsibility:7 IT:  Each instance of [so-called] epistemic irresponsibility is just an instance of purely non-epistemic irresponsibility/irrationality (either moral or instrumental). (Dougherty 2011b: 4)

IT is reductionistic about epistemic responsibility broadly construed because it identifies it with other forms of responsibility. It entails that there is no epistemic notion of responsibility other than the “responsibility” one has to believe in accordance with one’s actual evidence at a time. Why the scare quotes? I am not sure the term is apt when applied to belief, though I do think there is some kind of truly epistemic “ought.” What I think about that I will try to say below. Even an inept lawyer has an obligation to defend his client adequately, for that is her responsibility.8 I doubt this is mere metaphor. I hypothesize that this talk, and its appropriateness, stems from the notion that an agent who is aware of his or her intellectual telos (to be discussed below) is responsible for trying to bring it about that they fulfill that telos. But it may well be metaphorical

  6  See Axtell (1997) for details on responsibilism; see Baehr (2011) for the most well-developed character account in print.   7  Dougherty (2011b, 2011c, and 2012).   8  For similar thoughts, see Feldman (2004).

The ‘Ethics of Belief’ is Ethics (Period)  149 or even incoherent, in which case we should dispense with talk of purely epistemic obligations or purely epistemic responsibility altogether. It is easy to see how we could be suckered in to this way of speech. Because the entities in question are capable of conceptualizing their telos, we slip into thinking about their moral responsibility to achieve their telos. I will not pursue this line of thought further here, but I do think it is a promising explanatory gambit for why we use the language we do in sometimes seemingly odd ways. I want to make the dilemma I posed in “Reducing Responsibility” more explicit here. I think it is most usefully put in the form of a recipe or flowchart (Figure 8.19) for the reduction of an alleged item of epistemic irresponsibility to something else.10 In this way, so I say, any alleged case of epistemic irresponsibility can be seen to be either a standard case of lack of evidential fit, some non-epistemic shortcoming, or no problem at all. Lack of evidential fit is, as I will argue in the next section, a genuinely epistemic criticism. But, as we have seen, it is not the criticism behind paradigm examples of the use of ethical language to condemn belief. It is a form of criticism to which “ethics” cannot be literally applied, for it is the kind of criticisms one makes of a horse’s strength and a flower’s beauty. It is teleological in nature. I will illustrate the method of reduction with an example fitting to our opening example.11 Let us suppose there is an individual who is a “big wig” with lots of influence in a certain religious circle. We’ll call this ring leader “Big Al.” We’ll focus on one of Big Al’s followers, Little Kenny.12 Big Al’s intellect is dizzyingly brilliant and Little Kenny is star-struck. Big Al is a young earth creationist who believes the world was created in late October 4004 BC. Kenny knows that one of Big Al’s foes is Dapper Dan, who is himself revered among the secular elite of the academic industrial complex. Kenny checks his book out from the library, planning on reading it. But before he has a chance, Big Al sees it on his desk and exclaims, “Little Kenny! Burn this book immediately! It contains nothing but lies and sophistry! It is nothing but a bunch of Enlightenment hooey!” Such is Little Kenny’s veneration for Big Al that he immediately obeys, even though he worries that he’s not getting the whole story. We may stipulate, as part of the story, that as a result of turning a blind eye to a source of evidence, Little Kenny’s beliefs do in fact fit the evidence he has. That is, the proposition that the Earth was created in late October, 4004 BC is sufficiently supported by Little Kenny’s total evidence that belief is the most fitting

  9  Special thanks to Timmy Hsiao for creating the flowchart.   10  When I run cases through this flowchart, I use a pretty Humean account of practical reasoning, but one can supply one’s own favored view to get the same results.   11  Here, again, the example serves as a paradigm around which the target is to be defined. The intension is to be read of the extension here.   12  An anonymous reviewer wondered about the point of these references. In short, I am mocking people who treat Alvin Plantinga as a fundamentalist—for a classic treatment of this term see Plantinga (2000: 244– 5). Also, the Kenny alluded to is widely known as a bold and creative thinker, so the role here is ironic. If you know Kenny, tell him I said, “You’re welcome.”

1. Take the case of a particular instance a belief that seems normatively defective—say, S taking doxastic attitude A toward p in circumstance C at t—and ask the following questions.

2. Ask: Does A fit S’s evidence at t?

If not

2.1 Then there is a simple explanation of the seeming normative defect—lack of evidential fit. Since the alternative explanations are not mutually exclusive, proceed with further questions.

If so

3. Ask: Is p properly based on S’s evidence with regard to p?

If not

3.1 Then here is an explanation of what’s amiss.

If not

4.1 Then there is no basis for the charge of irresponsibility.

If so

4. Ask: In C, is there much at stake in A’s being inaccurate?

If so

5. Ask: Are the stakes those of S?

If so

5.1 Then not inquiring further into whether p was a failing of practical rationality.

If not

6. Ask: Are the stakes those of some individual to whom S owes a relevent duty of care?

If not 6.2. Then there is nothing at stake and so no basis for a charge of irresponsibility.

Figure 8.1  How to reduce responsibility

If so

6.1 Then not inquiring further into whether p is a failure of a moral duty.

The ‘Ethics of Belief’ is Ethics (Period)  151 attitude to take to that proposition. Prima facie, this seems to be a classic case of epistemic irresponsibility. Now let’s run through the flowchart. We’ll consider variants along the way. Does Little Kenny’s belief in young earth creationism fit his evidence at t? Yes, we have stipulated that it does. We may further stipulate that his belief in young earth creationism is held on the basis of the evidence that he does have which does support that thesis, due to his narrow reading. Is there much at stake in Little Kenny’s belief about the age of the earth being accurate? Here there are a lot of possibilities. Little Kenny’s standing in his community is at stake. To read the book and displease Big Al only to find out that Big Al was right all along would be a huge loss for no gain. And even if it turned out he was convinced by Dapper Dan’s arguments, what does he really gain? He gains a true belief, to be sure, but how important is it to have this true belief? It’s hard to see that much of consequence rides on it. Perhaps it would lead to a general disregard for science and contribute to some degree to his ignoring the evidence of global climate destabilization. But this, too, seems of little consequence practically speaking. Unless we include in the story that Little Kenny is himself a leader in the community who will have a significant influence on a number of people who would have otherwise believed in global climate destabilization, then the concern seem little more than a Slippery Slope fallacy. So the only thing that seems to be at stake is the bare fact that Little Kenny would have one more true belief and one less false belief. I am willing to grant that every truth has some value.13 But how much is one morally required to desire truth for its own sake? Does the possible gain of the bare value of one truth outweigh the risk of community scorn? It is not at all clear that it does. However, if the problem with Little Kenny is that he doesn’t have a strong enough desire for the truth as an end in itself, this is a moral criticism. It is saying he is a bad person, a person of bad values. Clifford, of course, would be very displeased by Little Kenny. Whoso would deserve well of his fellows in this matter will guard the purity of his beliefs with a very fanaticism of jealous care, lest at any time it should rest on an unworthy object, and catch a stain which can never be wiped away. It is not only the leader of men, statesmen, philosopher, or poet, that owes this bounden duty to mankind. Every rustic who delivers in the village alehouse his slow, infrequent sentences, may help to kill or keep alive the fatal superstitions which clog his race. Every hard-worked wife of an artisan may transmit to her children beliefs which shall knit society together, or rend it in pieces. No simplicity of mind, no obscurity of station, can escape the universal duty of questioning all that we believe. (Clifford 1999: 74–5)

Looking past the ornate Victorian prose which would make Ruskin blush, we seem to see a grave concern for any and all error which Queen Victoria herself would be proud of, with her and her time’s great concern for the least offence against decency. Yet looking a little harder we see that the superstitions are “fatal” and that society  

  See Kvanvig (2008).

13

152  Trent Dougherty will survive or spoil by these falsehoods. So let us suppose that, like Clifford’s shipowner, there is something at stake. In fact, we can simply take his shipowner as the paradigm. A shipowner was about to send to sea an emigrant ship. He knew that she was old, and not over-well built at the first; that she had seen many seas and climes, and often had needed repairs. Doubts had been suggested to him that possibly she was not seaworthy. These doubts preyed upon his mind and made him unhappy; he though perhaps he ought to have her thoroughly overhauled and refitted, even though this should put him to great expense. Before the ship sailed, however, he succeeded in overcoming these melancholy reflections. He said to himself that she had gone safely through so many voyages and weathered so many storms that it was idle to suppose she would not come safely home from this trip also. He would put his trust in Providence, which could hardly fail to protect these unhappy families that were leaving their fatherland to seek for better times elsewhere. He would dismiss from his mind all ungenerous suspicions about the honesty of builders and contractors. In such ways he acquired a sincere and comfortable conviction that his vessel was thoroughly safe and seaworthy; he watched her departure with a light heart, and benevolent wishes for the success of the exiles in their strange new home that was to be; and he got his insurance-money when she went down in mid-ocean and told no tales. (Clifford 1999: 70)

Clifford’s shipowner, let’s call him “Rico,” has a lot at stake. His is an “emigrant-ship” full of “unhappy families that were leaving their fatherland to seek for better times elsewhere.” Rico has some evidence that the ship needs repairs: “He knew that she was old, and not over-well built at the first; that she had seen many seas and climes, and often had needed repairs.” And his argument to the contrary is not very good: “He said to himself that she had gone safely through so many voyages and weathered so many storms that it was idle to suppose she would not come safely home from this trip also.” Yet, as with the case of Little Kenny, the problem does not seem to be that his beliefs at the time he decides to set sail do not fit the evidence he has at that time. For “[h]‌e would dismiss from his mind” doubts and suspicions and “[i]n such ways he acquired a sincere and comfortable conviction that his vessel was thoroughly safe and seaworthy.” My interpretation is not forced by the text, but it fits it well, and we may stipulate it to hold. The real problem here is, as Leiter complains above, the turning of a blind eye to the information available. In truth, the evidence that the boat needs repairs is far from conclusive. We are told that he had doubts but the only basis we are given for them consists in three things: (1) that the ship was old, (2) that the ship was not exactly the best kind when it was new, and (3) that the ship had needed repairs in the past. All this is true of my truck, but I have no particular evidence that it is currently in need of repairs. So it does not even seem to me that Rico is justified in believing the ship needs repairs. But what is interesting here, since we could just fill in the details as we wished without changing anything, is that given the consequences, it is immoral to risk the perilous journey without a thorough inspection of the ship, just as it would be imprudent of me to set out from here in Texas on a trip to Alaska via the trans-Canadian highway without having my truck thoroughly inspected, even if

The ‘Ethics of Belief’ is Ethics (Period)  153 there was not enough evidence to believe that repairs would be needed. This is simple decision theory. There is in Clifford’s condemnation of Rico an ambiguity. What shall we say of him [if the ship should sink]? Surely this, that he was verily guilty of the death of those men. It is admitted that he did sincerely believe in the soundness of his ship; but the sincerity of his conviction can in no wise help him, because he had no right to believe on such evidence as was before him. (Clifford 1999: 70, emphasis in the original)

Many readers of Clifford I have talked to have read this as saying that Rico’s belief in the soundness of the ship was, at the time it was formed, not justified by the evidence Rico had at that time. But I think this is the wrong interpretation. For it seems clear that by “before him” Clifford means available. But the whole problem is that Rico didn’t come to acquire the relevant evidence, due to his turning a cold shoulder to it, making himself blind to what was plain to sight. The passage just quoted continues thusly, He had acquired his belief not by honestly earning it in patient investigation, but by stifling his doubts. And although in the end he may have felt so sure about it that he could not think otherwise, yet inasmuch as he had knowingly and willingly worked himself into that frame of mind, he must be held responsible for it. (Clifford 1999: 70–1, emphasis added)

While Rico ended up with a belief that was epistemically justified, it remains true that at the time at which he made the decisions not to follow up on the doubts, the expected utility of not doing so was clearly negative. And since he owed a duty to the passengers to protect their safety, his lack of investigation is straightforwardly wrong, morally wrong. If the contents of the ship were merely his cargo, then it would be merely imprudent. In either case, however, there is nothing distinctively epistemic going on here. The wrongness of not investigating is exactly the wrongness of, say walking across an icy pond without testing it or investing someone’s retirement money which you have been entrusted with in risky securities. It’s just plain irresponsible. Adding “epistemic” in front does nothing but mislead.

8.3  The Nature of the Normativity I have just argued that the ethics of belief as standardly conceived is not a distinctively epistemic notion. I do, however, think there is a useful notion of distinctively epistemic normativity, though it doesn’t lead to the kind of robust “ethics of belief ” that responsibilists want. My position is that the content of the “epistemic ought” is exhausted by considerations pertaining to evidential fit. In short, I endorse the following thesis: (EO) One epistemically-ought to believe p if and only if p fits one’s evidence.14

  14  One might wish to qualify this and allow for justified mistakes about what one’s evidence shows or is. If so, then one may amend it to say “. . . what one is justified in believing is one’s evidence.”

154  Trent Dougherty Put in more fine-grained terms, one ought—as Locke and Hume had it—to “assign” precisely that degree of certainty—or confidence, or belief, or what have you—in a proposition as is supported by, or fits with, or is justified by the evidence.15 We can accommodate both the ternary and continuous notions of belief by the phrase “One ought to believe in accordance with one’s evidence.” Here is the thesis I endorse as specifically aimed at the graded notion: (EOG) One ought to have n degree of certainty in p iff p is probable to degree n on one’s evidence.16 My own view about the nature of the epistemic ought is that it is teleological, but not in the way criticized by Grimm (2009). In short, the (possibly quite complex) sub-system17 of the mind that forms doxastic attitudes has as its proximate function to assign doxastic attitudes that fit the evidence available to that sub-system. The person qua person may have higher functions such as to believe or know the truth or even to “know, love, and serve God in this world, and enjoy him forever in the next” (Baltimore Catechism). But regardless of the broader functions of broader parts of the organism of which the doxastic sub-system is a part, I do not think the doxastic part aims directly at truth. And since the norm is a teleological one, there is a clear usage of “ought” connected with it which does not imply can. A rose ought to bloom, but sometimes it is in adverse conditions where it just can’t do so. One ought to believe in accordance with one’s evidence even when one is in unfortunate conditions that preclude it. Though I have something in mind more robust than I think Feldman (2000, 2004) does with his “role oughts”— “ethics of belief ”—this is one way of putting flesh on those bones.18 I believe that among most, perhaps all, propositions there exist logical or quasi-logical epistemic support relations that constitute logical probability or objective evidential probability or more broadly degree of epistemic support.19 Put in these terms, one’s subjective probability ought to match this objective evidential probability. I take this to be the sole uniquely epistemic ought and to be teleological in nature. This “ought” is synchronic and does not imply can and so bears no logical relation to   15  The contribution of the scare quotes around “assign” is to convey that I do not at all think this is a direct act one can make. At best, one can perform certain actions with the purpose of bringing it about that the degree of certainty they find themselves with is the degree of certainty they judge they ought to have. One might be doxastically incontinent—see Heil (1984) and Audi (2008)—or one might be doxastically retentive, like the skeptic, even when one judges that one should not be so. I do not here intend to rehearse the literature on doxastic voluntarism. I endorse Feldman’s (2000 and 2008) defense of the thesis that “ought” does not always imply “can”—see also Kornblith (2001) and Ryan (2003).   16  I set aside problems of vagueness here.   17  I assume very little about the “modularity” of the mind here. Since it is obvious that we go from sensory inputs somehow to doxastic outputs, we can describe a simple function regardless of how complex the realizing structure is. What I say here can be translated into any sensible theory of mind.   18  Feldman has much to say about the relationship between ethics, belief, and the ethics of belief. See, in addition to the other items referenced in this paper, Feldman (1988a, 1988b, 1998, 2000, 2001, 2002, 2003, 2004, and 2008).   19  Caveat: That the relation would be sensitive to what the agent in question thinks the evidential probability is would not stop it being objective. A theory of right behavior which was exclusively a function of the agent’s subjective desires would still be an account of the property “being right.”

The ‘Ethics of Belief’ is Ethics (Period)  155 inquiry. The role of the doxastic system—in the fine-grained mode—is to ascertain the evidential probability of a targeted proposition and to produce a degree of certainty equal to it. In this regard, a mind with justified false belief is doing better than one with an unjustified false belief. We shall consider below the relative value of having a justified false belief and an unjustified true belief. Here’s an example I like which illustrates the teleological picture I’ve been advocating. The module that assigns credences takes as input the testimony of the senses and other basic faculties—e.g., memory, intuition, introspection. Its job is to do the best with what it has been given, though it is part of a larger thing that is aimed at accuracy. The doxastic module has a role like an artist. The witness comes in and describes what the assailant looked like to them. A great police sketch artist will represent the mental image of the witness. Now, the witness might go wrong in a number of ways, and this can prevent the overall goal—an accurate representation of the criminal—from being achieved, but this is no failing of the sketch artist. The sketch artist performs his proximate telos perfectly if she perfectly captures in sketch what the witness describes. This is so regardless of how accurate the witness herself was in describing what she saw. Thus I endorse Strong Value Evidentialism (SVE). It is the conjunction of the following theses: EJ: Doxastic attitude D toward proposition p is epistemically justified for S at t if and only if having D toward p fits the evidence S has at t (Conee and Feldman 2004: 83). Super-EJ:  EJ exhausts, without remainder, the nature of epistemic obligation.20 At the risk of being redundant, I will affirm one last time that the kind of obligation that I admit is distinctively epistemic and will not support an “ethics” of belief in a literal sense, for these role-“oughts,” or teleological norms, do not imply “can.” And since the traditional reference of “the ethics of belief ” is reducible to non-epistemic notions, there is literally nothing distinctively epistemic for “the ethics of belief ” to refer to.

8.4 Debunking DeRose I said in footnote 2 that I had nothing more to add than I and others have said in response to a common kind of case used by responsibilists to argue against evidentialism. I do, however, have a bit to add to one of these cases: Keith DeRose’s Henry case. Conee and Feldman’s (2011) response is necessarily compressed by the context, and   20  Note that the claim here is not about what is epistemic and what is not epistemic but rather about which epistemic notions pertain to genuinely epistemic obligation and which don’t. The property “well-foundedness” is distinct from justification. See Conee and Feldman (2004). To the normative property of justification, well-foundedness adds that a belief has actually been formed and based on the evidence, which makes belief the fitting attitude for the person involved. An agent might fail to base their belief on the evidence it fits either through moral shortcomings like laziness, bad luck—e.g., an illness which causes them not to be able to concentrate, bad upbringing—e.g., a religious believer who has true beliefs they have adequate evidence for but who is taught that evidence has nothing to do with faith, or a deviant causal chain somewhere in the process of the formation of the belief.

156  Trent Dougherty parts of it are liable to be considered question begging in some regards to those who make such accusations. This is not true for their whole response, however, and I will mention part of their response below, since it overlaps with some of what I say here. Here is the case. Henry: Suppose that Henry firmly believes that p—it doesn’t matter much what p is—and has excellent evidence for p (evidence that’s strong enough to adequately support the firm and confident attitude Henry has adopted toward p). Suppose further that Henry doesn’t possess evidence against p, so the attitude toward p that fits all the evidence Henry possesses is the confident belief that p which Henry in fact holds. But suppose that Henry doesn’t believe p on the basis of the excellent evidence for it that he possesses. Indeed, Henry hasn’t even considered p in the light of this excellent evidence, and the fact that he possesses good evidence for p is no part of the explanation for why Henry believes that p. Rather, Henry believes that p on the basis of some other beliefs of his that he considers to be good evidence for p, but which in fact constitute absolutely lousy evidence for p. Henry has no business believing p at all on the basis of the evidence he is in fact using, much less believing it as firmly as he does. If you want some details—maybe your intuitions are helped out by filling in the story a bit—suppose that p is a proposition to the effect that someone that Henry hates very much has done some horrible thing. It’s Henry’s hatred and resulting desire to believe his opponent has done this horrible thing that causes him to think of his lousy evidence as strong evidence and to thereby believe that p. (DeRose 2011: 138)

He then considers the question: EQ: What ought Henry to believe when it comes to p? where the “ought” is understood to be an “epistemic ought.” He says that an epistemic “ought” should to be connected tightly with knowledge and this pushes him in the direction of saying Henry ought not believe p because Henry would not know p even if p turned out to be true. Conee and Feldman point out that there are all kinds of things that have gone wrong with Henry which explain the impulse to give a negative answer to the epistemic justification question. The most important is that the belief—because it is not based on the evidence that justifies it—does not meet the well-foundedness condition on knowledge. To understand this response it is necessary to keep in mind the distinction between propositional justification—the property of a proposition such that it is epistemically supported by a target individual’s body of evidence21—and doxastic justification—the property of a belief such that that belief is properly based on evidence which epistemically supports it.

  21  We might, by an ordinarily innocent synecdoche, say of S when p is justified on S’s evidence, “S is justified in believing p.” And of course when p has the property of being justified by S’s evidence, S has the property of having evidence that justifies p. Here, the underlying subject of discussion is the quasi-logical epistemic support relation that holds between some propositions—which might have the accidental property of constituting someone’s evidence—and others—which might have the accidental property of being believed by someone.

The ‘Ethics of Belief’ is Ethics (Period)  157 Propositional justification holds whether or not an individual actually forms a belief that p or even does (or, on some views of evidence, even can) consider whether p. S only comes into the equation because it is S’s evidence we are considering p relative to. It is almost universally agreed22 that for an actually formed belief to constitute an instance of knowledge, that belief must be properly based23 on the evidence that supports it. Thus, if p is justified for S—if S’s evidence bears epistemic support relations to p— it is emphatically not necessarily the case that were S to believe (true) p, then S would know p. This brief review allows us to show what is wrong with DeRose’s argument against understanding the epistemic ought in terms of propositional justification. DeRose notes that Henry would not know p even if p turned out true. And this, he thinks, dooms the evidentialist’s ought (“evi-ought”) from be the epistemic ought (“episte-ought”), because Henry evi-ought to believe p. DeRose’s argument seems to be something like this. 1. Datum: Henry doesn’t know p. 2. The best explanation is: He isn’t believing as he ought, in a way that is tightly connected to knowledge and, thus, rightly called an “epistemic ought.” 3. But he is believing as he evi-ought. 4. Therefore, the evi-ought is not part of the best explanation of why he doesn’t know. 5. Therefore, the evi-ought is not the episte-ought. With our review above, it is easy to see that Premise 2 is false, or at least unmotivated. An explanation of why knowledge doesn’t obtain is to be had in the failure to hold of some necessary condition(s). There is one necessary condition for knowledge that glaringly doesn’t hold in the picture DeRose paints: proper basing. In DeRose’s story, it is clearly stated that Henry doesn’t base his belief on the good evidence he has. Of course, it is possible that multiple things are going wrong, and, indeed, they are: Henry is a lazy jerk. But not being a lazy jerk isn’t a condition on knowledge: lots of lazy jerks know lots of things. In fact, no moral property is a logically necessary condition on knowing anything. The bottom line is that we have a perfectly good explanation from within the evidentialist theory of knowledge for why Henry doesn’t know, and so there is no reason here not to identify epistemic justification with propositional justification. And the simplicity, thought of as parsimony, and explanatory power of the hypothesis are good reasons to accept it. I will address one final issue from DeRose’s Henry case. DeRose considers three different sets of evidence to which we might relativize epistemic justification. 1. The evidence he actually brought to bear on p. 2. All and only the evidence he actually had. 3. His actual evidence plus what evidence he “easily could have.”   Notable exceptions are Lehrer (1971) and Kvanvig (2003).   See Korcz (2000) for an overview of the basing relation.

22 23

158  Trent Dougherty DeRose can’t see anything special about Option 2. I don’t know that Option 2 just glows or something, but it is by far the most natural and least problematic of the options. Consider first Option 1. There is something very strange about this option. It’s this: if we relativized the epistemic ought this way, then we can’t make judgments about what someone epistemically ought to believe until they have actually formed the belief. What someone should believe, when they are just considering propositions, is, it seems, simply indeterminate. So ahead of time, if Henry said, “What (episti-)ought I to believe?” we would have to say, “There is no answer to that question.” But then, when Henry actually formed the belief, we’d cry, “Foul!” That’s just plain weird. Now consider Option 3. It immediately suggests the question: How easily? Book by the nightstand? Gotta read it. Downstairs in the office? Gotta read it. At the work office? Hmmm. Next door? Maybe. Once one goes beyond actual evidence and suggests that one epistemically ought to believe on the basis of possible evidence, one will have very hard time indeed drawing a non-arbitrary line. Option 2 suggests a natural and non-arbitrary line. Option 2 clearly stands out as the best overall option, even given problems fixing what evidence is had, since Option 1 and Option 3 face similar issues. Thus, DeRose’s case of Henry seems to pose little if any challenge to the evidentialist picture.

8.5  Overcoming Objections Objection: Isn’t ‘Believe the truth’ an epistemic norm? For the “epistemic” in that question to do any work, the command would have to embody a distinctively epistemic normativity, rather than merely express a moral duty of an agent. The notion epistemic justification comes from the discussion in the Meno and Theaetetus about how knowledge differs from mere true belief. Obviously, since truth is common between knowledge and mere true belief, it is does not pertain to distinctively epistemic normativity. What accounts for the difference between knowledge and mere true belief is that in a case of knowledge the believer of the true has some kind of cognizance of or connection to what makes the proposition true—and, of course, the believer must properly24 base their belief on that evidence. So there may be some kind of human telos to believe the truth, but that would be a kind of Aristotelian moral virtue, a condition of human flourishing, not an epistemic norm, but the function of Reason is to weigh reasons. Theoretical rationality weighs theoretical reasons— evidence: signs of the truth of a proposition or its negation. Practical rationality weighs practical reasons—chances of success: signs that an act will further ends. When we engage in practical deliberation we end up with a judgment of expected utility. If we act on this, our action is rational and normatively correct regardless of whether it actually has the best outcome. When we engage in theoretical deliberation we end up with a   24  Part of proper basing, on my view, is not essentially depending on a falsehood, which is also my account of what goes wrong in Gettier cases.

The ‘Ethics of Belief’ is Ethics (Period)  159 judgment of probability—one that may be used in practical deliberation. If we believe in accordance with this probability, then our beliefs are normatively correct whether or not they turn out to be true. The truth is not a norm of belief or the anchor of epistemic justification or rationality. Just as success is a non-normative externality in the realm of practical rationality—a rational decision cannot be criticized for having a bad outcome—so truth is a non-normative externality in the realm of theoretical rationality—a rational belief cannot be criticized for coming out false.25 Objection: You just talked of a distinction between practical rationality and theoretical rationality, but isn’t practical rationality with respect to the goal of believing truths or getting knowledge inherently epistemic? Richard Foley has suggested this, but no.26 Practical rationality is about achieving one’s aims. Suppose someone cares deeply about the knowing the truth. Then they will have a practical reason to engage in considerable efforts to get knowledge. But that the end is epistemic does not mean that there is a new kind of rationality that is epistemic. To see this, consider one’s goal to be fit: this does not delineate a new sui generis kind of fitness rationality. Furthermore, sometimes practical rationality with respect to the truth goal and epistemic rationality can conflict. Here is an example based on Conee (2004). Suppose a genie tells you that if you manage to believe that the moon is made of green cheese, he will make it the case that in the future, any time you wonder whether p you will the next moment know whether p. Since all your evidence supports that the moon is not made out of green cheese, it would be epistemically irrational to believe it, if you could somehow get yourself to do so, even though in this case it is practically rational to do it from the standpoint of the truth goal. Since the two notions can come into conflict, they cannot be the same thing. Objection: Doesn’t your view exclude epistemic virtues from epistemology? The phrase “the ethics of belief ” has not traditionally been used to refer to a virtue ethics of belief. So reducing such a view lies beyond the scope of this chapter. I can, however, sketch how such a project would go. Some of what goes by that name “epistemic virtue” seem clearly to me to be moral virtues with epistemic payoffs when things go well. But I am not committed to the thesis that there is any natural category of such virtues, for they seem a scattered lot to me. For example, consider the so-called “epistemic virtue” of intellectual courage. In an environment where the majority is stubbornly refusing to countenance new research, intellectual courage might have veritistically beneficial results. But this is purely contingent: it might just as well lead one to misleading evidence. Objection: But if it did that, it wouldn’t be intellectual courage, it would be some vice like intellectual rashness.    

  For more, see the debate between David (2005) and Kvanvig (2005).   See Foley (2005) and a critique by Kelly (2003).

25

26

160  Trent Dougherty That makes epistemic virtue theory sound tautological and unfalsifiable. But it’s just a misguided picture. Consider standard courage. There is no reason to think that genuine courage won’t sometimes lead to failure or worse results than cowardice. Thus there is no logical relation between such virtues and their teloi. Thus, it is a contingent relation and subject to empirical research.27 I find the work of Baehr (2011) and Roberts and Wood (2007) fascinating, illuminating, and helpful, but it is—and at least Roberts seems to agree—a kind of psychological investigation. Roberts does use the term “regulative epistemology,”28 but I think that is an elliptical reference to ways of regulating one’s life in an epistemically beneficial way. Objection: You ended section 8.2 by saying that adding “epistemic” to “irresponsible” adds nothing. But it does tell us what kind of irresponsibility it is, that it involved the formation of a belief. Fine, so it does something, but that in no way delineates a proper sub-field of epistemology. Consider the example of imprudently walking across a pond. Is that “pondological irresponsibility”? Is that a new kind of normativity? Are there sui generis pond-walking duties? Surely not. There are diets—focused on so-called “brain food”—that focus on the enhancing of the reliability of our belief-forming processes. Does the fact that this culinary art and science focuses on belief formation make it part of epistemology? It is absurd to think that focus on belief formation is a logically sufficient condition to make something epistemology. Objection: Your theory gives too much away to malefactors. In an important, early treatment of doxastic responsibility, James Montmarquet (1993: 4ff) suggested that investigating epistemic responsibility was very important to investigating moral responsibility, for if someone responsibly believed falsehoods that made sense of gravely immoral actions, then failing to criticize doxastic character might be to give such persons a pass. And, of course, we probably agree that the moral dimension is more important than the epistemic one and, ultimately, by far. Thus, saying that some epistemic malfeasor’s doxastic attitude fits his evidence isn’t giving them much at all. It is certainly not giving them enough to counter-balance on the whole their moral shortcomings. On the contrary, according to my way of thinking, the malefactor has made a much worse error than an epistemic error; they have made a moral one. Given the greater gravity of moral badness over epistemic badness, to relegate doxastic irresponsibility to the merely epistemic risks trivializing grave moral evils.   27  The arguments above against sui generis normativity apply here as well. Courage in cycling surely does not indicate a new kind of courage “cycling courage.” No, there is just courage and one can be courageous in lots of things. Prefacing courage by “epistemic” or “cycling” creates the misimpression that the virtue is some new kind of virtue over and above the “plain ol” moral virtue of courage. Furthermore, the value of these virtues is likely in something like the intent behind them: desiring some good. This is essentially admirable, so the display of virtue is always good no matter what the outcome. But this kind of goodness is Aristotelian moral goodness.   28  But see Baehr (2007).

The ‘Ethics of Belief’ is Ethics (Period)  161 Furthermore, we must consider the notion of credit. On my view, even though the epistemically justified but (morally) irresponsible (with respect to beliefs) individual’s doxastic attitude has positive epistemic-axiological status, they deserve no credit for this. Their belief fits the evidence they have, but they are irresponsible for having the evidence they have. This might be so either because they were lazy and didn’t investigate enough29 or because they were sloppy and inattentive30 or trusted the testimony of individuals they acknowledge as being sketchy.31 This vicious activity seems to screen them off from the any credit for the epistemically justified belief and so further assuage the concern that it’s a concession to grant their beliefs positive epistemic status. Objection: Isn’t teleological normativity moral normativity? I have characterized epistemic normativity teleologically. But a failure to achieve one’s telos is a failure to flourish and so, classically speaking, a moral shortcoming. In “Reducing Responsibility” (Dougherty 2011b), I slipped and treated a kind of proximate teleological dysfunction as moral. But it is easy to see that this is mistaken. Not all failures to flourish are moral. The proximate telos of the eye is to see, but blindness is not a moral failing in any sense. If the doxastic system of my mind functions poorly and does not assign credences as it ought, this kind of epistemic poor insight is also not a moral failing. There may be a form of moral evaluation of an agent that emerges organically from an all-things-considered survey of how various teloi are going. However, this is far from the concerns of the responsibilist. Objection: That putative cases of epistemic irresponsibility are coincident with moral or pragmatic irresponsibility does not entail that there is no emergent property of epistemic irresponsibility. This is true. It does not entail it. But it justifies the belief that it is so, for we should not multiply properties without necessity, or at least very good reason. The burden of proof is on the person who wants us to expand our ontology, who wants to graft another branch onto our normative taxonomy. For we are already committed to the other forms of normativity involved in some form or other. That is, responsibilists already grant the existence of moral normativity and pragmatic normativity. So far we are agreed. And I have never seen or heard a responsibilist deny that there is such a thing as a relation of fit between evidence and doxastic states. Certainly it would be a cost for responsibilists to deny the existence of probabilistic relations between propositions. So surely the responsibilist will admit that in some substantive sense one ought to believe in accordance with one’s evidence, just as a consequentialist and a deontologist agree that there is such a thing as moral normativity. So we agree that there is evidential normativity,

  29  They failed to investigate as much as the importance, or at least the importance they attributed, warranted.   30  They manifested a form of “intellectual sloth”—i.e., sloth with respect to belief formation.   31  They did not practice “safe belief.”

162  Trent Dougherty what we could call “pure epistemic normativity.” Now, in addition to these kinds of normativity that we all agree on in some form or other, the responsibilist wants to add a new, sui generis, form of normativity that is “epistemic.” My recipe for reduction shows that this multiplication of properties is not necessary. Objection: You say the ethics of belief is only ethics and not epistemology, but don’t disciplines sometimes overlap? “Disciplines overlap” is ambiguous between, at least, two things. One reading is, roughly, this:  “Some topics count as being in two different disciplines, or sub-disciplines, at the same time.” Another reading is this:  “Some phenomena are studied by people in two different disciplines.” There is a bit of a generality problem here, for it is fairly trivial that nature is the subject of study of physicists, chemists, and biologists. (This may not be the best example because some think, as I once did but no longer do, that biology and chemistry are completely reducible to physics, but we can leave that to one side.) The lever might be studied by both physicists and engineers. (But this isn’t a good example either, since much of engineering is just applied physics.) Perhaps this will help. Aristotle said that some disciplines take as their first principles items proved in lower disciplines. Aquinas, in applying this point to the relationship between natural and sacred theology, uses as an example that the science of music is based on the science of arithmetic. But that does not make musical composition a branch of mathematics. An especially skilled and ambitious composer might also be a mathematician, just as many exceptional physicists are also skilled mathematicians, while ordinary physicists just use a lot of math. “Mathematical physics” seems to me to be a high-level subset of physics to practice that one must also be a mathematician. But that does not entail that one act is, under the very same aspect, both the instance of a physics and of mathematics. One might wear both hats, but not at exactly the same time—though they may change them so quickly we don’t see! Here is a commonsense way to think of this applying to the present concern.32 Qua epistemologist, I  devise a theory of evidential fit. Then, qua doxastically-oriented applied ethicist, I devise a theory of appropriate inquiry aimed at performing actions that bring it about that the evidence I have—and am “obliged” to believe in accordance with—is of the highest quality. If, qua epistemologist, I had erroneously come to the conclusion that epistemic justification consisted in beliefs formed by reliable belief-forming processes, then, qua applied ethicist of belief, I might have devised a theory of inquiry concerning which belief-forming processes are reliable. Or I might, qua epistemologist, be an epistemic value pluralist, thinking that there are multiple good-making features of beliefs, and, qua applied ethicist of belief, devise a theory of inquiry concerning actions aimed at bringing it about that one’s beliefs have any number of these good-making properties. So inquiry can serve epistemic ends, but that does not make it epistemic in any substantive sense. And of course not all inquiry is  

32

  This worry is based on a comment by Robert Audi.

The ‘Ethics of Belief’ is Ethics (Period)  163 conducted outside one’s head. An important source of evidence is reflection. And this is so not just by running thought experiments on ourselves, which is “very a priori” but also in reflecting about our reasoning and even our past inquiry practices. So, if I should inquire via reflection whether my practices of inquiry are liable to lead to high-quality evidence and find that they are not, I thereby acquire evidence against the proposition I believed on the basis of this evidence. But there isn’t even a whiff of commingling of natures here, for it is not news that inquiry yields evidence. Introspection is just one kind of inquiry and higher-order evidence is just one kind of evidence. Objection: We can stipulate out stakes of any kind, moral or prudential, and stipulate out desires or intentions that could be relevant for directing evaluations of instrumental rationality to decisions made in the case. A student infers Q from P and P → Q, but by applying some simple invalid rule other than modus ponens—one that overgeneralizes it, say. Let’s also suppose that the student has propositional justification for P and P → Q. How do you explain why your belief in Q here is not doxastically justified.33 First, not every instance of belief in a proposition for which you have propositional justification is, thereby, doxastically justified. For the belief must also be properly based on that evidence. This entails, among other things, that there is no essential dependence on a falsehood. In this case, there is essential dependence on the invalid rule of inference.34 Furthermore, not assuming the first point, either the student was justified in believing that the erroneous rule was valid or she was not. If she was, then she plausibly has doxastic justification. If she was not justified in believing the inference rule, then, since stakes were stipulated out, it’s just an error, and not all error requires blame. Even if she were “responsible,” as in “the mistake could be attributed to her and not another” (assuming it wasn’t just random, which such mistakes may well be), there is no reason at all to allege epistemic irresponsibility. If she were a TA charged with responsibility for teaching others, then perhaps it would have been irresponsible for her not to double-check her work. But that is different than alleging irresponsibility for making the mistake. Now, “error” seems normative. Calling it a “mistake” indicates something went wrong. But this is mere misfortune. Winning the lottery is a good thing to have happen to you and making a mistake is a bad thing to have happen to you. But there is nothing agential or normative in either case. The mistake only becomes thickly normative when there is something at stake. Maybe she desperately wants to be excellent at logic or maybe she’s a logic TA with a responsibility not to make such a mistake. In such cases we can blame her for not being more careful. But these are clearly failings of practical rationality or morality. The lesson here as elsewhere is: no stakes, no problem.

  33  This is based on an objection by Kurt Zylvan raised in correspondence.   34  For a theory of knowledge that includes no essential dependence on a falsehood example, see Feldman (2003). This move requires rejecting Klein’s notion of a useful falsehood (Smith 2008). For some reason to think that this rejection is warranted, see Dougherty (2010).

164  Trent Dougherty

8.6  Taking Taxonomy Seriously Why do I care about the taxonomy so much? Why is it so important to me to exclude responsibilist concerns from the realm of epistemology proper? I don’t and it’s not. Not much anyway. But I think that taxonomy is a teacher. It is a way of reminding ourselves—those of us, like me, who are very interested in the ethics of inquiry—that we should not be looking for sui generis properties answering to “epistemically irresponsible.” We should really, truly be doing the ethics of belief—that is, ethics. And we should, therefore, educate ourselves more in the way of ethics—especially, in my view, decision theory. And instead of trying to, in my opinion, “pervert” epistemology by remaking it in the image of ethics, we should “become perverts” and actually practice ethics.35

References Audi, Robert. (2008). “The Ethics of Belief:  Doxastic Self-Control and Intellectual Virtue,” Synthese 161(3): 403–18. Axtell, Guy. (1997). “Recent Work on Virtue Epistemology,” American Philosophical Quarterly 34(1): 1–26. Axtell, Guy. (2011a). “Recovering Responsibility,” Logos and Episteme (3): 429–54. Axtell, Guy. (2011b). “From Internalist Evidentialism to Virtue Responsibilism,” in Trent Dougherty (Ed.), Evidentialism and Its Discontents. Oxford: Oxford University Press, 71–87. Axtell, Guy. (2012). “(More) Springs of My Discontent,” Logos and Episteme (1): 131–37. Baehr, Jason. (2007). “Review of Robert C. Roberts, W. Jay Wood, Intellectual Virtues: An Essay in Regulative Epistemology,” Notre Dame Philosophical Reviews (7), URL  =  . Baehr, Jason. (2011a). “Evidentialism, Vice, and Virtue,” in Trent Dougherty (Ed.), Evidentialism and Its Discontents. Oxford: Oxford University Press, 88–102. Baehr, Jason. (2011b). The Inquiring Mind:  On Intellectual Virtues and Virtue Epistemology. Oxford: Oxford University Press. Clifford, W.K. (1999), “The Ethics of Belief,” in T. Madigan (Ed.), The Ethics of Belief and Other Essays. Amherst, MA: Prometheus, 70–96. (Original edition, 1877.) Conee, Earl. (2004). “The Truth Connection,” in Earl Conee and Richard Feldman (Eds.), Evidentialism: Essays in Epistemology. Oxford: Oxford University Press, 242–58.

  35  Thanks are due to the contributors to a thread on this subject on the blog Certain Doubts, especially Clayton Littlejohn, Chris Gadsden, Kurt Sylvan, and the ever effervescent Mr. Andrew Moon. I wish to thank the department of philosophy at the University of Iowa where a version of this chapter was read, especially Brett Coppenger, Evan Fales, Richard Fumerton, Ali Hasan, John Komdat, and Jeremy Shipley for helpful comments. I also presented a version of this chapter at the Rocky Mountain Ethics Congress in 2010. Thanks also go to Robert Audi’s epistemology seminar at the University of Notre Dame, especially Robert Audi, Meg Schmitt, and Kathryn Pogin. Kathryn was a frequent interlocutor and source of objections during my time at Notre Dame. She was a great help, and I thank her. Indeed, she should perhaps be considered an honorary co-author as several of my replies in the objections section began as emails to her. Ironically, she probably thinks much of this utterly irresponsible.

The ‘Ethics of Belief’ is Ethics (Period)  165 Conee, Earl and Feldman, Richard. (1985). “Evidentialism,” Philosophical Studies 48(1): 15–34. Conee, Earl and Feldman, Richard. (2004). Evidentialism:  Essays in Epistemology. Oxford: Oxford University Press. David, Marian. (2005). “Truth as the Primary Epistemic Goal:  A  Working Hypothesis,” in Matthias Steup, Ernest Sosa, and John Turri (Eds.), Contemporary Debates in Epistemology, 2nd edn. Malden, MA: Blackwell, 363–77. DeRose, Keith. (2011). “Questioning Evidentialism,” in Trent Dougherty (Ed.), Evidentialism and its Discontents. Oxford: Oxford University Press, 137–46. Dougherty, Trent. (2010). “Review of Quentin Smith (Ed.), Epistemology: New Essays,” Notre Dame Philosophical Reviews (1), URL= . Dougherty, Trent (Ed.). (2011a). Evidentialism and Its Discontents. Oxford:  Oxford University Press. Dougherty, Trent. (2011b). “Reducing Responsibility:  An Evidentialist Account of Epistemic Blame,” European Journal of Philosophy 20(4): 534–47. Dougherty, Trent. (2011c). “Re-Reducing Responsibility: Reply to Axtell,” Logos and Episteme II: 4, 625–32. Dougherty, Trent. (2012). “Internalist Evidentialism and Epistemic Virtue: Re-Reply to Axtell,” Logos and Episteme III: 2, 281–89. Dougherty, Trent. (MS). “Baehr on Evidence and Virtue:  E-relevant or Irrelevant?” URL= . Feldman, Richard. (1988a). “Epistemic Obligations,” Philosophical Perspectives 2: 235–56. Feldman, Richard. (1988b). “Subjective and Objective Justification in Ethics and Epistemology,” The Monist 71: 405–19. Feldman, Richard. (1995). “Authoritarian Epistemology,” Philosophical Topics 23: 147–70. Feldman, Richard. (1998). “Epistemology and Ethics,” in Peter Klein and Richard Foley (Eds.), Routledge Encyclopedia of Philosophy. London and New York: Routledge. Feldman, Richard. (2000). “The Ethics of Belief,” Philosophy and Phenomenological Research 60(3): 667–95. Feldman, Richard. (2001). “Voluntary Belief and Epistemic Evaluation,” in Matthias Steup (Ed.), Knowledge, Truth, and Duty:  Essays on Epistemic Justification, Responsibility, and Virtue. Oxford: Oxford University Press, 77–92. Feldman, Richard. (2002). “Epistemological Duties,” in Paul Moser (Ed.), The Oxford Handbook of Epistemology. Oxford: Oxford University Press, 361–84. Feldman, Richard. (2003). Epistemology. Upper Saddle River, NJ: Prentice Hall. Feldman, Richard. (2004). “The Ethics of Belief,” in Earl Conee and Richard Feldman (Eds.), Evidentialism: Essays in Epistemology. Oxford: Oxford University Press, 166–96. Feldman, Richard. (2008). “Modest Deontologism in Epistemology,” Synthese 161(3): 339–55. Foley, Richard. (2005). “Justified Belief as Responsible Belief,” in Matthias Steup and Ernest Sosa (Eds.), Contemporary Debates in Epistemology. Malden, MA: Blackwell, 313–25. Grimm, Stephen. (2009). “Epistemic Normativity,” in Adrian Haddock, Alan Millar and Duncan Pritchard (Eds.), Epistemic Value. Oxford: Oxford University Press, 243–64. Heil, John. (1984). “Doxastic Incontinence,” Mind 93(369): 56–70. Kelly, Thomas. (2003). “Epistemic Rationality as Instrumental Rationality:  A  Critique,” Philosophy and Phenomenological Research 66(3): 612–40.

166  Trent Dougherty Korcz, Keith Allen. (2000). “The Causal-Doxastic Theory of the Basing Relation,” Canadian Journal of Philosophy 30(4): 525–50. Kornblith, Hilary. (1983). “Justified Belief and Epistemically Responsible Action,” Philosophical Review 92(1): 33–48. Kornblith, Hilary. (2001). “Epistemic Obligation and the Possibility of Internalism,” in A. Fairweather and L. Zagzebski (Eds.), Virtue Epistemology: Essays on Epistemic Virtue and Responsibility. Oxford: Oxford University Press, 231–48. Kvanvig, Jonathan. (2003). “Justification and Proper Basing,” in Erik Olsson (Ed.), The Epistemology of Keith Lehrer. Dordrecht: Kluwer Publishing Co, 43–64. Kvanvig, Jonathan. (2005). “Truth is not the Primary Epistemic Goal,” in Matthias Steup and Ernest Sosa (Eds.), Contemporary Debates in Epistemology. Malden, MA: Blackwell, 285–95. Kvanvig, Jonathan. (2008). “Pointless Truth,” Midwest Studies in Philosophy 32(1): 199–212. Lehrer, Keith. (1971). “How Reasons Give Us Knowledge, or the Case of the Gypsy Lawyer,” Journal of Philosophy 68(10): 311–13. Leiter, Brian. (2010). “Foundations of Religious Liberty: Toleration or Respect?” San Diego Law Review 47: 935–59. Montmarquet, James. (1993). Epistemic Virtue and Doxastic Responsibility. Lanham, MA: Rowman & Littlefield Publishers, Inc. Plantinga, Alvin. (2000). Warranted Christian Belief. Oxford: Oxford University Press. Roberts, Robert C. and Wood, W. Jay. (2007). Intellectual Virtues:  An Essay in Regulative Epistemology. Oxford: Oxford University Press. Ryan, Sharon. (2003). “Doxastic Compatibilism and the Ethics of Belief,” Philosophical Studies 114(1–2): 47–79. Ryan, Sharon. (2013). “Wisdom,” Stanford Encyclopedia of Philosophy, URL  =  . Smith, Quentin. (2008). Epistemology: New Essays. Oxford: Oxford University Press.

PA RT  I I

The Ethics of Belief: Social

9 The Psychology of Knowledge Formation Its Impetus, Mechanism, and Social Context Arie W. Kruglanski and Lauren M. Boyatzi

The process of knowledge formation has been of principal interest to philosophers since at least the time of Plato. Its fascination is easy to grasp. If having knowledge entails having true beliefs, then unraveling the mystery of knowledge formation promises a method of attaining a most advantageous possession, affording effective actions and desired outcomes. Throughout the epoch of modernity, science has been hailed as the embodiment of such method, and therefore presumed capable of valid discoveries and true insights. Karl Popper (1949) proclaimed truth to be but a “regulating ideal,” one that serves to guide our epistemic pursuits, though it is in principle beyond human reach. Accordingly, subsequent scholars, such as Thomas Kuhn (1962), Paul Feyerabend (1970), and Imre Lakatos (e.g., with Musgrave 1970), re-directed the discourse to descriptive analyses of how knowledge1 is actually acquired. The descriptive stance toward knowledge formation highlights the importance of psychology as a source of individual-level insights into the epistemic process. Indeed, epistemic phenomena have been at the heart of much research carried out over the last half century or so in the domain of social cognition. Though not explicitly devoted to knowledge formation per se, such work addressed, nonetheless, a variety of topics with distinct epistemic relevance. Thus, work on attitude formation and change has dealt with phenomena of evaluative knowledge, i.e., judgments concerning the “goodness” or “badness” of various attitude objects. Research on cognitive consistency issues (see, e.g., Abelson et al. 1968; Gawronski and Strack 2012) dealt with the impact of logically incompatible propositions on judgment formation (Kruglanski   1  In this essay, we will be using the term “knowledge” in its subjective sense, to refer to beliefs held with a degree of confidence above a certain threshold. This reader should note that this practice is common in social psychology but not in philosophy.

170  Arie W. Kruglanski and Lauren M. Boyatzi 1989; Kruglanski and Shteynberg 2012). Social cognitive work on attribution processes pertained to the formation of causal knowledge; research on stereotyping referred to knowledge about persons, whereas inquiries into modes and processes of judgment formation (e.g., Kruglanski and Gigerenzer 2011; Kruglanski, Bélanger, Chen, and Köpetz 2012) referred to ways and means of forming knowledge on a variety of topics. Even though concerns about epistemic processes have been implicit in all such work, the different research programs in this domain have touched on separate aspects of knowledge formation and/or different knowledge contents (e.g., causal contents, social categories, stereotypes). Consequently, the field of psychological epistemics has been fragmented and has lacked a unified framework within which the different insights could be integrated. The theory of lay epistemics (Kruglanski 1989, 2004) has aimed to furnish such a synthesis. The theory has three components, corresponding to the three problems posed by the phenomenon of knowledge formation. The first of these is an inferential element, which addresses the how question of knowledge formation—the mechanism of knowledge creation from relevant evidence. The second is a social element, which addresses the who of knowledge formation—the trusted source of information that we treat as evidence for our judgments. The third is a motivational element, which addresses the why question of knowledge formation—the impetus to form knowledge in the first place. The first, inferential, component assumes that our knowledge derives from subjective (or intersubjective) rules of evidence leading in a quasi-syllogistic fashion from facts to derivative judgments. We are assuming that some rules of inference, like those in the realm of physical perception, are hard-wired, and hence are common to all members of the species. Other rules may be unique to specific epistemic communities whose distinct cultures may contain different beliefs as to which facts warrant which judgments. Other inference rules may be more idiosyncratic, e.g., those subscribed to by a paranoiac whose esoteric conclusions derive from information others may discard as valid evidence for those judgments. The second, social, component concerns the perceived “supply side” of our knowledge. Over the course of our development and socialization, we learn who in our social environment knows what. We then treat their endorsement as inherent evidence for the validity of their pronouncements. In our epistemic quests we rely a great deal on other people. For example, we trust known experts in our community (e.g., physicians, car mechanics, etc.) for answers to specific questions we might have (e.g., how to cure our illness, how to repair our automobile). The touted division of labor within groups is in part a division of epistemic expertise. Transactive memory (Wegner 1986), also known as cognitive interdependence between individuals (Wegner, Giuliano, and Hertel 1985), provides a compelling example. Transactive memory is a system of memory including encoding, storing, and retrieving information that is shared among a set of individuals (Wegner, Giuliano, and Hertel 1985). To the extent that the group learns who within the group knows what, group members

The Psychology of Knowledge Formation  171 may assign memory items to specific group members as well as optimally coordinate retrieval of the information (Wegner 1995). These shared memory structures are better suited than single individuals to successfully perform a wide variety of tasks that call for diverse types of expertise. The third, motivational, component assumes that knowledge formation is purposeful. It is driven by a goal or goals attained by the sense of knowing that one has developed. The purposive nature of our epistemic quest implies selectivity in the information we would attend to and input as grist into our epistemic mill. We have specific questions we desire to answer. These focus us on information we deem relevant to that answer. We may also differ in the magnitude of our epistemic desires. In some cases, we strongly desire epistemic certainty and find ambiguity intolerable. In other cases, we are reluctant to commit to a given judgment lest it be mistaken. Finally, we often have preferences concerning answers that we reach. These motivational concerns may bias the knowledge formation process in directions perceived as likely to yield the desired judgments. In what follows, these aspects of the theory of lay epistemics are described at some length. Collectively, they attest to the centrality of epistemic phenomena to a broad variety of psychological and social psychological issues. We begin this discussion with the psychology of inference and its place in knowledge formation.

9.1  Evidence and Inference: On the Centrality of Rules Cognitive scientists agree that much human knowledge is derived according to rules. These are implicational mental structures of the “if E, then C” type that an individual holds true. The reasoning from evidence to conclusions is syllogistic: the “if E, then C” statement is a major premise, its instantiation, E, in a given situation is the minor premise, and C is the conclusion. Syllogistic reasoning can be represented probabilistically, namely in terms of conditional probabilities as in Bayes’ theorem. According to syllogistic analysis, inferences are made by modus ponens, arrival at the consequent through affirmation of the antecedent. For instance, one may hold a rule that “if the light turns red, the traffic will come to a standstill.” If the light did turn red on a given occasion, one could infer, then, that the traffic would halt. Granting that epistemic judgments are syllogistically derived from rules, this invites the question of how rules are formulated in the first place. This pertains to the testing of hypotheses about possible implicational relations between events so that if event X occurred, event Y may be expected (e.g., if the individual’s goal was frustrated, he or she would feel angry). Hypotheses are tentative rules that need to survive empirical tests to be assuredly used later on. As to how hypotheses are tested, our position differs somewhat from Popper’s falsificationism. According to Popper (1959), hypothesis testing only lends itself to falsification rather then to verification. In contrast, we suggest

172  Arie W. Kruglanski and Lauren M. Boyatzi that hypotheses can be also verified by evidence, granting certain qualifying assumptions. Let us elaborate. Falsification is accomplished by modus tollens where negation of the consequent (i.e., failure to obtain the predicted result) entails a negation of the antecedent and, hence, a disconfirmation of the hypothesis being tested. One starts with the statement of the form “if Hypothesis H is true, then datum D should obtain.” As Popper rightly notes, obtainment of the datum does not prove that H is true because conceivably D could have obtained for reasons other than H as well. On the other hand, if D did not obtain then H could not be true, because H requires the presence of D. The problem is that scientists seem to pervasively violate the falsificationist injunction and happily verify their hypotheses, viewing successful instances of hypothesis testing as providing (positive) support for their hypotheses rather than as reflecting mere survival of falsification attempts. The question is whether scientists’ hypothesis testing behavior is logically inadmissible, and whether no genuine support for hypotheses is possible. Perhaps not. Crucial to Popper’s analysis was the assumption that hypothesis testing departs from an implicational premise of the “if H then D” variety. But what if the departing premise was “H if and only if D”. The latter assumption does allow the verification of hypotheses by relevant data. The question now is whether it is a legitimate assumption to make. We submit that the assumption is, indeed, legitimate within a context of a research design that controls for all the known plausible antecedents of D rather than H. Specifically, in designing an experiment the scientist is assuming, however tentatively, that only if the focal hypothesis (H) is true and all the rival hypotheses (controlled for in the design) are false then the datum (D) should be obtained. This allows the tentative confirmation of a hypothesis in accordance with syllogistic logic. Thus, although scientists may rarely attempt to falsify their hypotheses and may pervasively attempt to verify them, their behavior could be regarded as logically admissible after all.

9.1.1  Single Versus Dual Processes of Judgment Formation A broad epistemic question that has been hotly debated by social and cognitive psychologists has been whether syllogistic rule following is the only way in which individuals form judgments. Over the last several decades, an influential paradigm in psychological theorizing has been a dual process framework, whereby rule following is but one of two possible ways of judgment formation, the other being the associative or the intuitive route. But the question is whether or not the associative process is meaningfully distinct from the rule following process. We argue that they are not distinct processes. In a series of papers (Kruglanski and Thompson 1999a, 1999b; Erb, Kruglanski, Chun, Pierro, Manetti, and Spiegel 2003; Kruglanski, Pierro, Mannetti, and DeGrada 2006; Kruglanski, Pierro, Mannetti, Erb, and Chun 2007; Kruglanski, Dechesne, Orehek,

The Psychology of Knowledge Formation  173 and Pierro 2009; Kruglanski and Gigerenzer 2011), we have argued that the distinction between rule following and associations is unsustainable. The way associations work in affecting judgment and subsequent action is through implicit “if-then” rules (see also Mischel and Schoda 1995). These can be unidirectional of the “if A, then B” variety or bi-directional of the “if A, then B, and if B, then A variety.” That associative processes are, at root, rule-following processes has been acknowledged and commonly accepted in reference to such associative phenomena as classical and evaluative conditioning as well as in domains of unconscious and spontaneous inferences in perception.2 For instance with regard to perceptual inferences, a flat retinal cue may be transformed into a three-dimensional percept in accordance with a hard-wired rule:  “if certain properties of the retinal stimulus are present then the three-dimensional percept is experienced,” etc. It is true that some judgments come to mind quickly and effortlessly, seemingly popping out of nowhere, without much conscious awareness of their origins or of the manner of their formation. It is also true that other judgments seem deliberate: they arise from a lengthy and painstaking thought process that is transparent and accessible to awareness. But this does not require two qualitatively distinct psychological systems of judgment. Rather all of the results adduced in support of the dual systems framework (e.g., see Evans 2008; Evans and Stanovich forthcoming; and Kruglanski forthcoming) can be explained by the assumptions that (1) the consciousness with which rules are applied varies in degree, determined by the degree of routinization, and that (2) the extent of processing, that is, the degree to which rival alternative rules are generated and contrasted, is determined by the individuals’ motivation and cognitive resources (for discussion, see Kruglanski et al. 2012). To summarize, the manner in which rules are applied (i.e., more or less consciously, more or less effortfully) is determined by several continuous variables rather than being dichotomous as the dual systems or process theories have suggested.

9.1.2  Universal Inference Rules: Judgmental Heuristics So far, we have discussed the syllogistic form of the inference process. But what about the contents of such processes? Are they of interest? Or are they so idiosyncratic and multifarious as to be completely case-specific and ungeneralizable? Our answer is that it depends. Gerd Gigerenzer’s (Gigerenzer and Goldstein 1996) work on fast and frugal heuristics has identified some general rules that may have been shaped by the evolutionary process over the millennia of human existence. One such rule, Gigerenzer and his associates argued, is the recognition heuristic which states that if one of two alternatives is recognized, one should infer that it has the higher value on the criterion. For instance, it is larger, stronger, or more famous, whatever the question might be. Another rule example is the fluency heuristic whereby if both

 

2

  See Kruglanski and Gigerenzer (2011) for a discussion.

174  Arie W. Kruglanski and Lauren M. Boyatzi alternatives are recognized but one is recognized faster, one should infer that it has the higher value on the criterion. These general rules are presumed to be universal to the human species and, hence, to be used universally by humans regardless of culture. It is noted that the heuristics identified by Gigerenzer and his colleagues are meant to be descriptive rather than normative. They are assumed to have developed in the course of evolution, hence to be relatively adaptive, though not infallible. In fact, a major source of judgmental errors is the indiscriminate applications of heuristics to novel contexts where they have not been previously tested. In other words, heuristics are assumed to work only in circumscribed ecological contexts rather than generally (for discussion, see Kruglanski and Gigerenzer 2011).

9.1.3  Universal Inference Rules: Universal Rules of Perception Conceptually akin to Gigerenzer’s cognitive heuristics are various, hard-wired, perceptual rules that occasionally give rise to perceptual illusions. Consider Figure 9.1, for example. The dots on its left appear concave, receding into the surface and away from the observer. In contrast, the dots on the right of the figure seem convex, appearing to bulge and extend toward the observer. Intriguingly, these appearances reverse when the page is turned upside down. Now, the previously concave dots appear convex and vice versa. What explains these effects? The visual illusion appears to be based on an inferential rule that bets on two properties of the environment (Kleffner and Ramachandran 1992). One subconsciously assumes a three-dimensional world and uses the shaded parts of the dots to guess in what direction of the third dimension they extend. To make a good guess, one assumes that: 1. Light comes from above (in relation to retinal coordinates) and 2. There is only one source of light. These two structures describe mammalian history, when the sun and the moon were the only sources of light, and only one operated at a time. The brain exploits these assumed structures by using a simple rule of thumb: If the shade is in the upper part of the object then the dots recede into the surface; if the shade is in the lower part of the object then the dots project out from the surface. This visual illusion illustrates that unconscious, fast, and effortless intuitive processes can follow rules, specifically heuristic rules. It also illustrates that the rationality of the rule is ecological and, thus, resides in the match between rule and environment. If there

Figure 9.1  Unconscious inferences by assumed environmental properties: convex and concave perceptions as a function of shading

The Psychology of Knowledge Formation  175 is a three-dimensional world with the two properties described, the rule leads to good inferences; however, if this is not the case, as in the two-dimensional picture in Figure 9.1, the rule leads to a visual illusion. Systematic errors which are due to a reasonable bet on the environment but which fail due to specific, unexpected circumstances are “good errors.” Our point is that good errors are characteristic of every intelligent system (Gigerenzer 2005). Indeed, intelligence entails taking risks and making bets or, to use Jerome Bruner’s felicitous phrase (1973), going beyond the information given.

9.2  The Social/Developmental Aspect of Knowledge Formation: The Concept of Epistemic Authority In our work on the theory of lay epistemics, we have addressed the issue of rule universality versus specificity of inference rules under the rubric of epistemic authority. This term represents the degree to which a given source is assumed to provide valid information about a topic representing a rule of the kind “if source X believes the information to be true, then it is indeed true.” In relevant empirical research it is found that young children view their adult caretakers (typically, their parents) as general epistemic authorities (Raviv, Bar-Tal, Raviv and Houminer 1990), and presumably this is so universally in humans and perhaps in other mammals as well. Subsequently, in the course of development and socialization, trusted epistemic sources are differentiated in accordance with their perceived domains of expertise. Importantly, this includes development of one’s own epistemic authority, or the confidence that one’s own impressions and interpretations are correct. This has to do with one’s own expertise in specific domains as well as with one’s susceptibility to persuasion and the degree to which one regards the authority of an external source as higher or lower than one’s own epistemic authority. One study, for example, examined one’s own epistemic authority in the domain of mathematics (Ellis and Kruglanski 1992). The results show that, controlling for actual mathematical ability, individuals who had lower self-ascribed epistemic authority in mathematics benefited more from the explanations of an instructor (an external source of authority in math) and less from actual experience in solving mathematical problems, than individuals who had a higher self-ascribed epistemic authority in math.

9.2.1  Do Different Cultures Think Differently? Existence of different inferential rules across life span development relates to the question of whether different cultures think or know differently, a claim advanced a number of years ago by Richard Nisbett and his colleagues. Specifically, these authors argued that whereas members of the Western cultures think analytically, members of

176  Arie W. Kruglanski and Lauren M. Boyatzi the Eastern cultures think holistically (Nisbett, Peng, Choi and Norenzayan 2001). This was assumed to mean that the Easterners pay more attention to the context of an action whereas the Westerners pay more attention to the actor (for discussion, see Nisbett et al. 2001). Easterners were assumed to be more tolerant of contradictions whereas Westerners to be more upset by contradictions. Easterners were assumed to base their judgments more on experience whereas Westerners were assumed to base their judgments more on logic (Nisbett et al. 2001). Easterners were assumed to pay particular attention to the background or the context and Westerners were assumed to pay greater attention to the figure or the actor. However, the relevant questions are at what level of analysis should one interpret the undeniable cultural differences observed by Nisbett et al. (2001), and should they be viewed as attesting to fundamental differences in thought processes between distinct cultures. The empirical data cited by Nisbett et al. (2001) can be readily reinterpreted to suggest that whereas the contents of inferential rules may differ across cultures, the general rule-following process seems to be common to all cultures. For instance, Easterners may subscribe to a greater extent than do Westerners to rules assigning greater causal force to the context (e.g., “if behavior X occurred, then context C was the cause”), whereas Westerners may subscribe to a greater extent to rules that assign greater causal force to the individual (e.g., “if behavior X occurred then the actor A was the cause”). Other data cited by Nisbett et al. (2001) can be explained by greater accessibility of some rules to members of the Eastern versus Western cultures, or greater availability of such rules to some rather than other cultures (for a discussion, see Kruglanski, Dechesne, and Chun 2004). Accessibility of rules may be determined by the situational saliency of their components. These, in turn, may be affected by motivations shared by members of a given culture. For example, the presumed differential attention to the figure, rather than the background, of Westerners versus Easterners (Masuda and Nisbett 2001) may be affected by the differential saliency of each for Easterners and Westerners. The saliency of the figure and ground, in turn, may be driven by differences between Eastern and Western cultures in collectivistic versus individualistic motivations (Markus and Kitayama 1991). In this vein, Kuhnen and Oyserman (2002) demonstrated that in a condition where American participants were induced to think of themselves as dependent upon others, or exhibit a more collectivistic orientation, as is the case in the Asian cultures, they showed greater attention to the context or the background, just as the Asian subjects did in Masuda and Nisbett’s (2001) studies. Therefore, it may be argued that both cultures pay attention to what is motivationally salient. The ground is typically, though not universally, more salient in collectivistic Asian cultures, whereas the figure is typically more salient for members of the individualistic Western cultures. In turn, differential salience of ground versus figure may determine the invocation of different inference rules incorporating those categories, which ultimately produces different effects for members of the different cultures from seemingly the same array of information.

The Psychology of Knowledge Formation  177 Research (Morris and Peng 1994; Nisbett et al. 2001) has also shown that Eastern and Western cultures differentially attribute another’s actions to the situation versus the individual’s disposition. Specifically, Asians are more likely to use situational factors to explain events while Americans are more likely to utilize dispositional influences (Morris and Peng 1994). Thus, one may argue that Asians follow the rule “if making an attributional judgment, pay attention to situational influences” and Americans follow the opposite rule “if making an attributional judgment, focus on dispositional factors.” Other research, however, provides evidence that a stronger focus on situational or dispositional influences can both be found within the American culture and that one’s natural tendency this way or that may be reversed by appropriate experimental manipulations (Krull 1993; Webster 1993). Thus, it appears that both Eastern and Western cultures have in their repertories actor-based as well as situation-based inference rules and it is the momentary saliency of the actor or the situation that determines which of these two types of rules will be rendered accessible and actionable.

9.3  The Motivational Component of Epistemic Behavior: The Aim of Knowledge Formation A quintessential question for philosophers of science has concerned its aims. Translated into psychological terms, this raises the issue of epistemic motivation, namely the question: what needs, motives, or desires are individuals striving to satisfy in forming a judgment? This issue may be more complex than meets the eye. Ostensibly, conscious judgment formation aims to supply the individual with a confident belief. Intriguingly, these two components have opposing effects on the judgment formation process. An emphasis on confidence shortens the process of considering alternative data possibly relevant to a conclusion (i.e., it cuts short the consideration of alternative rules linking those data to some conclusion). In contrast, the emphasis on truth lengthens that process and prompts a careful evaluation of multiple inference rules. We have ample experimental evidence that increasing the emphasis on confidence invites closed mindedness and imperviousness to new information, whereas the emphasis on accuracy or truth invites open mindedness and “unfreezes” prior judgments. For example, the emphasis on confidence or cognitive closure augments the use of stereotypes that afford quick and relatively-thoughtless judgment, whereas the emphasis on the validity of one’s judgment liberates people from stereotypic thinking and prompts them to consider case specific or individuating information (Kruglanski and Freund 1983; Jamieson and Zanna 1989; Dijkersterhuis, Van Knipperberg, Kruglanski, and Schaper 1996). Concern with confidence and closure, versus truth and accuracy, has intriguing implications for human social behavior. We have evidence (Kruglanski and Webster 1996; Kruglanski 2004; Kruglanski, Pierro, Mannetti, and DeGrada 2006) that heightening the need for closure increases individuals’ striving for consensus, intolerance of

178  Arie W. Kruglanski and Lauren M. Boyatzi diversity, ingroup favoritism and political conservatism (Jost, Glaser, Kruglanski, and Sullaway 2003). It should be noted that exploring the potential disconnect between confidence and truth does not exhaust the complexity of the motivational basis underlying judgment. Whereas ostensibly the aim of judgment is confident truth, whatever it might be, philosophers, psychologists, and others have long known that judgments may be motivationally biased. In other words, people may not only desire to have the truth, but rather for a particular kind of truth, or a particular kind of closure that satisfies an ulterior need. Intriguingly, those biasing motivations operate unconsciously. They bias individuals’ judgments in the desired direction while the individual is assuming all the time that confident truth is all that matters. After all, “wishing does not make it so” and reaching conclusions just because they are desirable conflicts with the intended aim of judgment, namely truth. Accordingly, biasing motivations unconsciously privilege information that is consistent with the desired conclusions and generate rules that favor such conclusions while at the same time suppressing evidence contrary to such conclusions. Several studies provide evidence that an individual’s motivation can bias cognition by providing participants with the same information but evoking different responses based on the individuals’ motivations or goals. For example, Sanitioso, Kunda, and Fong (1990) instructed participants to consider autobiographical memories that were relevant to the personality dimension of introversion-extraversion. Some participants were told that introversion was desirable whereas others were told that extraversion was a valuable trait. Results showed that participants who believed that introversion was desirable recalled instances of introversion first and wrote down more of these instances than examples of extraversion, and vice versa. Arrival at motivationally-desirable conclusions can require quite a bit of cognitive work involved in suppressing, counter-arguing, or explaining away undesirable information while privileging information and inference rules that support the desired conclusions. In turn, such work may require an appreciable amount of mental energy. The energetic aspects of motivational biases have only recently come into focus of social cognitive research (for discussion, see Kruglanski et al. 2012).

9.3.1  The Motivation for Cognitive Consistency: Is It Real? The process of judgment formation and the motivational basis of subjective knowledge is, finally, relevant to the issue of cognitive consistency, a topic that has arguably occupied social psychologists more than any other and has given rise to such well know theories as Festinger’s (1957) theory of cognitive dissonance, Heider’s (1958) theory of cognitive balance, and other cognitive consistency frameworks. The cognitive consistency theme has prompted the appearance of a famous source book on this topic (Abelson, Aronson, McGuire, Newcomb, Rosenberg, and Tannenbaum 1968). A contemporary version edited by Bertram Gawronski and Fritz Strack (2012) was published recently.

The Psychology of Knowledge Formation  179 It is of interest to note that contributors to both handbooks have implied that cognitive consistency constitutes a fundamental human motive. From the present theoretic perspective, however, this isn’t quite so. On this account, cognitive consistency pertains to a confirmation of an abstract hypothesis by a relevant observation. The inference rule could be “if X was affirmed, then Y was appropriately concluded.” As this example shows, cognitive consistency represents a positive outcome of a test in which the hypothesis survived its potential disconfirmation, allowing for it to be maintained. In contrast, cognitive inconsistency refers to disconfirmation of a hypothesis. An inconsistency is a marker that something is wrong and that one’s belief system contains an inaccuracy. Crucially, however, neither hypothesis confirmation nor disconfirmation have any unique motivational significance as such. Instead, the motivational impact of confirmation and disconfirmation depends on whether the particular knowledge that the hypothesis represents was desirable or undesirable to the individual in given circumstances (Kruglanski 1989). Accordingly, the affective consequences of cognitive consistency or inconsistency are contingent: they depend entirely on the individual’s epistemic motivation, which is a motivation to validate or invalidate a given knowledge claim (Kruglanski 1989). Consistency theorists typically implied that the validation of knowledge, afforded by the cognitive consistency outcome, is generally more desirable than invalidation. It is for this reason that they regarded inconsistency as aversive and tension-inducing (Festinger 1957). However, common experience tells us otherwise. Whether validation or invalidation is more desirable should depend entirely on the kind of knowledge that the inconsistency undermines, and in particular on the positive or negative value the individual places on such knowledge. To the extent that the knowledge was deemed highly desirable, and its possession constituted an important goal, undermining it might indeed be upsetting to the individual. In these circumstances she or he might well invest effort in resolving the inconsistency (see Abelson 1968), so that coherent knowledge on this topic may be formed. However, if the knowledge was only of minor or no importance to the person, the upset occasioned by the inconsistency should be correspondingly minimal. In other cases the knowledge in question might be downright undesirable for the individual (e.g., the belief that she is afflicted by a dangerous illness, that she had failed an important exam, or that her loved ones were involved in a serious accident). Hence, its unraveling by an inconsistency would represent a welcome relief. Now, cognitive consistency theorists claimed that epistemic motivations matter. However, because they conceived of consistency as a goal in itself rather than as the means of knowledge confirmation, they were rather vague regarding the conditions under which inconsistency will be more or less upsetting. For instance, with regard to balance striving, several writers (e.g., Freedman 1968: 502; Zajonc 1968: 347) suggested that they are of a relatively weak intensity, more like aesthetic preferences rather than driving forces. And in Festinger’s (1957) formulation, the magnitude of dissonance is a function of the importance of the cognitions involved. Yet, Festinger (1957) did not

180  Arie W. Kruglanski and Lauren M. Boyatzi specify clearly what “importance” is. Nor did other cognitive authors who attempted to address the issue (e.g., Brehm and Cohen 1962: 228–31; Pepitone 1966;1968; Wicklund and Brehm 1976: 256). The present approach to this problem is straightforward: we distinguish between inconsistency, which constitutes a means of knowledge dissolution, and the goal of knowledge maintenance or confirmation, of which magnitude and valence can vary widely across persons and situations. We believe that the foregoing analysis solves the problem and delineates the conditions under which inconsistency will be more or less upsetting to individuals, and the preference for consistency will be more or less intense.

9.4 Conclusion Like many other topics that have been initially treated in philosophy, the process of knowledge formation (whether lay or scientific) has been further developed by psychological and social psychological researchers. In this sense, three parallels between philosophical and psychological concerns are striking. First, what in the philosophy of science was discussed as the logic knowledge formation (Popper’s 1934 Logik der Forschung) in psychology has been studied as the mechanism of inference. Second, what in philosophy of science was discussed as the aims of science was in psychology studied under the rubric of epistemic motivations. Finally, philosophers’ interest in paradigms (Kuhn 1962) and research programs (Lakatos and Musgrave 1970) finds its parallels in the concepts of epistemic authority and of shared reality (Hardin and Higgins 1996), including common inference rules and heuristics, and common motivational commitments characterizing members of a given epistemic collectivity.

References Abelson, R. P., Aronson, E., McGuire, W. J., Newcomb, T. M., Rosenberg, M. J., and Tannenbaum, P. H. (1968). Theories of Cognitive Consistency: A Sourcebook. Rand-McNally: Chicago. Brehm, J. W., and Cohen, A. R. (1962). Explorations in Cognitive Dissonance. New York: Wiley. Bruner, J. S. (1973). Beyond the Information Given:  Studies in the Psychology of Knowing. Oxford: Norton. Dijksterhuis, A. P., Van Knippenberg, A. D., Kruglanski, A. W., and Schaper, C. (1996). “Motivated Social Cognition: Need for Closure Effects on Memory and Judgement,” Journal of Experimental Social Psychology 32: 254–70. Ellis, S., and Kruglanski, A. W. (1992). “Self as Epistemic Authority: Effects on Experiential and Instructional Learning,” Social Cognition 10: 357–75. Erb, H.P., Kruglanski, A. W., Chun, Y.W., Pierro, A., Manetti, L, and Spiegel, S. (2003). “Searching for Commonalities in Human Judgment:  The Parametric Unimodel and its Dual-Model Alternatives,” European Review of Social Psychology 14: 1–49. Evans, J. S.  B. T. (2008). “Dual-Processing Accounts of Reasoning, Judgment and Social Cognition,” Annual Review of Psychology 59: 255–78.

The Psychology of Knowledge Formation  181 Evans, J. S.  B. T. and Stanovich, K. E. (2013). “Dual-Process Theories of Higher Cognition: Advancing the Debate,” Perspectives on Psychological Science 8(3): 223–41. Festinger, L. (1966). A Theory of Cognitive Dissonance. Stanford:  Stanford University Press. (Original edition, Evanston, IL: Row and Peterson, 1957.) Feyerabend, P. K. (1970). “Consolation for the Specialist,” in I. Lakatos and A. Musgrave (Eds.), Criticism and the Growth of Knowledge. Cambridge: Cambridge University Press, 197–230. Freedman, J. L. (1968). “How Important is Cognitive Consistency?” in R. P. Abelson, E. Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, and P. H. Tannenbaum (Eds.), Theories of Cognitive Consistencies: A Sourcebook. Chicago: Rand McNally, 497–503. Gawronski, B., and Strack, F. (Eds.) (2012). Cognitive Consistency: A Fundamental Principle in Social Cognition. New York: Guilford Press. Gigerenzer, G. (2005). “I Think; Therefore, I Err,” Social Research 72: 195–218. Gigerenzer, G. and Goldstein, D.G. (1996). “Reasoning the Fast and Frugal Way:  Models of Bounded Rationality,” Psychological Review 103: 650–69. Hardin, C. D., and Higgins, E. T. (1996). “Shared Reality: How Social Verification Makes the Subjective Objective,” in E. T. Higgins and R. M. Sorrentino (Eds.), Handbook of Motivation and Cognition: The Interpersonal Context, Vol. 3. New York: Guilford Press, 28–84. Heider, F. (1958). The Psychology of Interpersonal Relations. Hoboken, NJ:  John Wiley and Sons Inc. Jamieson, D. W., and Zanna, M. P. (1989). “Need for Structure in Attitude Formation and Expression,” in A. R. Pratkanis, S. J. Breckler, and A. G. Greenwald (Eds.), Attitude Structure and Function. Hillsdale, NJ: Lawrence Erlbaum Associates Inc., 398–401. Jost, J. T., Glaser, J., Kruglanski, A. W. and Sullaway, F. J. (2003). “Political Conservatism as Motivated Social Cognition,” Psychological Bulletin 129: 339–75. Kleffner, D. A., and Ramachandran, V. S. (1992). “On the Perception of Shape from Shading,” Perception and Psychophysics 52: 18–36. Kruglanski, A. W. (forthcoming). “Only One? The Default-Interventionist Framework as a Unimodel,” Perspectives on Psychological Science 8(3): 242–7. Kruglanski, A. W. (1989). Lay Epistemics and Human Knowledge: Cognitive and Motivational Bases. New York: Plenum Press. Kruglanski, A. W. (2004). The Psychology of Closemindedness. New York: Psychology Press. Kruglanski, A. W., Bélanger, J., Chen, X. and Köpetz, C. (2012). “The Energetic of Motivation: A Force Field Analysis,” Psychological Review 119: 1–20. Kruglanski, A. W., Dechesne, M. and Chun, W. Y. (2004). “Culture, Thought and the Unimodel,” Journal of Cultural and Evolutionary Psychology 2: 195–320. Kruglanski, A. W., Dechesne, M., Orehek, E. and Pierro, A. (2009). “Three Decades of Lay Epistemics: The Why, How, and Who of Knowledge Formation,” European Review of Social Psychology 20: 146–91. Kruglanski, A. W., and Freund, T. (1983). “The Freezing and Un-Freezing of Lay-Inferences: Effects on Impressional Primacy, Ethnic Stereotyping and Numerical Anchoring,” Journal of Experimental Social Psychology 19: 448–68. Kruglanski, A. W. and Gigerenzer, G. (2011). “Intuitive and Deliberative Judgments are Based on Common Principles,” Psychological Review 118: 97–109.

182  Arie W. Kruglanski and Lauren M. Boyatzi Kruglanski, A. W., Pierro, A., Mannetti, L., and DeGrada, E. (2006). “Groups as Epistemic Providers: Need for Closure and the Unfolding of Group-Centrism,” Psychological Review 113: 84–100. Kruglanski, A. W., and Shteynberg, G. (2012). “Cognitive Consistency as a Means to an End:  How Subjective Logic Affords Knowledge,” in B. Gawronski and F. Strack (Eds.), Cognitive Consistency: A Fundamental Principle in Social Cognition. New York: Guilford Press, 245–64. Kruglanski, A. W., and Thompson, E. P. (1999a). “Persuasion by a Single Route: A View from the Unimodel,” Psychological Inquiry 10: 83–110. Kruglanski, A. W., and Thompson, E. P. (1999b). “The Illusory Second Mode, or the Cue is the Message,” Psychological Inquiry 10: 182–93. Kruglanski, A. W., Pierro, A., Mannetti, L., Erb, H.-P., and Chun, W.Y. (2007). “On the Parameters of Social Judgment,” Advances in Experimental Social Psychology 39: 255–303. Kruglanski, A. W. and Webster, D.M. (1996). “Motivated Closing of the Mind: ‘Seizing’ and ‘Freezing’,” Psychological Review 103: 263–83. Krull, D. S. (1993). “Does the Grist Change the Mill? The Effect of Perceiver’s Goal on the Process of Social Inference,” Personality and Social Psychology Bulletin 19: 340–8. Kuhn, T. S. (1962). The Structure of Scientific Revolution. Chicago: University of Chicago Press. Kuhnen, U. and Oyserman, D. (2002). “Thinking About the Self Influences Thinking in General: Cognitive Consequences of Salient Self-Concept,” Journal of Experimental Social Psychology 38: 492–9. Lakatos, I. and Musgrave, A. (Eds.). (1970). Criticism and the Growth of Knowledge. Cambridge: Cambridge University Press. Markus, H. R. and Kitayama, S. (1991). “Culture and the Self:  Implications for Cognition, Emotion, and Motivation,” Psychological Review 98: 224–53. Masuda, T. and Nisbett, R. E. (2001). “Attending Holistically Versus Analytically: Comparing the Context Sensitivity of Japanese and Americans,” Journal of Personality and Social Psychology 81: 922–34. Mischel, W. and Shoda, Y. (1995). “A Cognitive-Affective System Theory of Personality: Reconceptualizing the Invariances in Personality and the Role of Situations,” Psychological Review 102: 246–68. Morris, M. W. and Peng, K. (1994). “Culture and Cause: American and Chinese Attributions for Social and Physical Events,” Journal of Personality and Social Psychology 67: 949–71. Nisbett, R. E., Peng, K., Choi, I., and Norenzayan, A. (2001). “Culture and Systems of Thought: Holistic Versus Analytic Cognition,” Psychological Review 108: 291–310. Pepitone, A. (1966). “Some Conceptual and Empirical Problems of Consistency Models,” in S. Feldman (Ed.), Cognitive Consistency. New York: Academic Press, 257–97. Pepitone, A. (1968). “The Problem of Motivation in Consistency Models,” in R. P. Abelson et al. (Eds.), Theories of Cognitive Consistency: A Sourcebook. Chicago:  Rand McNally and Cy, 319–26. Popper, K. R. (1934). Logik der Forschung. Vienna: Springer. Popper, K. R. (1959). The Logic of Scientific Discovery. London: Hutchinson. Raviv, A., Bar-Tal, D., Raviv, A., and Houminer, D. (1990). “Development in Children’s Perception of Epistemic Authorities,” British Journal of Developmental Psychology 8: 157–69.

The Psychology of Knowledge Formation  183 Sanitioso, R., Kunda, Z., and Fong, G. (1990). “Motivated Recruitment of Autobiographical Memories,” Journal of Personality and Social Psychology 59: 229–41. Webster, D. M. (1993). “Motivated Augmentation and Reduction of the Overattribution Bias,” Journal of Personality and Social Psychology 65: 261–71. Wegner, D. M. (1986). “Transactive Memory: A Contemporary Analysis of the Group Mind,” in B. Mullen and G. R. Goethals (Eds.), Theories of Group Behavior. New York: Springer-Verlag, 185–208. Wegner, D. M. (1995). “A Computer Network Model of Human Transactive Memory,” Social Cognition 13: 319–39. Wegner, D. M., Giuliano, T., and Hertel, P. (1985). “Cognitive Interdependence in Close Relationships,” in W. J. Ickes (Ed.), Compatible and Incompatible Relationships. New York: Springer-Verlag, 253–76. Wicklund, R. A., and Brehm, J. W. (1976). Perspectives on Cognitive Dissonance. Hillsdale, NJ: Erlbaum. Zajonc, R. B. (1968). “Attitudinal Effects of Mere Exposure,” Journal of Personality and Social Psychology 9: 1–27.

10 Perspectives on Social Knowledge Mark Dechesne and Charlotte de Roon

Two central issues in the ethics of belief are clarifying how a person ought (1) to gather and (2) to weigh evidence. One cannot achieve either of these aims virtuously without a proper regard for the nature of evidence. This chapter considers how social psychology has sought to gather and weigh evidence to develop an understanding of some of the most important knowledge domains: ourselves and our social world. What is the right methodology to understand ourselves and our social surroundings? What should be the structure of our beliefs about social life? And what type of evidence should we consider? The chapter highlights how challenging it is to come to social knowledge, not merely for “lay” people but even for “professionals.” Indeed, social psychology has a long history characterized by multiple answers to the questions just raised, whereby two quite distinct perspectives on social knowledge are particularly salient. Considering this history may inform virtue epistemology about perspectives on epistemic norms regarding ourselves and the social world. At the same time, we also seek to answer a fundamental question hardly ever asked: What is ethical for a social psychologist to believe? Before addressing this question at the end of the chapter, let us first consider in greater detail the two perspectives on social knowledge.

10.1  Wundt’s Heritage The founding father of psychology on the European side, Wilhelm Wundt, already came to the conclusion that psychology is not one discipline, but two.1 Two disciplines, with each having a unique tradition, set of assumptions to work from, and methodological approach. Wilhelm Wundt is the builder of one of the first, if not the first, experimental psychological laboratory in the world, in Leipzig. (Although some would argue that Wundt already had a lab in the Haubstrasse in Heidelberg.) With his laboratory he intended to create facilities that would enable the systematic empirical study   1  See for example Rieber and Robinson (2001).

Perspectives on Social Knowledge  185 of basic psychological processes. Using his laboratory, and methods derived from the natural science, he explored the fundamentals of human sensation and perception. But Wundt also realized there was more to psychology than basic human functions. There was more to human lives than motor responses, sensation, perception, memory, etc. There was also a need for a psychology of the people, or a folkpsychology. Humans, so Wundt reasoned, go about their daily lives engaging with others, talking to each other, creating shared understandings, and in doing so continuously create culture. Culture in the form of shared experiences and memories, informal or formal agreements, myths about daily life, and life’s ultimates. And if one wants to understand the human mind, then, these forms of culture are of critical importance. They regulate human behavior and the human mind in the context where it really matters: in people’s everyday life. And indeed, Wundt took his mission of a second, folk psychology very seriously. It was written down in a bulky ten-volume series (Wundt 1916). The implications of Wundt’s folkpsychology are certainly not restricted to the psychology of that time. They hold importance for psychologists up to this day, especially those that are concerned with understanding social life. Perhaps the most important implication is that, following Wundt’s ideas about the socio-cultural origins of mental processes, it makes no sense to study those aspects of human life that implicate social meanings (and what aspects of human life are not affected by meanings?) by isolating behaviors and interactions to a situation outside of the context that provides these meanings. That is, it makes no sense to study these fundamentally social and meaning-suffused processes by studying them in a laboratory. Rather, one needs to study mental processes where they occur: in their natural setting. Secondly, and also importantly, by imposing external, observer-based assumptions on what drives mental processes, we may miss out on the very relevant personal meanings. Thus, rather than starting from an hypothesis derived from abstract theory, it might be better, if we really were to understand mental processes in their social context, to see and ask how people themselves construe their lives. However, one needs to only look at a recent issue of a social psychological journal to find out that the actual practice, the dominant paradigm, if not unbreakable dogma, of contemporary social psychology suggests quite a different, perhaps even downright oppositional approach to understanding social life.2 When it comes to the most prestigious journals in the fields, such as the Journal of Personality and Social Psychology, or the Journal of Experimental Social Psychology, above anything else, the laboratory experiment seems to dominate the journal pages. Thus, within social psychology, knowledge is preferably obtained through experimentation. And thereby, social life is preferably understood by taking people outside of their natural social life. Furthermore, in order to be acceptable, the research should pertain, and preferably add to, existing theory in social psychology. Add to knowledge that is shared among

 

  See for example Wilson, Aronson, and Carlsmith (2010).

2

186  Mark Dechesne and Charlotte de Roon a group of scientists who prefer to be outside of social life and observe social life from an outsider perspective. This knowledge is certainly not shared among the group of people that live social life everyday on the basis of shared meanings, in a natural, unreflected context. Wilhelm Wundt’s thinking and the current state of affairs in social psychology are of interest in the light of a consideration of the ethics of belief. In particular, these contrasting perspectives can be considered cases that illustrate how people develop beliefs about themselves and their social context. They represent two separate perspectives on “virtuous” social knowledge. The experimental practice of the contemporary social psychology corresponds with a critical rationalist perspective, which relies on the empirical validation of theoretical viewpoints. Wilhelm Wundt’s folkpsychology can be equated to the social constructionists’ approach, which states that the focus of social research should be on the stories that scientists and laypeople construe to make sense of their lives and events. For each perspective, evidence is handled and weighed differently. What is considered evidence if one adopts one perspective is not necessarily considered evidence in the other perspectives. But both set the ambitious task of understanding social life. In this chapter, we would like to elaborate on the two perspectives on social knowledge: the experimentalists’ practice of contemporary social psychology and its social constructionists’ counterpart. The two perspectives may actually not be mutually exclusive, and seen from a particular angle, they may even be quite compatible. Further, if the perspectives are not incompatible, then we should also consider what the perspectives have in common. Do they, for example, rely on a similar knowledge structure? We want to argue that they do. And once this is argued, and the pros and cons of an integrative position are considered, we inevitably encounter the challenge to develop a unified framework to come to social knowledge.

10.2  Two Perspectives on Social Knowledge Let us consider the two perspectives on social knowledge in greater detail. We start with the most common idea in social psychology to come to knowledge about social life.

10.2.1  The Experimentalists The experimentalists refer to the strand of scholars that consider the test of hypotheses through experimentation that are derived from general, abstract theory, the primary practice to develop knowledge about the social world. For philosophers of science, this will bring to mind associations with traditions such as the constructive empiricism and critical rationalism. The latter has been developed by Karl Popper (1902–94). Critical rationalism is generally considered to be a critical response to logical positivism because it seeks to replace the method of verification, a process of empirical validation of hypotheses

Perspectives on Social Knowledge  187 based on systematic observation and inductive reasoning, with the method of falsification, a process of empirical elimination of hypotheses based on systematic testing of hypotheses and deductive reasoning (Ulrich 2006). But what these perspectives have in common, is that they rely on the empirical assessment of theoretical viewpoints by testing hypotheses. Hypotheses are often tested through experimentation, for which there are various experimental designs available. Popper explains his position: Observations or experiments can be accepted as supporting a theory (or a hypothesis, or a scientific assertion) only if these observations or experiments are severe tests of the theory—or in other words, only if they result from serious attempts to refute the theory, and especially from trying to find faults where these might be expected in the light of all our knowledge. (Popper 1994: 89)

Social scientists adhering to this perspective create an artificial social situation in a laboratory, where they seek to obtain evidence that either falsifies or corroborates hypotheses about the social world. The artificial “social situation” is often comprised of no more than a small cluster of cubicles at a university. Each cubicle has a computer on a desk. On the computer screen social stimuli appear. These stimuli are expected to elicit a response, whereby these responses are expected to differ across participants depending on the nature of the stimulus. This type of research is inherently hypothesis-driven. The whole setup only makes sense if the hypothesis under consideration is taken into account. There are no other indicators of the social life, other than what is studied. It is a common practice, indeed, to use this setup. A particularly famous experiment has been carried out by Stanley Milgram (1963) on obedience to authority figures. The study took place in a laboratory of Yale University in 1961. Inspired by the Eichmann trials of that year, the study intended to measure in an experimental setting the willingness of naive subjects to obey to an authority figure who instructed them to administer increasingly severe electrical shocks to a victim, an act that conflicted with their personal conscience. The main result was that rational, mentally sane adults were willing to go to almost any length on the command of an authority figure. More than half of the participants were willing to administer the experiment’s final massive shock, even though many were very uncomfortable doing so. Nowadays, the use of the experiments in social psychology is widespread. One sees it in the study of persuasion (Kruglanski and Thompson 1999) where it is used to specify how people are persuaded by particular messages or the extent to which evidence can alter beliefs. But one also sees it in the study of attitudes and social cognition where the experiment is used to investigate the antecedents of prejudicial cognitions and other social perceptions (e.g., Macrae and Quadflieg 2010). It is also used to study human motivation, including the study of the basic drivers of human action (Kesebir and Pyszczysnki 2012). In fact, the practice of experimentation is considered the standard for the field, and no one really questions its utility for coming to understanding the social components of the human mind. No one questions the use of the artificial environment of the laboratory, far removed from the homes, sororities, bars, where

188  Mark Dechesne and Charlotte de Roon the actual social life of the college student participants takes place. No one questions whether this artificial environment of the laboratory is most optimally suited if suited at all, to get to know people. Indeed, those who do question it, are no longer considered part of the same practice. They are no longer considered social psychologists. They have become “social constructionists.”

10.2.2  The Social Constructionists The social constructionists have their main proponent and vanguard in Kenneth Gergen (e.g., Gergen and Gergen 2010). In 1973 he produced the classic yet highly controversial article “Social Psychology as History.” It caused quite a stir indeed. In it, Gergen argued that social psychology was very much on the wrong track. If it were to be the true holder of knowledge about social life, it should better abandon its experimental practices. Experiments are considered to reveal the truth of general principles, that in Gergen’s view, may actually be more about contemporary history rather than about general truth. Social psychology is even assumed to be an inherently historical discipline, because in everyday interaction, there are no abstract principles that govern the process. There are no universal principles. Rather, people, in their interactions with others, continuously create, recreate, negotiate, and renegotiate, the rules, and laws (formal and informal) that guide their behavior, and their expectancies of others. We are back at Wundt’s original idea: if we want to understand the psychology of everyday life, of social life, and not the psychology of basic psychological processes, we should consider the way people makes sense of their lives, we should consider the stories they tell to themselves and to others. We should consider the myths they create and use. And importantly, we should do this at the place where the stories, myths, and rules, are created, used, and changed in the first place. The only place where it really matters: in everyday life. To understand social life means to go to the people: to ask what matters to them in real life, what they belief in, what sort of practical solutions they have come up with to deal with the flotsam and jetsam of everyday life. There are no abstract principles and rules. The validity of experiments is not to be considered superior over other forms of research. There are no universal principles that can be studied outside of the natural context in which social behavior occurs. Rather, there are social constructions, and we should focus on them if we want to understand social life. This seems to be the core message of the social constructionists’ stance. Concomitantly, inspired by Gergen, Vivian Burr (2003) identifies the following key assumptions at the heart of social constructionism. First, social constructionism denies the idea that our observations of the world are objective and unbiased. It cautions us to be ever suspicious of our assumptions about how the world appears to be. Second, Burr states that “the ways in which we commonly understand the world, the categories and concepts we use, are historically and culturally specific” (3–4). Thus, all ways of understanding are tied to culture and to be considered artifacts of that culture.

Perspectives on Social Knowledge  189 Third, because knowledge cannot be derived from the nature of the world as it really is, social constructionism assumes that people construct knowledge through social processes and interaction. These processes and interactions are, therefore, of great interest to social constructionists. Finally, knowledge and social interaction go together, in the sense that descriptions or constructions of the world afford particular patterns of social interaction and exclude others. In this context, patterns of interaction are bound up with power relations because these patterns allocate roles within a social hierarchy to individuals. Within the constructionists’ tradition qualitative methods are used to analyze the social meanings of attribution and discourse. The unstructured interview is a particularly cherished instrument: “Constructionist interviews are dialogical performances, social meaning-making acts, and cofacilitated knowledge exchanges” (Koro-Ljunberg 2008). An example can be found in the discursive psychology, which is primarily concerned with the performative functions of language, such as how people use language in their everyday interactions. As social constructionism, it aims to construct reality by analyzing everyday discourse.

10.3  Case: The Psychology of Terrorism It is of interest to consider the divide between the experimentalists and social constructionists in the light of a specific case. Let us consider the psychology of terrorism. This area of research has witnessed immense growth in the past decade after the attacks of September 11th of 2001 (e.g., Schmid 2011). And indeed, there are few areas of study as fascinating as the psychology of terrorism. We are dealing with a fairly rare phenomenon. But to understand it requires mastery of far more than one discipline. Studying terrorism takes one through psychology, political science, criminology, history, sociology, etc. And it yields knowledge that is not just of value to one discipline, but in fact, to many. It is also a fascinating disciple, because here, perhaps more than anywhere else, the two perspectives on social knowledge come into play. And in fact, they come into play in peculiar fashion, not necessarily as opposites. By all accounts, it should be apparent that the psychology of terrorism should have one, very specific goal: to predict. One needs to know where the next terrorist attack will take place, who to look for, and what circumstances to change. With respect to the task of prediction, the experimentalists may appear to have serious advantages. Prediction, on the basis of general hypothesis, is, after all, their game. And it is certainly not the mainstay of the social constructionists. Their message that each context produces its unique dynamics would lead to a response of disappointment to all those that deal with terrorism either professionally or personally. Yet, when describing the field of the psychology of terrorism, it is really striking how much it is dominated by historical analysis, anthropological methods, and case research. One finds, for example, one of the most prominent scholars in the field,

190  Mark Dechesne and Charlotte de Roon Martha Crenshaw (1995), advocating that terrorism should be studied in context. The complexity of terrorism, the unique circumstances and dynamics of each case of terrorism, simply warrant no further inference than what can be observed, in retrospect, in a particular context. In retrospect, indeed, terrorism is a field that is dominated by historians and anthropologists that tend to stress the importance of historical description, not the formulation of general rules or hypotheses that would predict future terrorism. Moreover, within the psychology of terrorism, there is the myth that one needs to have interviewed, talked to, or observed from a close distance, a terrorist. This clearly fits with the idea that one should study terrorism in context, not in an artificial environment such as the laboratory. In order to study terrorism in a real way, one needs to listen to the political ideals of the terrorists, to their justifications, and to their life stories. As Max Taylor, an authority in the field, has put it, “One big problem with terrorism policy altogether is that it tends to interpret things from our perspective, based on what makes sense to us,” Taylor says. “That’s not really the issue: The issue is what makes sense to people on the ground” (DeAngelis 2009). The psychology of terrorism is of interest in a discussion on the ethics of belief, because there seems to be a large discrepancy between the quite stringent ethics of belief held by scholars, on the one hand, and the pragmatics of belief that is implicated by the threat imposed by terrorism, on the other. While many scholars forewarn against any generalizations of their case description, such generalizations are actually what we need, pragmatically, in order to do something about terrorism. To know where it will occur next, one needs to generalize on the basis of prior occasions. And the stories and experiences with terrorists and terrorist history are actually quite often used pragmatically, as a justification for a particular policy, as a way of making sense of current events, as a way to warn against mistakes, or as a way, in fact, to predict future behavior. Also, in a related field, the psychology of mass violence (Dechesne 2011), it is quite apparent that the major theories on the topic developed in an attempt to make sense specific incidents, and these specific incidents shaped to a considerable extent the theories. The theories are thus to be considered generalizations of specific events that are then used to generate hypotheses that are investigated in other contexts.

10.4  Using Stories to Navigate in Time and Space This pragmatic use of stories or incidents, as is evident from the case of the psychology of terrorism and of mass violence, brings us to something more fundamental. Something that suggests that social constructions, yes, exactly the ones that were brought up to deny experimentalists claims, are used in daily life, in the same way as the stories of terrorists are used by security analysts—that is, as a way to justify a particular course of action, as a way to make sense of current events, as a way to warn against mistakes, or as a way to predict the future.

Perspectives on Social Knowledge  191 We argue that these stories, or more generally, social constructions, help the individual to navigate in space and time. These social constructions may even be considered hypotheses, and we stress here the word “hypotheses,” of how the world, the social world in particular, functions. It is this pragmatics of belief, that could lead one to suggest that the social constructionists and the experimentalists are not that different. In fact, they rely on exactly the same cognitive structure, a structure of hypotheses in the form of if-then rules. Let us explain in greater detail. Consistent with lay epistemic theory (Kruglanski 1989), we argue that everyday understandings of laypeople are not that different from the understanding of scientists. And the process through which people both (1) come to their understanding of events and (2) let these understandings guide their thinking and actions, is not that different from the process of hypothesis generation and hypotheses validation that can be found among scientists. And this similarity, we argue, has to a large extent to do, not such much with a shared interest in coming up with the right set of beliefs, but rather with the most actionable sets of beliefs. These beliefs allow us to justify our actions and decisions, to make sense of current events, to warn against mistakes, or to predict the future. And in doing so, they have a pragmatic function. They enable action. Even if one adopts a radical social constructionists stance, one is often still forced to admit that there is a use to all these social constructions. A construction may be specific to a situation, but the use of constructions is universal. Cultural psychologist Harry Kempen (1996), a fervent constructionist, was very apt at explaining how stories affect our daily life. In one of his papers, he draws a person standing in front of a morass, suggesting that the person was contemplating how to get across. In considering how to get across, the person makes use of constructions, or what Kempen calls “as-if behaviors” (Kempen 1996: 721). Constructions that would help him to get a grip on the situation, to get the right tools, and to outline a course of action that would enable him to jump across the canal. The more abstract principle behind this image is that constructions, including stories, have a clear function in the sense they help the individual to orient and take action. Stories help the individual to navigate; navigate in space and navigate in time. In these constructions the person finds beacons that enable him or her to orient. If I am following the Frysian ritual of fydlejeppen, then I should use this stick to get across the water. If Tjibbe, that Frysian legend, was able to get across, then I should be able to follow in his footsteps, and do so too. Or in a different context, if the history of the IRA tells us to be aware of harsh government responses to terrorism, then we caution against excessively violent responses to terrorist attacks. Note that these beacons are used in exactly the same way as hypotheses are by scientists. Both the scientist and the layperson use constructions, constructs if you will, to derive hypotheses on the actual situation and the right course of action. Both use an if-then contingency structure to make sense of their lives. And although the content of the if-then rule may be different, and in fact they differ among scientists,

192  Mark Dechesne and Charlotte de Roon as much as they differ among laypeople, the operating system, the basic structure of the belief system and its operations, is the same, wherever you go (Anderson and Lebiere 1998). Once we accept the if-then nature of human (cognitive) constructions, it makes perfect sense to study humans as knowers, independent of specific social contexts, in a laboratory setting. The process of knowledge formation, irrespective of whether one is studying person perceptions, group identification, persuasion, or whatever related to the social world, is the same. It is about studying the hypotheses people generate and the hypotheses people validate in a particular domain. In encounters with others, individuals develop particular rules of thumb or rely on heuristics to perceive people and predict their future behavior. In deciding what to believe, what to be persuaded by, people rely on their already established knowledge. Again, knowledge formation in the social domain is all a matter of hypothesis generation and hypothesis validation, and understanding the environmental, cognitive, and motivational factors that impact on this process (see Kruglanski 1989). And this can be studied in an artificial environment such as the psychological laboratory.

10.5  The Nature and Importance of Evidence The discussion so far may suggest that the experimentalists are right. But it is too early for such a call. One can study the process through which people construct their understanding of the world and use these constructions to guide their actions. And although social constructionists would disagree with this statement, very strongly in fact, the actual practice pleads against them. The social constructionist literature is full of (implicit) assumptions of how stories have a direct, and universal impact on action. And these assumptions all implicate the idea of an if-then universal structure of knowledge formation. The stories provide the content, not the structure of thought. But going beyond the question of whether the structure of knowledge formation is the same for the everyday knower, the layperson and the scientist, certainly not all is won for the experimentalists. One also needs to consider the way evidence is treated, and above all, what is treated as evidence. And now, we enter a whole new discussion. A discussion, in fact, with an outcome that may not particularly please the experimentalists, while it may please the social constructionists. What we treat as objective evidence for general hypotheses in social psychology (and we should again emphasize here that social psychology is meant to be our most advanced system to understand and know the social world), may in fact be very limited in scope. While we are dealing with a universal knower, the specific practices of social psychology may have led us away rather than brought us towards understanding the social knower. And this is primarily because of the treatment of evidence in social psychology. Two recent significant events in social psychology suggest this is the case.

Perspectives on Social Knowledge  193

10.5.1  We Have to Break Up In 2009, one of the great inspirational figures in the field, Robert Cialdini, had just published an article with the title “We have to break up” (Cialdini 2009). In it, he described his disaffection with the current state of affairs in the field of social psychology. The reader should know that Cialdini is one of the foremost scholars in social psychology. And this is particularly so because he has managed to engage not only the academic community, but also people in business and society at large. His ideas about social influence have not only spurred interest within the university walls, but have had a significant impact on the way companies go about advertising their products, and have encouraged more than one or two random people at a bookstore to try their luck in persuading others to sell their products, or convictions. Now this Cialdini writes an article, “We have to break up.” He wants to break up because, in essence, social psychology is no longer interesting. Or to put it somewhat differently, social psychology has turned inward, up to the point that it is has lost its appeal to the outside world of non-psychologists. It has become irrelevant to anyone outside of the field. And there are multiple reasons for this. But foremost, it is the emphasis on methodological rigor that has put many down, according to Cialdini at least. It is the emphasis, in other words, on a very restricted interpretation of what counts as evidence for social psychological propositions. In fact, one might say, in the effort to come up with solid, reliable evidence, social psychology has gotten out of touch with reality. Note here that Cialdini’s critique that science has become too restricted, too much focused on methodological rigor, and thereby has lost its touch with reality, does not pertain to the basic principles of doing science. Rather, it is about the type of evidence that is considered of value to corroborate or refute a scientific hypothesis. And to be sure, the assumption is not so much that each social situation implicates its own set of hypotheses, as the social constructionists would argue. But then again, if academics tend to focus on only particular types of evidence, and in the process of doings so, ignore reality, one can see how a peculiar type of knowledge is constructed. Experiments become social constructions, contrived by a select group of social psychological practitioners, far more than test areas to document universal processes.

10.5.2  The Stapel Case A far more shocking case, that directly pertains to the peculiar treatment of evidence, involves recent events in the Netherlands. One of the country’s proclaimed superstars of social psychology and of academia in general, for that matter, was forced to admit that he had reported data from experiments that actually were never administered. He had made them up. And this for the most productive social psychologists around, in the Netherlands and worldwide. His entire entourage of colleagues, graduate students, and postdocs, were not so much directly involved, but got their data through

194  Mark Dechesne and Charlotte de Roon his memory stick, which he controlled. Thus, a considerable number of a generation of young Dutch social psychologists have written their dissertations and their first articles on the basis of data given to them by Stapel. It has all become useless and without any value now. The official report (Levelt committee 2012) on the circumstances and the research culture that has made this grotesque fraudulent endeavor possible clearly implies that psychology, and social psychology in particular, is seriously damaged. And the damage primarily concerns reputation damage. The Stapel case provides yet another example of how the treatment of evidence, and yes, the complete illegitimate treatment of evidence, has created a socially construed sense of reality among the Dutch social psychological community. Exactly the ones who have been so resisting of social constructionist ideas. And it is only gradually dawning that the world of academic successes and quick publications, which has become the hallmark of Dutch social psychology, is no longer within reach. It has proved to be a bubble. A bubble that the constructionists would certainly consider as another illustration of their central argument. But again, it is not so much the universal operating system of hypotheses and hypotheses testing that is at stake, or under review, but rather the process through which people come to accept particular pieces of information as evidence for their hypotheses. The Stapel affair may elicit fundamental questions and debate regarding the practice of social psychology: about how social psychology is currently practiced and about how it should be practiced. These fundamental questions are certainly worth asking, particularly in a time like this. It is of importance to consider what type of evidence is necessary to substantiate scientific claims regarding our social world. And it is of importance to consider whether our thoughts and beliefs are operated through a universal mechanism, or consists of a great variety of situation dependent mechanisms. In a sense, then, if we understand correctly, it is of importance to consider the ethics of belief in social psychology.

10.6  The Ethics of Belief in Social Psychology Much of our daily life revolves around others and our dealings with others. Knowledge about the social may therefore be considered one our biggest prizes. But such knowledge doesn’t come easily. And in fact, after another fight with a loved one, a failed attempt to get a promotion, or another encounter that ends in estrangement, we realize how difficult it is to really understand others and ourselves. Social psychology is the field that is in principle best equipped to come up with that type of knowledge. But in practice, social psychologists too are divided on how to come up with the right types of beliefs. As we have tried to explain, social psychology comes in two. There are two perspectives on social knowledge: an experimentalist perspective, which focuses on universal hypothesis and experimental testing, and a social constructionist perspective, which focuses on the situation specific construction of cultural meanings.

Perspectives on Social Knowledge  195 In actuality, the two may not be that different. The stories that people develop and pass on in their everyday lives, may serve purposes that are very similar to the theories and hypotheses that scientists use. Just as scientists use their theories, people use their stories to justify a particular course of action, as a way to make sense of current events, as a way to warn against mistakes, or as a way to predict the future. The stories allow the individual to navigate through space and time, as the stories provide beacons to cling on to, or alter course when obstacles are encountered. The beacons, as we have explained, implicate the same if-then structure, that also characterizes hypotheses. It is not so much the structure of beliefs, then, that causes the problem in coming up with a unified perspective on social knowledge. In fact, the notion of an if-then production system may do quite well as a unified architecture of cognition (Anderson 1983). The contentious issue is the process through which people obtain evidence and judge whether a piece of information is evidence to substantiate a hypothesis or belief. The experiment may not be the best at our disposal. As experimentation has become the hallmark of mainstream social psychology, experts in the field, Robert Cialdini prominent among them, have increasingly questioned whether excessive methodological rigor brings us any closer to reality. And after the Stapel affair, who can be sure that the data was really obtained? In order to come to a social psychological ethics of belief, it is of importance to consider how pieces of information come to be considered evidence for a particular hypothesis. And in doing so, we should recognize that there are actually multiple viewpoints of what is most relevant. In what we have just written, one may actually recognize Arie Kruglanski’s Unimodel (see Chapter 9). With its emphasis on a universal production mechanism, i.e., if-then rules, but also with the specification of judgmental parameters, it may formalize what has been proposed by others to resolve the tension between the experimentalists and the social constructionists. William McGuire (1973) for example has advocated a “perspectivist” stance, which encourages researchers to take multiple perspectives and to engage in dialogue with opponents. A Unimodel account of the ethics of belief in social psychology goes beyond that. It goes beyond the intuitive notion of perspectivism by specifying the parameters that are relevant for engaging in dialogue. If a particular conclusion has been reached under heightened time pressure, or diminished processing capacity, we should scrutinize the validity of the conclusion under less time constraints or under enhanced processing capacity. If a particular conclusion is based a specific source deemed to have epistemic authority, we should scrutinize the validity of the conclusion in the absence of that epistemic authority. If a particular conclusion has been reached against the backdrop of directional or non-directional motivation, we should scrutinize the validity of this conclusion at the opposite side of these motivational dimensions. The Unimodel then may not only serve as a framework for understanding how laypeople make sense of their lives. It could also be used as a framework to consider how social psychologists should make sense of the lives of others. It provides a framework

196  Mark Dechesne and Charlotte de Roon for systemic perspectivism. And that is quite a step indeed. It has often been argued that laypeople are intuitive scientists. But here is an interesting reversal. Laypeople are not intuitive scientists, but rather, social psychologists are to be considered sophisticated laypeople. Laypeople, because social psychologists make use of the same logic to understand and predict the social world. But also sophisticated, because they are aware of the multiple perspectives one can have on social knowledge. And sophisticated because they are aware the social psychologist has the obligation to fully consider all pieces of evidence. The ethical social psychology, or “ethical social psychological theory,” then, is the one that reaches out to, and engages with multiple audiences. That considers the evidence for its hypothesis at different parameter values. It seeks to form coalitions between the different perspectives, and in doing so, it seeks knowledge alliances. Alliances within the academic community, but also alliances with the outside world: the outside world of fellow sophisticated knowers of the social world, such as politicians, and policy makers. But also to the people it seeks to study. Because the layperson, too, is likely to have ideas that may hint towards the drivers of his or her own behavior. Seen in this light, then, perspectives on social knowledge may not be as contentious after all. Once framed in a particular way, they all contribute to a legitimate and ethical understanding of one of the important, and most fascinating areas of study: the study of our social world.

References Anderson, J. (1983). The Architecture of Cognition. Cambridge, MA: Harvard University Press. Anderson, J., and Lebiere, C. (1998). The Atomic Components of Thought. Mahwah, NJ: Lawrence Erlbaum Associates Publishers. Burr, V. (2003). Social Constructionism, 2nd edn. London: Routledge. Cialdini, R. B. (2009). “We Have to Break Up,” Perspectives On Psychological Science 4(1): 5–6. Crenshaw, M. (1995). Terrorism in Context. University Park, PA: PSU Press. DeAngelis, T. (2009). “Understanding Terrorism,” APA Monitor on Psychology, URL = . Dechesne, M. (2011). The Psychology of Mass Violence:  A  Unified Field Theory. University of Leiden—Campus The Hague. Gergen, K. J. (1973). “Social Psychology as History,” Journal Of Personality And Social Psychology 26(2): 309–20. Gergen, K. J., and Gergen, M. M. (2010). “Scanning the Landscape of Narrative Inquiry,” Social And Personality Psychology Compass 4(9): 728–35. Kempen, H. G. (1996). “Mind as Body Moving in Space:  Bringing the Body Back into Self-Psychology,” Theory and Psychology 6(4): 715–31. Kerlinger, F. N. (1973). Foundations of Behavioral Research, 2nd edn. Holt, Rinehart and Winston. Kesebir, P., and Pyszczynski, T. (2012). “The Role of Death in Life:  Existential Aspects of Human Motivation,” in R. M. Ryan (Ed.), The Oxford Handbook of Human Motivation. New York: Oxford University Press, 43–64.

Perspectives on Social Knowledge  197 Koro-Ljungberg, M. (2008). “A Social Constructionist Framing of the Research Interview,” in J. Holstein and J. Gubrium (Eds.), Handbook of Constructionist Research. New York: Guilford Press, 429–44. Kruglanski, A. W. (1989). Lay Epistemics and Human Knowledge: Cognitive and Motivational Bases. New York: Plenum Press. Kruglanski, A. W., and Thompson, E. P. (1999). “Persuasion by a Single Route: A View from the Unimodel,” Psychological Inquiry 10(2): 83–109. Levelt Committee. (2012). Flawed Science: The Fraudulent Research Practices of Social Psychologist Diederik Stapel, URL = . Macrae, C., and Quadflieg, S. (2010). “Perceiving People,” in S. T. Fiske, D. T. Gilbert, and G. Lindzey (Eds.), Handbook of Social Psychology, Vol. 1, 5th edn. Hoboken, NJ: John Wiley and Sons Inc., 428–63. McGuire, W. J. (1973). “The Yin and Yang of Progress in Social Psychology: Seven Koan,” Journal Of Personality And Social Psychology 26(3): 446–56. Milgram, S. (1963). “Behavioral Study of Obedience,” Journal of Abnormal and Social Psychology 67: 371–78. Popper, K. (1994). The Myth of the Framework:  In Defence of Science and Rationality, N. A. Notturno (Ed.). London: Routledge. Rieber, R., and Robinson, D. (Eds.) (2001). Wilhelm Wundt in History: The Making of a Scientific Discipline. New York: Kluwer Academic. Schmid, A. (Ed.) (2011). The Routledge Handbook of Terrorism Research. London: Routledge. Ulrich, W. (2006). “Rethinking Critically Reflective Research Practice: Beyond Popper’s Critical Rationalism,” Journal of Research Practice 2(2), URL = . Wilson, T. D., Aronson, E., and Carlsmith, K. (2010). “The Art of Laboratory Experimentation,” in S. T. Fiske, D. T. Gilbert, and G. Lindzey (Eds.), Handbook of Social Psychology, Vol. 1, 5th ed. Hoboken, NJ: John Wiley and Sons Inc., 51–81. Wundt, W. (1916). Elements of Folk Psychology:  Outlines of a Psychological History of the Development of Mankind, Edward L. Schaub (Trans.). London: Allan and Unwin.

11 Contagion, Community, and Virtue in Hume’s Epistemology Rico Vitz

The whole is a riddle, an enigma, an inexplicable mystery. Doubt, uncertainty, suspence of judgment appear the only result of our most accurate scrutiny, concerning this subject. But such is the frailty of human reason, and such the irresistible contagion of opinion, that even this deliberate doubt could scarcely be upheld. . . David Hume, The Natural History of Religion

Hume regularly employs a particular, two-fold strategy in analyzing the way that people acquire beliefs. He begins, in the negative phase, by showing that certain kinds of beliefs are not founded in reason and goes on, in the positive phase, to explain the way that they are grounded in human nature. Probably the most famous use of this strategy is his analysis of people’s beliefs concerning matters of fact. After arguing that such beliefs are not founded in reason—demonstrative or probable—he claims that they are grounded in human nature, arising from custom, or habit. Perhaps Hume’s most ambitious use of this strategy is evident in the distinctively irreligious aspect of his philosophical program. In the negative phase, developed most notably in the Dialogues concerning Natural Religion and his essay on miracles, he argues that traditional religious beliefs are founded neither in compelling arguments from natural religion nor in successful arguments from revealed religion. In the positive phase, developed in the Natural History of Religion, he argues that the belief in invisible intelligent power originated among our ancestors from a secondary principle that was triggered by difficult circumstances and powerful emotions. His irreligious use of this strategy is particularly interesting. After completing his systematic analysis concerning the acquisition of religious beliefs, he concludes by noting that human reason is so frail that even deliberate doubt concerning such beliefs can scarcely be maintained in light of “the irresistible contagion of opinion”

Contagion, Community, and Virtue  199 (NHR 15.13).1 This concluding comment raises a number of intriguing questions about Hume’s epistemology. How are beliefs supposed to be “contagious”? Why is the contagious nature of beliefs so powerful? How does one resist this contagion? Under what conditions is it virtuous to do so? The purpose of this essay is to answer these questions. Providing answers to such questions is particularly important for, at least, two reasons—one historical and one more broadly philosophical. First, with respect to the history of philosophy, both our interpretation and our evaluation of Hume’s philosophical system is, and will remain, incomplete until proper scholarly care has been taken to explicate his positive epistemic views in the kind of detail that his negative epistemic views have been explicated.2 Second, with respect to philosophy in general, since Hume’s philosophical system is an important forerunner to various views in ethics and epistemology, understanding his answers to the kinds of questions raised above has the potential to offer fruitful insights to contemporary debates in the ethics of belief. In light of the twofold significance of answering the questions raised above, my aim in this chapter is, similarly, twofold. I hope to provide an example of how (1) careful analysis in the history of philosophy can (2) elucidate contemporary debates about philosophical issues. My analysis of Hume’s account of the contagion of belief will unfold in three parts. In the first section of the chapter, I will offer a summary of Hume’s account of the nature of beliefs concerning matters of fact, since it is this type of beliefs that are at issue in his comments concerning the contagion of opinion. In the second section, I will elucidate his account of the “contagion of opinion” itself. In so doing, I will answer the first three questions raised above, explaining how beliefs are contagious, why the contagious nature of beliefs is so powerful, and how people can resist the contagion. In the third

  1  In this essay, I use the following abbreviations and conventions to refer to Hume’s works: • References to A Treatise of Human Nature are to Hume (2000) and are abbreviated as “T,” followed by the book, part, section, and paragraph numbers of the relevant passage. Each reference also uses the abbreviation “SBN” to note the corresponding page(s) in Hume (1978). • References to An Enquiry concerning Human Understanding are to Hume (1999a) and are abbreviated as “EHU,” followed by the section and paragraph numbers of the relevant passage. Each reference also uses the abbreviation “SBN” to note the corresponding page(s) in Hume (1975). • References to An Enquiry concerning the Principles of Morals are to Hume (1999b) and are abbreviated as “EPM,” followed by the section and paragraph numbers of the relevant passage. Each reference also uses the abbreviation “SBN” to note the corresponding page(s) in Hume (1975). • References to Essays: Moral, Political, and Literary are to Hume (1987) and are abbreviated as “EMPL,” followed by the page number of the relevant passage. • References to The Natural History of Religion are to Hume (2007) and are abbreviated as “NHR,” followed by the section and paragraph numbers of the relevant passage.   2  To illustrate this point, let me highlight an example from one aspect of his philosophical system: his account of religious belief. The overwhelming majority of the scholarly work on Hume’s analysis of religious belief focuses on the negative aspects of Hume’s program, as described in the Dialogues (1980), the essay on miracles (EHU 10), the essay on a future state (EHU 11), and so forth. In fact, if (i) my 2008 survey of the scholarly literature of works focusing on “Hume” and “religion” was correct and (ii) the status of this literature has not changed dramatically over the past five years, the proportion of published articles focusing on the negative aspects of Hume’s irreligious program is over 95%.

200 Rico Vitz section, I will elucidate the significance of Hume’s account of contagious beliefs for his account of doxastic virtues. In so doing, I will answer the final question, explaining the conditions under which it is virtuous to resist contagious beliefs.

11.1  The Nature of Belief Hume identifies two kinds of objects of enquiry: relations of ideas and matters of fact (see, e.g., EHU 4.1; SBN 25). When explaining the nature of belief both in the Treatise and in the first Enquiry, he focuses principally on the latter. In the Treatise, he sets up his discussion of belief by noting that the issue in question is beliefs concerning matters of fact—particularly, beliefs concerning causation. He says, [T]‌he belief or assent, which always attends the memory and senses, is nothing but the vivacity of those perceptions they present; and that this alone distinguishes them from the imagination. To believe is in this case to feel an immediate impression of the senses, or a repetition of that impression in the memory. It is merely the force and liveliness of the perception, which constitutes the first act of the judgment, and lays the foundation of that reasoning, which we build upon it, when we trace the relation of cause and effect. (T 1.3.5.6; SBN 86, emphasis mine; cf. T 1.3.5.1; SBN 84)

Immediately after his famous descriptions of belief as “a lively idea related to or associated with a present impression” and as “a lively idea produc’d by a relation to a present impression” (T 1.3.7.5–6; SBN 96–7), he notes that these descriptions are of beliefs concerning matters of fact (T 1.3.7.7; SBN 628).3 He, then, announces that the purpose of the subsequent section (T 1.3.8) is to explain the causes of the kind of beliefs described in T 1.3.7—namely, beliefs concerning matters of fact (cf. T 1.3.8.1; SBN 98). In the ­following section (T 1.3.9), he confirms, yet again, that his interest is beliefs con­ cerning matters of fact—specifically, “judgments concerning cause and effect” (see, e.g., T 1.3.9.2–3; SBN 107–8; cf. T 1.3.9.10; SBN 112). In the first Enquiry, he leads into his explanation of belief by discussing causation. Towards the end of the first part of EHU 5, he says, “All belief of matter of fact or real existence is derived merely from some object, present to the memory or senses, and a customary conjunction between that and some other object” (EHU 5.8; SBN 46, emphasis mine).4 He prefaces his subsequent descriptions of belief as “nothing more but a lively, forcible, firm, steady conception of an object” and as consisting “not in the peculiar nature or order of ideas, but in the manner of their conception, and in their feeling to the mind” (EHU 5.12; SBN 49; cf. EHU 5.13; SBN 50), by noting that he has in mind belief concerning matters of fact (EHU 5.9; SBN 47). Thus, in his seminal discussions about the nature of belief, both in the Treatise and in the first Enquiry, Hume is principally—in fact, almost exclusively—concerned with    

  See also T App.2–9; SBN 623–7, as well as T 1.3.5.4, 1.3.10.10–2; SBN 627–8, 630–2.   Cf. T 1.3.7.6, 1.3.8.1, 7, 11, 15, 1.3.9.13; SBN 96–8, 101, 103, 105, 114; ATHN 21; SBN 653–4.

3

4

Contagion, Community, and Virtue  201 attempting to elucidate the nature of beliefs about matters of fact. Regarding such beliefs, however, Hume offers what might appear to be a number of different descriptions. For instance, he identifies a belief about a matter of fact as “something felt by the mind, which distinguishes the ideas of the judgment from the fictions of the imagination” (EHU 5.12; SBN 49; cf. EHU 5.8, SBN 46), “nothing but a more forcible and vivid conception of an idea” (T 1.3.9.2; SBN 107, emphasis mine; cf. T 1.3.13.19, App 21; SBN 153, 654), “a lively idea related to a present impression” (T 1.3.8.1; SBN 98, emphasis mine), “a lively idea related to or associated with a present impression” (T 1.3.7.6; SBN 96, emphasis mine; cf. T 1.3.6.15; SBN 93), “a lively idea produc’d by a relation to a present impression” (T 1.3.7.6; SBN 97, emphasis mine; cf. T 1.3.8.11; SBN 103), “a strong and lively idea deriv’d from a present impression related to it” (T 1.3.8.15; SBN 105, emphasis mine), and “an act of the mind arising from custom” (T 1.3.9.13; SBN 114, emphasis mine; cf. EHU 5.11–3; SBN 48–50). In his discussion of the causes of beliefs about matters of fact, he argues, It is certain we must have an idea of every matter of fact which we believe. It is certain that this idea arises only from a relation to a present impression. It is certain that the belief superadds nothing to the idea, but only changes our manner of conceiving it, and renders it more strong and lively. The present conclusion concerning the influence of relation is the immediate consequence of all these steps; and every step appears to me sure and infallible. There enters nothing into this operation of the mind but a present impression, a lively idea, and a relation or association in the fancy betwixt the impression and idea; so that there can be no suspicion of mistake. (T 1.3.8.7; SBN 101, emphasis mine)

Therefore, according to Hume, a belief about a matter of fact is an act of mind5 in which a present impression triggers a psychological mechanism, such as the principle of custom, to enliven an idea to which the impression is related. In other words, on Hume’s account, a person believes a proposition about a matter of fact if he or she has a mental representation (e.g., an idea), which can be signified by a proposition, and his or her mental representation is enlivened by a psychological mechanism, like the principle of custom, acting on a present impression to which the idea is related.6 I have only claimed, so far, to identify criteria that are sufficient for belief, but are these criteria not also necessary? In the Treatise, Hume says that people “believe Caesar   5  See also T 1.3.7.5n20, 1.3.7.7, 1.3.9.13, 1.4.1.8, App 9; SBN 96n1, 628–9, 114, 183–4, 627.   6  Cf. Johnson (1995: 173–4), Price (1969: 189–220), Price (1967: 43ff), and Russell (1921: 245–6). Three points of clarification, though not essential to my main line of argument here, are helpful for an accurate understanding of Hume’s account of belief concerning matters of fact. First, on Hume’s account, a belief is an enlivened idea—i.e., an idea that is “strong,” “lively,” “forceful,” “vivid,” and so forth. To say that an idea is enlivened, however, does not mean merely that the vivacity of the idea approaches that of an impression. For instance, no idea will become a belief merely by an increase of vivacity via poetical enthusiasm (T 1.3.10.10; SBN 630). According to Hume, the terms “strong,” “lively,” “forceful,” “vivid,” and so forth are merely inadequate ways of trying to capture that special feeling which is caused in a particular way and which distinguishes beliefs from mere ideas of the imagination—no matter how vivacious (T 1.3.7.7, 1.3.10.10; SBN 629, 630–1).   Second, Hume claims that an enlivened idea is related to a present impression in one of three ways. (Strictly speaking, this claim could be a bit misleading since the relation in question is not about ideas and

202 Rico Vitz was killed in the senate-house on the ides of March” (T 1.3.4.2; SBN 83). He also claims that people have beliefs about principles such as “like objects, plac’d in like circumstances, will always produce like effects” (T 1.3.8.14; SBN 105). He does not seem to suggest, however, that people cease to have these beliefs the moment they no longer have lively mental representations that correspond to these propositions. Moreover, he claims that people have a number of beliefs to which they have been accustomed from their infancy (T 1.3.9.17; SBN 116). Beliefs such as these are acquired dispositions— habits, as Hume suggests—that have taken “deep root” in people and are “impossible” for them to eradicate (T 1.3.9.17; SBN 116). Hence, on Hume’s account, not all beliefs are mental acts; some beliefs are mental habits, or dispositions. In more contemporary terminology, we might say that on Hume’s account, not every belief is occurrent; some are dispositional.7 Hume clarifies his account of habitual, or dispositional, beliefs concerning matters of fact in T 1.3.9. He notes that some of people’s beliefs are habits that they have acquired from infancy (T 1.3.9.17; SBN 116)  and suggests that frequently repeated occurrent beliefs (e.g., acts of judgment) cause habitual beliefs (cf. T 1.3.9.19; SBN 117). For instance, to say that a person has a habitual, or dispositional, belief that Caesar died on the ides of March, is to say that the person is disposed such that when he or she has an idea of Caesar dying on the ides of March, he or she will enter the mental state that is constitutive of the act of believing. Having clarified the context of T 1.3.9, we can rightly recognize Hume’s comment at T 1.3.9.16 (SBN 115–6) as a summary of his criteria for habitual, or dispositional, beliefs. He says that if in all past experience, we have found two objects to have been always conjoined together, it is evident, that upon the appearance of one of these objects in an impression, we must, from impressions considered as acts of the mind, but about ideas and impressions considered as objects of the mind. To account for this distinction, perhaps it would be better if we employed some contemporary terminology and read Hume as saying not that the impressions bear these relations to the ideas; but that the contents of the impression bear these relations to the contents of the idea.) First, the contents of an idea may resemble the contents of the present impression that enlivens it (cf. T 1.3.8.3–4; SBN 99–100; EHU 5.15–16; SBN 51–2). For instance, the contents of the present impression one gets from seeing a picture of one’s friend resembles the contents of the idea of one’s friend. Second, the contents of an idea may bear the relationship of contiguity to the contents of the present impression that enlivens it (cf. T 1.3.8.5; SBN 100; EHU 5.17; SBN 52). For example, looking out the window on her flight home to New York, a U.S. citizen who has been in Europe on business for a week sees the Statue of Liberty and the New York skyline. The contents of the impressions of the statue and the skyline bear what Hume calls the relationship of contiguity to the contents of the idea of her family members, who live in the city. Finally, the contents of an idea may have the relationship of causation to the contents of the present impression that enlivens it (cf. T 1.3.8.6; SBN 100–1; EHU 5.18–9; SBN 53). Consider, for instance, a book that was written by one’s colleague. Since the colleague is the cause of the existence of the book, the contents of one’s impression of the book bears what Hume calls the relationship of causation to the contents of the idea of one’s colleague. Third, Michael Gorman suggests that Hume does not include a causal notion in his definition of belief (1993: 98–9), but Gorman’s claim is rather misleading. According to Hume’s definition, a belief is “a lively idea related to or associated with a present impression” (T 1.3.7.6; SBN 96). This “association” to which Hume refers in his definition of beliefs about matters of fact is causal, as Hume implies at EHU 5.8 (SBN 46–7).   7  Some of Hume’s commentators mistakenly suggest that Hume’s account does not allow for dispositional beliefs—see, e.g., Pears (1991: 50).

Contagion, Community, and Virtue  203 custom, make an easy transition to the idea of that object, which usually attends it; and by means of the present impression and easy transition must conceive that idea in a stronger and more lively manner than we do any loose floating image of the fancy.

This passage is somewhat ambiguous. Consider it in light of the following case: a person has a frequent and uniform experience of thunder always following lightning. We can read the passage in two ways. First, we can read Hume as suggesting that the person has a disposition to form the occurrent belief thunder is coming on the occasion of seeing lightning, but does not have the dispositional belief thunder is coming. Second, we can read it as a passage in which Hume suggests that the person has the dispositional belief thunder always follows lightning. These readings are compatible. Thus, we can read the passage as one in which Hume is explaining both how people have dispositions to believe and how people have dispositional beliefs.8 On Hume’s account, each of these two things is true. First, a person has a disposition to believe that a proposition is true if he or she is naturally disposed9 such that if he or she has an experience of the proposition being true, he or she will form the occurrent belief that the proposition is true. Second, the person has a dispositional belief that a proposition is true if he or she is habitually disposed such that when he or she has an idea, which can be represented by a proposition, and his or her idea will be enlivened by the principle of custom acting on a present impression to which the idea is related, as in the case of beliefs about matters of fact. So that the significance of the distinction is neither lost nor obscured, let me elucidate the difference between dispositions to believe and dispositional beliefs in a bit more detail. Consider, again, the previous example of the relationship between thunder and lightning. Suppose a person is naturally disposed such that if she has an experience that corresponds to the proposition thunder follows lightning, then she will form the occurrent belief that thunder follows lightning. Thus, we are supposing that she has the disposition to believe that thunder follows lightning, but this disposition alone is not a dispositional belief. Once she has the experience, she will form the occurrent, but not necessarily the dispositional, belief that thunder follows lightning. Only if she forms the occurrent belief a sufficient number of times10 will she form a dispositional belief that thunder follows lightning. Thus, the process of forming a dispositional belief has four stages. First, a person has a natural disposition to believe (occurrently), under conditions, that a certain proposition is true. Second, the person has an experience that satisfies the conditions. Third, the person has the occurrent belief that the proposition is true.   8  Cf. Audi (1994).   9  For the sake of clarity: As I am using the term, to say that a person is “naturally” disposed to A is equivalent to saying, colloquially, that he or she is “wired up” to A.   10  Note that the “sufficient number” of times could be as low as one since some experiences might be so striking that they form in a person a disposition after merely one occurrence. For instance, a young child might need only one frightening experience of a particular vicious dog, Spike, to form the dispositional belief that Spike is mean.

204 Rico Vitz Fourth, the person experiences the second and third stages a sufficient number of times such that he or she has an acquired dispositional belief that the proposition is true.11 Before I move on, let me summarize, briefly, Hume’s account of the nature of belief. According to Hume, a person believes a proposition (about a matter of fact), if either (1) he or she has an idea, which can be represented by a proposition, and his or her idea is enlivened by a psychological mechanism, like the principle of custom, acting on a present impression to which the idea is related, or (2) he or she is habitually disposed such that when he or she has an idea, which can be represented by a proposition, his or her idea will be enlivened by a psychological mechanism, like the principle of custom, acting on a present impression to which the idea is related.

11.2  The “Contagion of Opinion” Having summarized Hume’s account of the nature of beliefs concerning matters of fact, I will now proceed to explain his conception of contagious beliefs. I will begin by clarifying his account of contagion, in general. I will, then, explain both the way that people acquire contagious beliefs and the nature of contagion’s power.

11.2.1  Contagion in General Hume’s more widely discussed references to “contagion,” and its cognates, concern the passions. In the Treatise, for instance, he says, “The passions are so contagious, that they pass with the greatest facility from one person to another, and produce correspondent movements in all human breasts” (T 3.3.3.5; SBN 605). Likewise, in the second Enquiry, he discusses people’s ability to “catch” sentiments “by a contagion or natural sympathy”

  11  Any dispositional account of belief that fails to distinguish between dispositions to believe and dispositional beliefs runs the risk of having deeply counterintuitive implications. Here is why. Take a generic, dispositional account of belief, DB, that is committed to the following two points: first, it fails to distinguish between dispositions to believe and dispositional beliefs; second, according to DB every occurrent belief is merely the manifestation of a dispositional belief. With these points in mind, consider the following fact, healthy people are disposed from birth to form occurrent beliefs about basic logical truths. Therefore, according to DB, it seems that such people have beliefs from birth regarding basic logical truths, like DeMorgan’s Laws. It seems rather implausible, however, that when my son was an infant, he believed that the negation of a disjunction is logically equivalent to the conjunction of the negation of each of the disjunction’s disjuncts. Thus, endorsing a dispositional account of belief like DB comes with a high price—namely, attributing beliefs to people who, clearly, seem not to have them. On my reading, Hume’s account of belief avoids this problem; however, other dispositional readings of Hume seem to attribute to him the problematic view of DB, or something very much like it—see, e.g., Loeb (2005: 70–1, 75).

Contagion, Community, and Virtue  205 (EPM 7.2; SBN 251cf. EPM 7.21; SBN 257–8).12 He makes similar comments elsewhere. In his essay on national characters, for example, he says, The propensity to company and society is strong in all rational creatures; and the same disposition, which gives us this propensity, makes us enter deeply into each other’s sentiments, and causes like passions and inclinations to run, as it were, by contagion, through the whole club or knot of companions. . . . If we run over the globe, or revolve the annals of history, we shall discover every where signs of a sympathy or contagion of manners . . . (EMPL 200–4)13

Thus, in general, Hume uses the term “contagion,” and its cognates, to refer to the process by which people enter into the sentiments of others. He does not restrict, however, the product of this process to people’s sentiments. He also claims that people’s opinions, or beliefs, can be contagious. In the Treatise, for instance, he notes that by the mechanism of sympathy people “enter . . . into the opinions and affections of others” (T 2.1.11.7) and suggests that [n]‌o quality of human nature is more remarkable, both in itself and in its consequences, than that propensity we have to sympathize with others, and to receive by communication their inclinations and sentiments, however different from, or even contrary to, our own. This is not only conspicuous in children, who implicitly embrace every opinion proposed to them; but also in men of the greatest judgment and understanding, who find it very difficult to follow their own reason or inclination, in opposition to that of their friends and daily companions. (T 2.1.11.2; SBN 316)

Thus, on Hume’s account, beliefs can be acquired from others in much the same way that sentiments can—namely, as the product of the mechanism of sympathy and the process of contagion. In this respect, the operation of sympathy is strikingly similar to the operation of the understanding, insofar as each functions to enliven ideas (cf. T 2.1.11.3; SBN 317). In fact, Hume claims that the operation of sympathy is “exactly correspondent to” the operation of the understanding (T 2.1.11.8; SBN 320).

11.2.2  The Contagious Acquisition of Beliefs So far, I have highlighted textual evidence showing that Hume regards beliefs as contagious. Having established that point, I may now proceed to explain, in some detail, how contagious beliefs are acquired, on Hume’s account. For ease of presentation, let

  12  On Hume’s account, “sympathy” itself is “nothing but the conversion of an idea into an impression by force of the imagination” (T 2.2.6.8; SBN 427; cf. T 2.1.11.7–8; SBN 319–20). There has been some debate among Hume scholars regarding whether Hume changes his view about the sympathy from the Treatise to the second Enquiry. In this essay, I am adopting the view for which I and others have argued elsewhere—namely, that Hume’s position is substantially unchanged from his earlier to his later work. For defenses of this view, see Altmann (1980), Abramson (2001), Debes (2007a), Debes (2007b), Vitz (forthcoming), and Vitz (2004); cf. Vitz (2002).   13  See also “Of the Rise and Progress of the Arts and Sciences” (EMPL 111–12) and “Of the Liberty of the Press” (EMPL 604), where Hume makes similar and related comments on the nature of contagion.

206 Rico Vitz me simply elucidate three types of cases that emerge from Hume’s writings as plausibly most common. Sympathy. The first type of case focuses on the operation of sympathy. Consider, for example, a young child who is traveling with her father through a large metropolitan city. As they enter the bus station in the middle of the city after dark, the child is perfectly content playing in her surroundings. A woman nearby, however, who is an inexperienced traveler from a rural area is rather anxious in such unfamiliar circumstances. As a result of her anxiety, she begins to fear a group of young men loitering nearby. Consequently, she begins to believe that travelers, like herself and the father and his child, are in danger. Looking up from her play and studying the worried woman’s face, the child begins to sympathize with the women’s unspoken fears. In so doing, she acquires the woman’s belief that the travelers are in danger, despite the fact that unlike the woman, she neither notices nor fears the young men. Thus, the child comes to believe that she is in danger, though from what, she does not know. As this type of case suggests, the essential and most fundamental cause of contagious belief acquisition is the mechanism of sympathy. This need not be, however, the only mechanism that operates in such cases of contagion. Sympathy and Custom. The second type of case involves an operation of sympathy and an operation of custom. According to Hume, “Custom has two original effects upon the mind.” The first is to bestow “a facility in the performance of any action, or the conception of any object” (T 2.3.5.1; SBN 422). This facility is achieved, in incremental degrees, by repetitive experience (T 2.3.5.3; SBN 423). The second effect, and a consequence of the first, is to bestow “a tendency or inclination towards” the performance of the action or the conception of the object (T 2.3.5.1; SBN 422; emphasis mine). Once custom has facilitated the conception of an idea and, consequently, bestowed an inclination towards it, sympathy can enliven the idea, thereby causing a person to acquire a new belief by means of contagion. For instance, a young woman who moves from Belle Vernon, Pennsylvania to Pasadena, California, to begin her undergraduate studies at Caltech, may be exposed to the idea that the Los Angeles Lakers are a better sports franchise than the San Antonio Spurs. Even if she has absolutely no interest in sports or in the views of her peers on such issues, she may, by repetitive experience, begin to “kindle in the common blaze” (cf. EPM 9.9; SBN 275) and, over time, to have some inclination towards the idea that the Lakers are a better sports franchise than the Spurs. Consequently, sympathy may enliven the idea, thereby causing her to acquire the belief.14 As this type of case suggests, a second way in which a person can acquire a belief via contagion is by the operations both of sympathy and of custom.15   14  Cases of this second sort are similar to those in psychological studies on the influence of group effects on individual’s judgments, perhaps the most famous of which are those by Solomon Asch—see, e.g., Asch (1951, 1955, 1956).   15  Instances of beliefs acquired by what Hume calls “education” could be instances of contagious beliefs. In fact, I suspect that they probably are in many cases. It seems to me, however, that such beliefs could be acquired without the operation of sympathy and, hence, are not necessarily instances of contagious beliefs.

Contagion, Community, and Virtue  207 Sympathy, Custom, and Reason. The third type of case is similar to the second but is richer, taking into account the operation of yet another psychological mechanism in the contagious acquisition of beliefs—namely, the operation of reason. According to Hume, it is natural for people to embrace the opinions of others not merely from custom and sympathy, but also from reasoning, which makes us regard other people’s judgment as a kind of argument for what they affirm. In fact, on Hume’s account, sympathy and reasoning—or, more specifically, that subspecies of reasoning that he refers to as ‘authority’—influence “almost all our opinions” (T 2.1.11.9; SBN 320–1). For example, like the woman in our second type of case, a graduate student in philosophy might become accustomed to the views of his dissertation director concerning Hume’s moral psychology. Moreover, he might not only esteem her for her view but believe her to be a trustworthy authority on the matter in question. Consequently, the mechanism of sympathy might enliven his idea of her conclusion, facilitated not only by his attention and inclination to the position itself but also by the operation of reason, concerning not (necessarily) the content of the belief but the authority who holds it.16 As this type of case suggests, a third way in which a person can acquire a belief via contagion is by the operation of sympathy, of custom, and of reason—or, more specifically, of authority.17

11.2.3  The Power of Contagion To this point, I have explained, in some detail, how contagious beliefs are acquired, on Hume’s account. In so doing, I have not only answered the first question raised in the introduction—namely, how are beliefs supposed to be “contagious”? I  have also provided the essential elements for answering the second and third questions— namely, why is the contagious nature of beliefs so powerful, and how does one resist this contagion? The most significant reason that people find it so difficult to resist acquiring beliefs by contagion is that, in many cases, the process originates from a confluence of two or more particularly powerful psychological mechanisms: sympathy, custom, and reason. The way that people can reduce and, thereby, resist the power of the contagion is by intentionally acting in such a way as to eliminate, or at least to mitigate, the effects of one or more of these mechanisms. To illustrate the point, let me return to the particularly intriguing passage from the Natural History of Religion with which I began this essay. In the introduction, I merely mentioned it in passing. Now, let me quote it in full. Concerning religious belief, Hume says,

  16  For a helpful overview of the contemporary psychological research on cases of “epistemic authority,” see Kruglanski et al. (2009)—especially 174–81—and Kruglanski et al. (2010)—especially 945–7.   17  The example I use here involves a belief acquired from a person who is a special authority on the matter in question. This is an example of case of a belief acquired by sympathy, custom, and reason. For the sake of clarity, however, I should note that this third kind of case may also include instances of beliefs acquired from people merely perceived to be reliable authorities. Thus, being a special authority is not necessary condition of being a belief acquired by sympathy, custom, and reason.

208 Rico Vitz The whole is a riddle, an enigma, an inexplicable mystery. Doubt, uncertainty, suspence of judgment appear the only result of our most accurate scrutiny, concerning this subject. But such is the frailty of human reason, and such the irresistible contagion of opinion, that even this deliberate doubt could scarcely be upheld; did we not enlarge our view, and opposing one species of superstition to another, set them a quarrelling; while we ourselves, during their fury and contention, happily make our escape into the calm, though obscure, regions of philosophy. (NHR 15.13, emphasis mine)

According to Hume, one’s doubts concerning such belief “could scarcely be upheld” unless one enlarged one’s view of the competing religious hypotheses. By opposing “one species of superstition to another” and setting them “a quarrelling,” Hume does not mean to imply that the members of the various sects must be present. Rather, he means simply that one must consider a wider variety of competing religious hypotheses and the people who hold them.18 In so doing, one would accomplish three things. First, by considering the views of a variety of people, one would mitigate the influence of sympathy. Second, by attending to a more diverse group of ideas, one would mitigate the influence of custom. Third, by considering the tensions among the claims of various proponents of religious systems, one would undermine their claims to authority and, thereby, mitigate the influence of reason. In short, with respect to belief formation, the process of contagion derives its power from the confluence of two or more particularly influential psychological mechanisms. By mitigating, if not eliminating, the influence of these psychological mechanisms, one is able to resist the power of contagion.

11.3  Doxastic Virtues and Resistible Contagions Having explained, in some detail, Hume’s account of the “contagion of opinion,” I will now elucidate a social aspect of Hume’s epistemology and, thereby, answer the final question, raised in the introduction—namely, under what conditions is it virtuous to resist acquiring beliefs by contagion? Before I proceed, a preliminary methodological clarification might be helpful. What follows will be an extension of my interpretation of Hume’s conception of doxastic virtues. Since I have explained and defended that reading elsewhere,19 I will begin by merely summarize it here.

11.3.1  The Virtuous Belief Principle At the end of Book I of the Treatise, Hume endorses a belief forming norm, which Don Garrett calls the Title Principle. Beliefs formed in accordance with this norm have three notable characteristics. The first is a quality: such beliefs are lively, as opposed to being merely “vigorously conceived.” The second is a cause: such beliefs are produced by a   18  For an application of this kind of strategy, see, e.g., the second part of Hume’s discussion of miracles— EHU 10.24ff, SBN 121ff.   19  Vitz (2009)—see, especially, 220–4.

Contagion, Community, and Virtue  209 properly functioning psychological mechanism the natural effect of which is to produce true and thus, as a rule, useful beliefs. The third is an effect, or a set of effects: such beliefs have a propensity, for example, to command people’s assent or to influence their actions.20 Attributing to Hume a conception of ‘properly functioning’ psychological mechanisms may seem problematic, at first glance. My use of the phrase, however, is nothing more than a way (i) of identifying what Hume regards as psychological mechanisms that function a particular way in mature, healthy human beings and (ii) of contrasting them with those that function differently in people who suffer from what Hume refers to as “madness,” “defects,” “disorders,” “diseases,” and “perversions.”21 The Title Principle is a specific formulation of a more general norm concerning doxastic virtue. According to this norm, which I call Hume’s Virtuous Belief Principle: A person believes, virtuously, that a particular proposition is true if and only if he or she affirms, or is habitually disposed to affirm, an idea representing the proposition, which (1) manifests a certain set of belief-making qualities, such that it is not merely “vigorously conceived,” but is truly “forceful,” “lively,” “vivacious,” “firm,” “steady,” “stable,” and so forth, (2) has a certain cause, such that its belief-making qualities are produced either

(a) by a properly functioning psychological mechanism the natural effect of which is to produce true beliefs, or (b) by a properly functioning psychological mechanism the natural effect of which is to produce useful beliefs;22 and (3) has an effect, or set of effects, such that it has a propensity, for example, to command people’s assent or to influence their actions.

Of these, (1) and (3) are essentially criteria for truly having a belief; it is (2)—which I will call “the causal condition of the Virtuous Belief Principle”—that identifies the distinctively virtuous character of the belief. Thus, on Hume’s account, certain kinds of beliefs fail to be virtuous because they lack the right kind of cause. Rather than being caused either (a) by a properly functioning psychological mechanism the natural effect of which is to produce true beliefs, or (b) by a properly functioning psychological mechanism the natural effect of which is to produce useful beliefs, they are caused by a psychological faculty that produces, for example, delusions that are not merely false but also, in many cases, harmful to the person holding the belief or to those with whom the person interacts.

  20  Vitz (2009: 222).   21  For more on these points, see Vitz (2009: 222, 228n28, 228n29).   22  Previously, I  referred to these as properly functioning cognitive mechanisms—Vitz (2009:  223–4). I now think that phrase is misleading and that the broader phrase “psychological mechanisms” is preferable.

210 Rico Vitz One example of a kind of belief that fails to satisfy the causal condition would be a belief caused by a psychological mechanism that is malfunctioning due to something like fever, madness, or very strong emotions. As Hume notes, When the imagination, from any extraordinary ferment of the blood and spirits, acquires such a vivacity as disorders all its powers and faculties, there is no means of distinguishing betwixt truth and falsehood; but every loose fiction or idea, having the same influence as the impressions of the memory, or the conclusions of the judgment, is received on the same footing, and operates with equal force on the passions. A present impression and a customary transition are now no longer necessary to enliven our ideas. Every chimera of the brain is as vivid and intense as any of those inferences, which we formerly dignified with the name of conclusions concerning matters of fact, and sometimes as the present impressions of the senses. (T 1.3.10.9; SBN 123)23

If such beliefs were caused merely by a momentary and fleeting psychological malfunction, they would not evince a doxastic virtue, but they need not evince a doxastic vice either, since they were not caused by a durable principle of a person’s mind and, hence, not caused by a quality of his or her character. If, however, such beliefs were caused by a durable principle of a person’s mind and, hence, by a quality of his or her character, then they would evince a doxastic vice. On Hume’s account, the type of traditional religious beliefs that he describes as superstitious or delusional seem to be—in many, if not all, cases—examples of beliefs that evince doxastic vices. Let me illustrate the point with three examples. First, with respect to theism, in general, Hume says, Whatever weakens or disorders the internal frame promotes the interests of superstition: And nothing is more destructive to them than a manly, steady virtue, which either preserves us from disastrous, melancholy accidents, or teaches us to bear them. During such calm sunshine of the mind, these spectres of false divinity never make their appearance. On the other hand, while we abandon ourselves to the natural undisciplined suggestions of our timid and anxious hearts, every kind of barbarity is ascribed to the supreme Being, from the terrors with which we are agitated; and every kind of caprice, from the methods which we embrace in order to appease him. Barbarity, caprice; these qualities, however nominally disguised, we may universally observe, form the ruling character of the deity in popular religions. Even priests, instead of correcting these depraved ideas of mankind, have often been found ready to foster and encourage them. The more tremendous the divinity is represented, the more tame and submissive do men become his ministers: And the more unaccountable the measures of acceptance required by him, the more necessary does it become to abandon our natural reason, and yield to their ghostly guidance and direction. Thus it may be allowed, that the artifices of men aggravate our natural infirmities and follies of this kind, but never originally beget them. Their root strikes deeper into the mind, and springs from the essential and universal properties of human nature. (NHR 14.8, emphasis mine)

 

  Cf. T 1.1.1.1; SBN 2; EHU 2.1; 12.11; SBN 17, 153.

23

Contagion, Community, and Virtue  211 Second, with respect to alleged private religious revelations, he claims that “[m]‌adness, fury, rage, and an inflamed imagination” are often thought of as the very dispositions by which people “can have any immediate communication with the Deity” (NHR 6.3). Third, with respect to miracles, Hume suggests that beliefs in such events, resulting from testimony, are frequently influenced by people’s passions. For instance, he famously claims, The passion of surprise and wonder, arising from miracles, being an agreeable emotion, gives a sensible tendency towards the belief of those events, from which it is derived. And this goes so far, that even those who cannot enjoy this pleasure immediately, nor can believe those miraculous events, of which they are informed, yet love to partake of the satisfaction at second-hand or by rebound, and place a pride and delight in exciting the admiration of others. (EHU 10.16; SBN 117)

In essence, Hume contends that the minds of traditional theists—in many, if not all, cases—tend to acquire their religious beliefs by means of weak and disordered psychological mechanisms that are, frequently, affected both by the manipulative work of their religious leaders and by the anticipated admiration of their peers.

11.3.2  Virtuous Resistance So far, I have noted that Hume’s explanation of doxastic virtues and vices focuses on that which causes a person to believe that a certain proposition is true. As I will show next, however, he does not focus exclusively on the function of people’s psychological mechanisms in his analysis of virtuous belief formation. He also emphasizes a social aspect of doxastic virtues. As his reference to “the irresistible contagion of opinion” (NHR 15.13) and his discussion of miracles (see, e.g., EHU 10.16; SBN 117) suggest, Hume is acutely aware of the influence of what we might call a person’s “doxastic community”—i.e., a community of people with whom a person has sufficiently regular and intimate social contact that the person can acquire, by sympathetic contagion, the beliefs of the members of the community. To prevent confusion, let me make two points of clarification. First, as I am using the term here, a “doxastic community” is a community of doxastic influence, as opposed to a community of shared doxastic content. For instance, if there were a U.S. soldier who was held captive as a prisoner in the Soviet Union during the Vietnam War, the group of people who held the soldier captive would constitute a doxastic community. Even if they were to share, say, neither his religious nor his political views, the prisoner would have sufficiently regular and intimate social contact with his captors that he can acquire their beliefs by sympathetic contagion. Second, a person may be a member of multiple doxastic communities at a given time. For instance, the soldier in the previous example may be a member of two distinct doxastic communities— one constituted by his captors and another constituted by his fellow prisoners. Each group has the potential to influence his beliefs, though in strikingly different ways. His captors have the potential to influence him to believe that he should betray his

212 Rico Vitz fellow prisoners. His fellow prisoners have the potential to influence him to maintain his belief that he should be loyal to those who share his plight. Not only is Hume aware of the influence of a person’s doxastic communities, he is concerned with the profound power such communities have to cause people’s beliefs and the conditions under which it is virtuous for people to resist their influence. As his discussions of traditional religious beliefs suggest, Hume contends that given the profound “frailty of human reason,” virtuous people need to be vigilant, lest their social interactions cause them to form superstitious or delusive beliefs. In essence, he suggests that if people believe that their doxastic communities are causing, or are likely to cause, them to acquire beliefs that violate the causal condition of the Virtuous Belief Principle, then they ought to find ways to resist the influence of their communities, as he not only commits himself to doing but encourages others to do at the end of the Natural History. Thus, Hume’s account of doxastic virtue emphasizes not only the role of the operations of the psychological faculties that cause people’s beliefs, but also the role of the social circumstances that trigger the operations of those psychological faculties. In this respect, his conception of doxastic virtue is particularly rich since it encompasses a concern not only with (i) the gathering and assessment of evidence24 and (ii) the operations of psychological faculties that are sensitive to evidence—such as reason, memory, and imagination25—but also with (iii) the operations of other psychological faculties, such as sympathy and custom, and, consequently, with (iv) the way in which people deliberately manage their social circumstances.

11.4 Conclusion Let me close by summarizing Hume’s account, briefly, and by noting some of its more noteworthy strengths. According to Hume, beliefs about matters of fact are either   24  Here and in Vitz (2009), I have analyzed only Hume’s account of the conditions for evaluating people for the way in which they assess evidence virtuously and, hence, identified only his Virtuous Belief Principle. I  have not analyzed the his account of the conditions for evaluating people for the way in which they gather evidence virtuously and, consequently, identified what we could call his Virtuous Enquiry Principle. Nonetheless, he clearly seems to have in mind a set of conditions by which one rightly gathers evidence and, thereby, manifests certain doxastic virtues, like “just reasoning” and avoids manifesting certain doxastic vices, like “credulity, or a too easy faith in the testimony of others” (T 1.3.9.12; SBN 112; cf. T 1.3.9.13, 1.3.10.4, 1.4.7.6; SBN 113, 120, 267; EHU 10.1, 26; SBN 109, 124) or “supine indolence of mind” (EHU 5.1; SBN 41)—hence, for example, his emphasis on the necessary role of “experiments” in forming beliefs well. A section from the beginning of T 2.1.12 clearly illustrates his concern with virtuous enquiry. After offering a number of experimental observations for his account of the causes of pride and humility, he says, “It is usual with anatomists to join their observations and experiments on human bodies to those on beasts; and, from the agreement of these experiments, to derive an additional argument for any particular hypothesis. . . . Let us therefore apply this method of enquiry, which is found so just and useful in reasonings concerning the body, to our present anatomy of the mind, and see what discoveries we can make by it” (T 2.1.12.2; SBN 325–6, my emphasis).   25  Cf. T 1.3.4.2, 1.3.11.2, 2.3.1.12–16; SBN 83, 124, 403–6; EHU 4.1–4, 10.4, 12.6; SBN 25–7, 110, 151. I am using the term “imagination,” here, in Hume’s narrow sense of the term—that is, as opposed to reason—cf. T 1.3.9.19n22; SBN 117–18n1; see also Garrett (1997: 11–29).

Contagion, Community, and Virtue  213 (i) ideas enlivened by a psychological mechanism, like the principle of custom, acting on a present impression to which the idea is related, or (ii) acquired dispositions to have such ideas. Such beliefs can be acquired by contagion as a result of the operation of (at least) three psychological mechanisms: sympathy, custom, and reason. The confluence of all three of these mechanisms makes beliefs particularly contagious and difficult to resist. So, it is natural for people to acquire the beliefs of the members of their doxastic communities, especially if they view certain members of those communities as experts on a given issue. Nonetheless, virtuous people must resist the contagious influence of their doxastic communities, if these groups will cause them to acquire beliefs neither by a properly functioning psychological mechanism the natural effect of which is to produce true beliefs, nor by a properly functioning psychological mechanism the natural effect of which is to produce useful beliefs, but rather by a kind of psychological faculty that produces, say, superstitious delusions that are not merely false but also harmful to the people who hold them or to those with whom they interact. Hume’s account of doxastic virtue has at least two particularly noteworthy strengths. First, it encompasses the rich and complex nature of human beings, who—even with respect to the process of belief formation—are not merely reasonable but also social and active beings (cf. EHU 1.6; SBN 8–9). In this respect, it is superior to certain narrowly rationalistic conceptions of doxastic virtue that overemphasize the role of reason in the process of belief formation. Second, it recognizes that people can control their beliefs, indirectly, not only by influencing whether they will perceive the relevant evidence, but also by influencing how they will perceive the evidence. In this respect, it is superior to those that conceive of indirect doxastic voluntarism as little, if anything, more than an issue concerning the way in which people position themselves with respect to the evidence. Hume’s conception of the social aspect of doxastic virtue—particularly, his account of the “contagion of opinion”—would present yet another noteworthy strength if his position is supported by contemporary research in social psychology. In fact, a cursory look at the scholarly literature—especially, e.g., the research on “lay epistemics” and “epistemic authority”—looks favorable.26 For these reasons and others, Hume’s account of doxastic virtue should be of interest not only to the narrow array of philosophers who are doing research on the early modern period, but also to the broader band of philosophers and psychologists who are doing research at the nexus of ethics, epistemology, and cognitive science.27

  26  In addition to the psychological research that I mentioned above, see also, e.g., Raviv et al. (1990, 1993, 2003), Guimond and Palmer (1996a, 1996b), Leman (2002), Kruglanski et al. (2005), and Klucharev et al. (2008).   27  I would like to thank Don Garrett, Tom Holden, and Adam Green for helpful comments on earlier versions of this essay.

214 Rico Vitz

References Abramson, K. (2001). “Sympathy and the Project of Hume’s Second Enquiry,” Archiv Fur Geschichte Der Philosophie 83: 45–80. Altmann, R. W. (1980). “Hume on Sympathy,” Southern Journal of Philosophy 18: 123–36. Asch, S. E. (1951). “Effects of Group Pressure on the Modification and Distortion of Judgments,” in H. Guetzkow (Ed.), Groups, Leadership and Men. Pittsburgh: Carnegie Press. Asch, S. E. (1955). “Opinions and Social Pressure,” Scientific American 193: 31–5. Asch, S. E. (1956). “Studies of Independence and Conformity: I. A Minority of One against a Unanimous Majority,” Psychological Monographs: General and Applied 70: 1–70. Audi, R. (1994). “Dispositional Beliefs and Dispositions to Believe,” Noûs 28: 419–34. Debes, R. (2007a). “Has Anything Changed? Hume’s Theory of Association and Sympathy after the Treatise,” British Journal for the History of Philosophy 15: 313–38. Debes, R. (2007b). “Humanity, Sympathy and the Puzzle of Hume’s Second Enquiry,” British Journal for the History of Philosophy 15: 27–57. Garrett, D. (1997). Cognition and Commitment in Hume’s Philosophy. Oxford:  Oxford University Press. Gorman, M. M. (1993). “Hume’s Theory of Belief,” Hume Studies 19: 89–102. Guimond, S. and Palmer, D. L. (1996a). “Developmental Differences in Children’s Understanding of Epistemic Authority,” Journal of Applied Social Psychology 26: 683–702. Guimond, S. and Palmer, D. L. (1996b). “The Political Socialization of Commerce and Social Science Students:  Epistemic Authority and Attitude Change,” Journal of Applied Social Psychology 26: 1985–2013. Hume, D. (1975). Enquiries concerning Human Understanding and the Principles of Morals, 3rd edn. Oxford: Oxford University Press. Hume, D. (1978). A Treatise of Human Nature, 2nd edn. Oxford: Oxford University Press. Hume, D. (1980). Dialogues concerning Natural Religion, 2nd edn. Indianapolis: Hackett. Hume, D. (1987). Essays, Moral, Political, and Literary. Indianapolis: Liberty Fund. Hume, D. (1999a). An Enquiry concerning Human Understanding. Oxford: Oxford Uni­versity Press. Hume, D. (1999b). An Enquiry concerning the Principles of Morals. Oxford: Oxford University Press. Hume, D. (2000). A Treatise of Human Nature. Oxford: Oxford University Press. Hume, D. (2007). A Dissertation on the Passions; The Natural History of Religion. Oxford: Clarendon Press. Johnson, O. A. (1995). The Mind of David Hume. Urbana and Chicago: University of Illinois Press. Klucharev, V., Smidts, A., and Fernandez, G. (2008). “Brain Mechanisms of Persuasion: How ‘Expert Power’ Modulates Memory and Attitudes,” Social Cognitive and Affective Neuroscience, 34: 353–66. Kruglanski, A. W., Dechesne, M., Oreheka, E., and Pierro, A. (2009). “Three Decades of Lay Epistemics: The Why, How, and Who of Knowledge Formation,” European Review of Social Psychology, 20: 146–91. Kruglanski, A. W., Orehek, E., Dechesne, M., and Pierro, A. (2010). “Lay Epistemic Theory: The Motivational, Cognitive, and Social Aspects of Knowledge Formation,” Social and Personality Psychology Compass, 10: 939–50.

Contagion, Community, and Virtue  215 Kruglanski, A. W., Raviv, A., Bar-Tal, D., Raviv, A., Sharvit, K., Ellis, S., Bar, R., Pierro, A., and Mannetti, L. (2005). “Says Who? Epistemic Authority Effects in Social Judgment,” in M. P. Zanna (Ed.), Advances in Experimental Social Psychology, Vol. 37. New York: Academic Press, 345–92. Leman, P. J. (2002). “Argument Structure, Argument Content, and Cognitive Change in Children’s Peer Interaction,” Journal of Genetic Psychology 163: 40–57. Loeb, L. E. (2005). Stability and Justification in Hume’s Treatise. Oxford: Oxford University Press. Pears, D. (1991). Hume’s System: An Examination of the First Book of His Treatise. Oxford: Oxford University Press. Price, H. H. (1967). “Some Considerations about Belief,” in A. P. Griffiths (Ed.), Knowledge and Belief. Oxford: Oxford University Press, 41–59. Price, H. H. (1969). Belief. London: Allen and Unwin. Raviv, A., Bar-Tal, D., Raviv, A., and Abin, R. (1993). “Measuring Epistemic Authority: Studies of Politicians and Professors,” European Journal of Personality 7: 119–38. Raviv, A., Bar-Tal, D., Raviv, A., Biran, B., and Sela, Z. (2003). “Teachers’ Epistemic Authority: Perceptions of Students and Teachers,” Social Psychology of Education 6: 17–42. Raviv, A., Bar-Tal, D., Raviv, A., and Houminer, D. (1990). “Development in Children’s Perception of Epistemic Authorities,” British Journal of Developmental Psychology 8: 157–69. Russell, B. (1921). The Analysis of Mind. London: George Allen and Unwin. Vitz, R. (2002). “Hume and the Limits of Benevolence,” Hume Studies, 28: 271–95. Vitz, R. (2004). “Sympathy and Benevolence in Hume’s Moral Psychology,” Journal of the History of Philosophy, 42: 261–75. Vitz, R. (2009). “Doxastic Virtues in Hume’s Epistemology,” Hume Studies, 35: 211–29. Vitz, R. (forthcoming). “The Nature and Functions of Sympathy in Hume’s Philosophy,” in P. Russell (Ed.), The Oxford Handbook of David Hume. Oxford: Oxford University Press.

12 Understanding Epistemic Normativity in Feminist Epistemology Heidi Grasswick

Elizabeth Anderson characterizes feminist epistemology as “The branch of social epistemology that investigates the influence of socially constructed conceptions and norms of gender and gender-specific interests and experiences on the production of knowledge” (Anderson 1995a).1 As noted by Fred Schmitt in his 1994 anthology on social epistemology, concerns of normativity are often cited as a feature of social epistemology that distinguishes it from “mere” sociology of knowledge (Schmitt 1994, 1). Feminist epistemology, as a form of social epistemology particularly interested in how social relations of gender affect knowledge-seeking enterprises, has likewise always had markedly normative interests. However, feminist epistemology has been interpreted by some as either lacking normative content and thus failing as a robust form of epistemology, whatever interesting insights it might provide for sociology of knowledge, or as having the wrong kind of normative content to constitute a genuine epistemology. I argue that these criticisms and misunderstandings actually stem from different views concerning central tasks of epistemology and, correspondingly, what is required from epistemic normativity. After first explaining why feminist epistemology is sometimes understood to lack sufficient epistemic normativity, I  identify four features of the normative projects feminist epistemologists engage in, articulating these features as evolving out of the core feminist view of knowing as socially situated. These features include prioritizing questions concerning knowledge-seeking activities and practices, contextualizing

  1  This chapter has benefited from comments from audience members at the 2011 Ethics of Belief conference at University of North Florida and I am indebted to the organizers of the conference (and editors of this volume) Jon Matheson and Rico Vitz. I would also especially like to thank Rico Vitz for his extensive and thoughtful feedback on an earlier draft.

Epistemic Normativity in Feminist Epistemology  217 epistemic norms, accepting the entanglement of the epistemic and the ethical, and attending to the perspective of epistemic agents. By understanding the normative repercussions of a view of socially situated knowing, it becomes apparent that some of the disagreements about the normative in feminist epistemology are fundamentally about differences in what various theorists expect of a normative epistemology. Moreover, I illustrate points of contact between feminist epistemology and some of the interesting and important recent developments in epistemology including naturalism, contextualism, and virtue epistemology. These intersections show that feminists are not alone in developing alternative understandings of epistemic normativity, even as they arrive at their conclusions via a distinctive route, a route dedicated to understanding the role of social power relations in epistemic pursuits.

12.1  The Normative Interests of Epistemologists and Feminists Epistemology is a normative enterprise in so far as it attempts to guide our belief formation, or regulate our belief-forming practices. As James Maffie notes, the normative is “intimately connected with human conduct and motivation” and our epistemic concepts, facts, and judgments “proffer reasons to believe and tell us what actions are valuable, desirable, or worthwhile from the epistemic point of view” (Maffie 1996, 334). There are clearly a number of approaches epistemologists have taken in articulating “the epistemic point of view,” and a number of different epistemological questions whose answers have normative content in the senses that Maffie describes. However, it is worth noting one very influential epistemological framework within the Anglo-American canon that has been referred to as the “traditional framework” (Millar 2010). Those who work within this framework understand knowledge in terms of belief that satisfies certain conditions, with truth being one of these. For these theorists, one of the major tasks for epistemology and that which has acquired the lion’s share of attention is the “traditional analytic project” of providing necessary and sufficient conditions for the truth of “S knows that p” claims (Millar 2010), often with a focus on the condition(s) needed to supplement true belief in order for something to qualify as knowledge. In seeking such an abstract and general formulation of the conditions of knowledge that will hold for any subject S and any proposition p, this approach commits to the idea of the universality of epistemic norms, and the ability to generate such norms through conceptual analysis (Knowles 2003). Attempts to defeat the skeptic (by articulating conditions of knowledge that people are able to fulfill successfully), outlines of the structure of justification (such as theories of foundationalism and coherentism), proposed solutions to Gettier problems, the evolution of the internalism/externalism debates, and developments in reliabilism can all be understood as developments that stem from this primary traditional analytic project.

218  Heidi Grasswick The traditional analytic project is very much a normative project. “Knowledge” is a success term, and these traditional epistemologists have concerned themselves with conceptual analyses of it in order to clarify how it is differentiated from other forms of belief such as opinion. The difference discovered through such analysis is expected to capture the characteristics that make knowledge good or recommended as something worthy of seeking, or at least worthy of seeking when compared with other forms of belief. By providing a conceptual analysis of knowledge, normative assessments of beliefs can be made in turn, asking for a particular belief: Is it true? Is it justified? Is it knowledge? In the case of particularists, such as Chisholm, particular instances of beliefs are identified as clearly justified, followed by attempts to derive general principles of justification out of such cases (Chisholm 1989). For those engaged in the traditional analytic project who focus on the status of beliefs, answers to more systemic and agent-centered questions, such as how to be good inquirers, what practices we should engage in, and what activities we must undertake in order to be epistemically praiseworthy must then be derived from investigations into the general category of knowledge itself. To give just one example, Alvin Goldman’s veritistic social epistemology begins by defining “knowledge” (in the weak sense)2 as true belief, and then proceeds to consider which social epistemic practices have a favorable impact in terms of leading us to true beliefs (Goldman 1999). The acquisition of pieces of knowledge is understood as the central epistemic goal, and the normative work is done at the level of providing an abstract analysis of the conditions of knowledge that supposedly could then be applied to particular epistemic situations. Feminists enter the arena of epistemology with more than just a general interest in understanding knowledge as a category. Their interests in epistemic normativity are in large part generated out of commitments to political programs of social justice. Feminists recognize that their progressive social and political projects depend on an ability to know the world well, where the phrase “know well” (here and throughout this chapter) expresses the epistemic goal of engaging in epistemically sound practices and activities that tend to lead to a rich understanding of the world appropriate to our needs.3 Feminists obviously make knowledge claims as part of their social and political projects—claims about such things as the existence of oppression, the workings of oppression, and the injustice of practices of sexism, for example. But they also seek to justify their political projects of social change—eliminating oppression, for example—through reference to many of these knowledge claims, holding them up as

  2  Goldman’s “weak” sense of “knowledge” is meant to be distinguished from a “strong” sense that includes some additional element such as a justification or warrant clause.   3  I consciously use the term “appropriate to our needs” rather than something like “that which directly addresses our needs.” I seek to avoid a position that implies that knowledge has only instrumental value derived from our practical goals and needs. It may well have some intrinsic value as well. Nevertheless, I do mean to build into the idea of “knowing well” a sense that collections of practices of knowing that deliver some forms of knowing, but fail to deliver those types of knowing that are important for our needs, would mark an epistemic failing.

Epistemic Normativity in Feminist Epistemology  219 epistemically superior to alternative claims, such as the denial that oppression exists in society. For this task, feminists need epistemological standards and methods of distinguishing what constitutes knowing well and knowing poorly, where “knowing poorly” would be understood as engaging in practices and activities that fail to deliver a rich understanding of the world appropriate to our needs.4 In short, feminists must engage in the normative work of epistemology. Expressed in more general terms, Sally Haslanger argues that the heart of epistemic value, and hence the source of epistemology’s normativity itself, lies in how our epistemic practices allow for a certain kind of intrinsically valuable autonomous agency for beings like us. In short, having beliefs is “a constitutive requirement of a kind of agency that is part of a eudaimon life” and “truth is a constitutive value of belief.” Hence, evaluations of how well our epistemic practices contribute to this kind of autonomous agency is central to the work of epistemology (Haslanger 1999). Turning to the specific interests of feminists, their work as agents of social and political change depends upon having such autonomous agency as Haslanger outlines. It should be obvious that for feminist epistemologists, simply describing how social relations of gender are implicated in our past and present knowledge-seeking endeavors won’t suffice. After all, part of the point of feminist epistemology has been to reveal the ways in which sexist and androcentric biases, gendered epistemic injustices, and oppressive power relations in a variety of forms have worked against good knowing. Evaluating the appropriate role of social relations of gender and other axes of social power in practices of knowing well is crucial. When do particular social relations get in the way of good knowing, and how can we know well within situations of differential social power? Such questions are central to projects of feminist epistemology. Because of feminists’ need to criticize certain knowledge practices on epistemic grounds in order to support their social and political goals, it has been argued that feminists have an even stronger motivation than many social epistemologists for ensuring that they are able to maintain a normative epistemology as they incorporate social dimensions of knowing into their analyses (Grasswick 2006).

12.2  Critiques of Normativity in Feminist Epistemology In spite of their motivation, and a clear sense amongst feminist epistemologists that they are doing normative work, some critics have interpreted feminist epistemology as

  4  A practice might lead to “poor knowing” in a few different ways. It might simply be unreliable, generating numerous false claims, or it might succeed in generating true claims, but not very many of them, or not very important ones. Though “knowledge” is a success term, making the phrase “poor knowledge” oxymoronic, and “good knowledge” redundant, the terminology of “knowing well” (or “good knowing”) and “knowing poorly” (or “poor knowing”) does not work in quite the same way. As I have defined them they do not speak to single instances of knowing, but particular patterns of epistemic success.

220  Heidi Grasswick failing to provide a robust normative epistemology, by providing either the wrong kind of normativity or no normativity at all. How can this be, given the strong motivation for feminist epistemology to have normative content? There are two different characteristics of feminist epistemology that account for such difficulties in interpretation. The first concerns the relationship between epistemic and ethical normativity. Many epistemologists, particularly those involved in the traditional analytic project, insist that epistemic normativity must be clearly distinct from ethical normativity. For these theorists, one of the important epistemological tasks is to isolate the normativity specific to epistemology and the twin cognitive goals of attaining truth and avoiding falsity. Of specific concern to them is that we not confuse the pragmatic values of particular beliefs—e.g., how they allow us to achieve our specific goals in the world, including ethical goals—with their epistemic value, understood as restricted to their connections with the value of truth (Sosa 2007). For these epistemologists truth-conduciveness—understood in a broad sense, including avoidance of falsity—is as far as epistemic normativity can reach. For theorists who narrowly define epistemic value in terms of truth-conduciveness and strive to isolate the epistemic value of beliefs from any other value they might carry, the ethico-political motivation of feminist epistemology immediately appears suspicious. The complaint here is that the normative concerns of feminist epistemologists are actually mislabeled as epistemic concerns, when they really amount to nothing more than political or moral concerns. According to such critics, feminist epistemologists do not offer any epistemic normativity but simply want, and allow, ethical and political normative concerns of developing a socially just society to trump authentic epistemic concerns. Susan Haack, for example, argues that the idea of a “feminist” epistemology encourages the politicization of inquiry, which she judges always to be unsound (1998: 119). Further, she worries that a genuine feminist epistemology would legitimate the idea that “feminist values should determine what theories are accepted,” rather than let concerns of truth dictate the result (1998: 128). According to such critiques, feminist epistemologists offer theories that permit feminists to select knowledge claims and theories according to how well they serve their political purposes, or match their social goals, with less regard, or rather an inappropriately low regard, for the truth of the claims or theories. An example of work that some, such as Haack, have interpreted this way comes in feminist philosopher of science Helen Longino’s discussion of theory choice. According to Longino, it is acceptable for feminists to prefer one theory over another because that theory ultimately supports feminist values better. More precisely put, following Kuhn’s recognition that there are multiple epistemic values at play—with his primary candidates being accuracy, consistency, scope, simplicity, and fruitfulness— Longino argues that one may legitimately prioritize one particular epistemic value over another because the chosen one will do a better job of revealing gendered relations. This selection and prioritization of certain epistemic values over others contributes to theory choice itself. In one of Longino’s examples, she contrasts the epistemic values of

Epistemic Normativity in Feminist Epistemology  221 simplicity versus ontological heterogeneity applied to the study of households in economics. Embracing the epistemic value of simplicity, many economists have used the interests of the male (assumed) head-of-household as a stand-in for household interests. Longino argues instead that in this case a theory embracing the epistemic value of ontological heterogeneity might be legitimately preferred and selected over the standard theories that embrace simplicity. The preference can be justified because a theory that appeals to the virtue of ontological heterogeneity by accepting the possibility of gendered differences in interests within the household will do a better job revealing these gendered relations, should they be present. Longino takes the view that a criteria for a feminist choice of a particular epistemic value, or virtue, is that it “does not let gender disappear” (1994: 481) and that if a context is structured by gendered power asymmetries, “inquiry guided by these virtues is more likely to reveal it or less likely to preserve its invisibility than the traditional virtues” (1997: 50). In the economics example, selecting a theory that trades off greater simplicity for greater ontological heterogeneity is more likely to reveal the important gendered relations in the household. It is important to note that Longino never argues that epistemic values such as ontological heterogeneity can replace what many consider to be the core epistemic value of empirical adequacy, and this point should lead us to question the merits of the criticisms of Longino that interpret her as allowing feminists to select theories on the basis of one’s (non-epistemic) values without regard for truth, or in Longino’s empiricist framework, empirical adequacy.5 But regardless of assessing the merits of the criticism of Longino, my point here is that the very fact that feminist epistemology stems from and is framed by ethico-political concerns makes it susceptible to this kind of misinterpretation by which feminists are seen as prioritizing ethical normativity over epistemic normativity and mislabeling that ethical normativity as epistemic normativity. The persuasiveness of such critiques, however, depends on the premise that epistemic assessments must be restricted to mere truth-conduciveness, and that epistemic normativity must remain pure, untainted by any ethical values. As I discuss in more detail later, the separability of ethical and epistemic normativity and the ability to isolate the epistemic completely is one major claim that feminist epistemologists are questioning through their work. Another critique concerns the role descriptions of social practices play in feminist epistemology and argues that no epistemic normativity at all is provided by feminist accounts. Rather, the claim is that feminist epistemology, and in particular feminist critiques of science, are merely engaged in providing interesting sociological descriptions of social practices and insights concerning how certain knowledge claims have come to hold credibility within society without engaging in the normative work of epistemology. Indeed, feminist critics of science do offer a vast array of examples of   5  Elizabeth Anderson makes this point very clearly in her criticism of Haack’s interpretation of Longino in her piece “How Not to Criticize Feminist Epistemology: A Review of Scrutinizing Feminist Epistemology” (2008).

222  Heidi Grasswick how gender bias, among other biases, have affected various scientific claims and theories and their shelf life for the worse, from the dominance of man-the-hunter theories in archaeology (Hubbard 1983; Bleier 1984; Nelson 1993), to the preponderance of social dominance theories in primatology (Haraway 1989), to inadequate evolutionary theories of female orgasm (Lloyd 2005), to the gendered stereotypes invoked in standard (yet now understood as poor) explanations of the comparative roles of the egg and sperm in fertilization (Martin 1991). The critics’ argument here is that the attention feminist epistemologists give to the social histories of the generation of knowledge claims and the role of power relations in such knowledge production consists only in descriptions of cases which ultimately fail as knowledge—i.e., cases where such social influences have gotten in the way of good knowing—or cases which succeed as knowledge but where the interesting social history provided has had no bearing on this epistemic success. The origin of this kind of critique lies in the traditional approach to epistemic normativity that I described above: that of engaging in the traditional analytic project of setting out the general conditions of knowledge as an entity and then answering other normative questions that are given priority, such as what the status is of any particular belief or whether that belief was well-formed, in light of the abstract conception of knowledge provided. If one takes the core normative work of epistemology to be the articulation of the necessary and sufficient conditions of knowledge as a product, then for any claim we examine, it either stands or falls as normatively praiseworthy based on its satisfaction of these necessary and sufficient conditions. Descriptions of the social practices that resulted in the claim’s formation will be extraneous.6 “Knowledge” is conceptualized as an ideal and for those working in the traditional framework, this ideal is developed primarily through conceptual analysis with no acknowledgement of the potential relevance of descriptions of how current practices operate.7 If the epistemic results satisfy the criteria for knowledge, then the gender biases of the practices that generated them cannot be criticized on epistemic grounds; if the results fail as knowledge, they do so on the basis of their failure to satisfy the abstract conditions of knowledge that were generated through a conceptual analysis that does not concern itself with the social details of current practices. But this focus on the status of particular knowledge claims and the ideal of “knowledge” misses a key interest of feminist epistemologists: understanding the systematic nature of our epistemic practices and assessing these practices accordingly. Feminist

  6  This is especially so for the majority of traditionalists who have conceptualized the justification condition of knowledge as involving sound individual reasoning, with little consideration of how social conditions might contribute to a belief ’s justification.   7  Of course, some contemporary analyses of knowledge have built into them reference to how the knowledge was itself acquired. In particular, reliabilist and causal theories of knowledge include a condition of knowledge being the result of particular (reliable) processes. Reliabilists have accordingly been increasingly drawn to more specific (and at times empirical) assessments of which practices count as reliable, though the focus usually remains on assessments of the status of the beliefs themselves.

Epistemic Normativity in Feminist Epistemology  223 epistemologists are less interested in the status of any particular androcentric claim than they are in demonstrating the systematic distortion of our knowledge production that can and does happen as a result of gendered power relations in society. That is to say, they are less concerned with identifying any particular scientific theory as epistemically “wrong” than they are in showing how something like androcentric bias in science—or racist bias, or any bias that operates as part of the practices of oppression— can set up a pattern in which we are less likely to achieve sound epistemic results. Assessments of individual knowledge claims, such as found in the work of many epistemologists engaged in the traditional analytic project, can’t do the work that feminists need done.8 Here again, the dispute between feminist epistemologists and their critics is better understood as a dispute regarding what kind of normativity we need from our epistemology rather than whether or not there is normativity present at all. As I argue below, unlike many epistemologists working within the traditional framework, feminist epistemologists have not taken the question of the proper conception of knowledge and corresponding assessments of particular knowledge claims to be primary, but rather have framed the normativity of their projects in response to other epistemological questions they take to be more pressing, especially those questions concerning the activities and social practices of “knowing well.” There is a sense in which the traditional framework and feminist epistemology operate like two different Kuhnian paradigms, each with their own set of anomalies or challenges to be worked out. For the remainder of this chapter, I look more closely at feminist epistemology itself, identifying certain characteristics of its conceptualization of epistemic normativity. Along the way, I illustrate these characteristics through examples of feminist work, and I draw attention to parallels that can be found between feminist epistemology and several recent developments in epistemology that also challenge aspects of the traditional framework.

12.3  Socially Situated Knowing The wide variety of approaches and projects that fall under the category of feminist epistemology makes it difficult, and intellectually dangerous, to discuss feminist epistemology in general terms.9 However, a central thesis common across feminist   8  Importantly, many feminist epistemologists have argued more strongly that the traditional framework of epistemology has not only failed in those specific normative tasks that feminists seek to deliver on, such as offering critiques of gender-biased practices, but that in fact, this traditional framework has made such tasks harder. For example, feminists argue that by conceptualizing epistemic subjects abstractly, as interchangeable with each other, the traditional framework allows masculine perspectives to be masked as the view-from-nowhere (Code 1991). Although these critiques of the traditional framework form an important part of the field of feminist epistemology, I do not engage them in this chapter. My primary focus is instead on understanding the differences in the frameworks and how they need to be understood on their own terms.   9  Additionally, amongst audiences not well versed in feminist epistemology, there are common misunderstandings concerning what feminist epistemology is, one of the most prominent being the idea that feminist epistemology uncritically embraces “women’s ways of knowing” (Rooney 2012). Feminist epistemology

224  Heidi Grasswick epistemologies is the idea that knowing is socially situated (Haraway 1988; Code 1991; Harding 1991; Anderson 2011). My goal in this section is to articulate a rather minimal interpretation of the situated knowing thesis, in order to capture the core components that can be identified across a range of feminist works and examine their implications for how questions of epistemic normativity become structured. This approach will be helpful in coming to understand where the tensions are, and where the attractions are, between feminist approaches to epistemologically normative questions and those found within other recent epistemological developments that also diverge from the traditional framework such as naturalism, contextualism, and virtue epistemology. As Phyllis Rooney notes, it is important to understand the connections and overlaps between feminist epistemology and other recent developments in epistemology, even as feminist epistemology does not map perfectly onto any one of them and follows its own trajectory (Rooney 2012). The first feminist articulation of the idea of situated knowing came in Donna Haraway’s article “Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective” (1988). Originally written as a commentary on Sandra Harding’s work, Haraway introduced the idea of situated knowing to characterize the challenges facing feminist epistemologists as they negotiated their way between social constructionism and realism, emphasizing the uneliminable role of the social in shaping knowledge, yet maintaining a commitment to knowledge being oriented toward gaining the truth about the world. To claim that knowing is socially situated is to claim that people know the world from particular perspectives, and necessarily so. Their social locations shape and limit their knowledge, with the implication that knowers are not all similarly situated from an epistemic point of view, at least not in socially stratified societies. Additionally, the feminist thesis of socially situated knowing recognizes that people’s differences in social situatedness contribute to their epistemic dependencies on each other, making clear the need to negotiate through various socially situated perspectives in their epistemic tasks. Because situated knowing is most closely associated with feminist standpoint theory, it is worthwhile pausing for a moment to give a brief overview of standpoint theory for the uninitiated. At the same time, however, it is important to stress that the basic ideas of situated knowing that I have set out thus far are more broadly employed across a variety of feminist epistemologies. Feminist standpoint theory, with its origins in Marxist social theory, maintains that certain underprivileged social positions offer the potential for a kind of epistemic privilege in particular areas of inquiry, especially those areas concerned with knowledge of social relations. In one version of the theory, the potential for epistemic privilege results from the socially underprivileged’s dual vision on the social world (Hartsock 1983): the dominant class structures the social world in certain ways, and because the is clearly distinguishable from such versions of “feminine epistemology” in its attempt to understand the connections between power relations of gender and knowledge (Grasswick 2011).

Epistemic Normativity in Feminist Epistemology  225 oppressed live within those structures, they too have access to this dominant vision of the world of social relations. But additionally, the oppressed have access to their own experiences of being exploited, something the dominant class lacks. With this double vision the oppressed have the potential to develop an epistemically privileged standpoint that allows them to see the oppressive social relations for what they really are, and this marks an epistemic achievement that is much more difficult for a member of the dominant class. The standpoint of the oppressed then is epistemically privileged in the sense that it offers a better and less partial understanding than does the perspective of the dominant. An important detail of standpoint theory is the fact that a standpoint is an achieved stance, not merely a situated perspective. According to standpoint theorists, the oppressed do not automatically carry epistemic privilege in virtue of their position, but rather their situation permits the development of a political consciousness that integrates the dual nature of their vision, enabling them to have a better understanding of the social world than those in a dominant position.10 There has been much debate within feminist epistemology with respect to both how to understand the idea of epistemic privilege, and what its full implications are (see for example Nelson 1990; Bar On 1993; Longino 1993; Wylie 2003). But it is not necessary to work these out here, or even to commit to a standpoint version of feminist epistemology. For the purposes of this chapter, a minimalist interpretation of socially situated knowing that does not require a commitment to any particular thesis of or argument for epistemic advantage or privilege will suffice. This minimal interpretation commits only to the idea that social situation both shapes and limits knowing, presenting feminists with the important epistemological task of negotiating through such differences across epistemic subjects. Socially situated knowing is first and foremost a thesis about knowing subjects, offering a rejection of what feminist epistemologists have labeled the atomistic view of individual knowers (Jaggar 1983; Code 1991), according to which knowers can be understood from the epistemic point of view as abstract, generic or interchangeable, and self-sufficient individuals (Grasswick 2004). According to socially situated knowing, knowers cannot be assumed to be interchangeable, but must be understood epistemically as differentially socially located. Social location makes a difference to how and what a person can know. But social situatedness is of course also a thesis about the knowing relationship itself. It amounts to a denial of the traditional conception of the knowledge relation as constituting a view-from-nowhere representation of the world (Code 1991). This is the ‘limiting’ nature of socially situated knowing, and it has ramifications for how other epistemological questions are answered. Both these parts of the situated knowing thesis impact feminists’ conception of the normative work to be done in epistemology.   10  Recognizing a standpoint as an achieved stance allows standpoint theorists to acknowledge that many of the socially underprivileged may experience internalized oppression, and may fail to achieve the dual vision that can give them epistemic insight into the structure of oppressive social relations.

226  Heidi Grasswick These two aspects of the minimal interpretation of the socially situated knowing thesis—that one’s knowing is differentially shaped by social location, making it epistemically relevant to examine a knower’s social location, and that the knowing relation must always be limited and perspectival presents feminist epistemologists with a key challenge: how are we to understand the relationship between particular socially located perspectives and an epistemically appropriate perspective? This is where the normative work of feminist epistemology is done. A central epistemological, and normative, task for those who adopt a situated approach becomes the interrogation of the epistemically relevant features of situation. As Lorraine Code explains: ‘situation’ is not just a place from which to know, as the language of ‘perspectives’ might imply, indifferently available to anyone who chooses to stand there. Situation is itself a place to know whose intricacies have to be examined for how they shape both knowing subjects and the objects of knowledge; how they legitimate and/or disqualify knowledge projects . . .  It is an achieved epistemic stance, knowledgeably chosen as a place that can be mapped to facilitate responsible knowing. (2006: 40–1)

Feminist epistemologists have used the idea of socially situated knowing to interrogate and explain how and why power relations within society affect the quality and content of our knowing. Beginning with gender, but taking up many lines of social stratification (Alcoff and Potter 1993), much of the work has focused on providing evidence of the thesis of socially situatedness, showing that claims that have been understood as strong examples of knowledge bear the social marks of their creators. At the same time, feminists have used the framework of social situatedness to investigate the ways in which we can expect epistemic perspectives to differ, using that understanding to develop normative recommendations for how we can negotiate through such differences and in the process improve our knowing practices. For example, Harding has increasingly stressed the methodological guidance feminist standpoint theory has to offer, arguing that its use as a methodology in various disciplines is where the theory has been most productive (Harding 2009). Given the fact that the perspectives of marginalized groups will not be reflected in the dominant knowledge forms, Harding argues that social science researchers should begin their research by thinking from the lives of women and from the lives of the racially marginalized in an attempt to incorporate the insights available from those perspectives into the research project.11 Taking a different approach, feminist empiricists such as Longino (1990, 2002) have used social situatedness, coupled with a procedural conception of objectivity, to explain how increasing social diversity within research communities can offer epistemic resources to those communities, improving the practices of knowledge production as a result.12   11  Harding continues to develop the idea of “thinking from the lives of the marginalized,” considering the many different marginalized positions that may not be reflected in the direction of scientific research. More recently she has focused her attention on the marginalized positions of those in the global south and their relationship to discourses of modernity (2008).   12  For a discussion of the many complications that come with how to achieve the benefits of diversity within academe, see Fehr (2011).

Epistemic Normativity in Feminist Epistemology  227 For Longino, the results of a scientific community are objective—and, thus, epistemically praiseworthy—to the extent that they have been subject to certain procedural conditions ensuring that the claims have been subjected to transformative criticism.13 Recognizing that these practices of transformative criticism occur within the confines of a research community and that people bring to the table different background assumptions (some connected to their social situatedness), Longino argues that maximally transformative criticism can only occur within a research community that has robust social diversity within its ranks. Harding and Longino are both working with the situated knowing thesis as they undertake the normative work of describing an epistemically appropriate stance that can emerge from social situatedness. The objective results that Longino describes will still offer only a limited take on the world, as will the results of Harding’s endeavor to incorporate the perspectives of the marginalized into research. The results will not represent a view-from-nowhere. However, they will represent more than any singular socially situated perspective on the world, and they will be attuned to the epistemic goals of the communities of inquirers. The above serve as examples of the normative work being done by feminists that is premised on the situated knowing thesis. In what follows, I explain in more detail four identifiable features of the normative work that feminist epistemologists engage in. I articulate how these features stem from this minimal interpretation of the thesis of socially situated knowing, and I examine similarities and contrasts with some other recent developments in epistemology that challenge parts of the traditional framework.

12.3.1  Prioritizing the Activities of Knowledge-seeking As Catherine Elgin notes, it is a mistake to understand competing epistemological theories as simply providing different answers to the same questions. Rather, different theorists have understood the core questions of epistemology differently from each other. As she writes, “to view (epistemological theories) as supplying alternative answers to the same questions is an oversimplification. For they embody disagreements about what the real questions are and what counts as answering them” (1996: 3). Elgin’s insight supports an interpretation of the debates between feminist epistemologists and their critics according to which at least some of their disagreements are a result of a difference in priorities concerning which epistemological questions need answering. Earlier I noted that epistemologists working within a traditional framework have commonly focused on the normative question of what constitutes knowledge, understood as a success term, only then answering other normative questions, such as how we should go about knowing, in light of that analysis. For example, in Matthias Steup’s

  13  Longino’s (2002) conditions include that there be publicly recognized forums for criticism, uptake of criticism, publicly recognized standards of evidence, and tempered equality of intellectual authority.

228  Heidi Grasswick entry on epistemology in The Stanford Encyclopedia of Philosophy, he describes epistemology as follows: Defined narrowly, epistemology is the study of knowledge and justified belief. As the study of knowledge, epistemology is concerned with the following questions: What are the necessary and sufficient conditions of knowledge? What are its sources? What is its structure, and what are its limits? As the study of justified belief, epistemology aims to answer questions such as: How we are to understand the concept of justification? What makes justified beliefs justified? Is justification internal or external to one’s own mind? (Steup 2012)

The questions Steup characterizes as central and defining of the field of epistemology all concern the analysis of the concepts of knowledge and justification themselves, with essentially no mention of questions concerning how we attain knowledge. Although we can imagine that such questions as the sources and structure of knowledge cannot be fully answered without attending to how we go about seeking knowledge, it is remarkable that this description of epistemology is structured in such a way as to omit any explicit mention of core questions concerning inquiry into the activities that would constitute knowing well. In contrast, the situated approach of feminists is far more pragmatic in orientation, prioritizing normative questions concerning how we are to engage in the activities of knowing and inquiring well from within our particular situations rather than more abstract questions concerning definitions and adequate formulations of knowledge and justification.14 The primary normative work for feminist epistemologists is the provision of guidance for individuals and communities to undertake and improve the on-the-ground practices and activities of knowledge-seeking. This type of normative endeavor falls under what Earl Conee (1998) calls “advisory epistemology”. For feminists, it is because the situated knowing thesis understands knowing as necessarily occurring within embodied social situations that they consider their primary normative work to consist in the provision of practical guidance for knowing activities within such situations. Furthermore, the limited and socially shaped nature of knowing means that questions concerning practical guidance cannot be adequately answered simply by the application of an abstract and general conception of knowledge such as many epistemologists working within the traditional framework have sought to provide. More needs to be encompassed in the feminist analysis. Of course, how we know and what knowledge are not issues that can be fully isolated from one another, and I do not mean to suggest that feminist epistemologists can or

  14  Feminists’ pragmatic focus makes for strong connections with developments in pragmatist epistemology (Siegfried 1996; Rooney 2012). I do not discuss these connections in detail here only because I have chosen to draw attention to feminist connections with those recent developments in epistemology that are now fairly well integrated into the discussions of “mainstream” epistemology—e.g., naturalism, contextualism, and virtue epistemology. Relatedly, I do not discuss feminist epistemology explicitly in relation to social epistemology, since I opened this chapter with a conception of feminist epistemology that placed it squarely within the field of social epistemology.

Epistemic Normativity in Feminist Epistemology  229 do ignore completely issues of what constitutes knowledge and justification. But feminists are more likely to answer questions concerning knowledge and justification in light of their answers concerning our activities of knowledge-seeking, rather than the other way around. This approach can be seen explicitly in Longino’s response to Philip Kitcher’s criticisms suggesting that her approach fails to provide a concept of knowledge that will do the normative work we need of distinguishing between knowledge and social acceptance of a claim (Kitcher 1991, 1994; Longino 1994, 2002). Longino’s procedurally-based conception of objectivity is clearly normative in recommending that epistemic communities engage in social practices that encourage transformative criticism of the theories and claims put forward.15 For her, the primary normative question of epistemology concerns what kind of epistemic practices we ought to engage in, not what an adequate definition of knowledge is. When pressed by Kitcher and his worries that her practice-based conception fails to capture adequately the normative distinction required to be able to distinguish between knowledge and social acceptance of a claim, Longino complies by providing an explicit account of knowledge as product; but it is an account that understands knowledge as the result of a particular set of practices that demand transformative criticism. Without outlining her entire account of knowledge here, it will suffice to note that Longino acknowledges that her definitions result in relativizing “various concepts of knowledge to communities, but those communities themselves satisfy certain conditions in order that the cognitive activities occurring within them qualify as knowledge or as knowledge-producing” (Longino 2002: 138). What counts as knowledge will be the outcome of certain communal practices that themselves satisfy certain conditions. The priority placed on practical normative guidance for the activities of knowing also explains why work in feminist epistemology tends to involve a good deal of description of our practices of knowing, something for which they have been criticized by traditionalists. According to feminists, sound normative guidance within social situations of knowing can only be ascertained through careful study of these practices and how they are doing in terms of reaching our situated epistemic goals. Feminist epistemologists are not alone in arguing that careful attention needs to be paid to describing our knowing practices. Naturalistic epistemologies are also known for emphasizing the descriptive component of epistemology, though with marked differences amongst them concerning the degree of normativity they consider themselves able to provide and the degree to which knowledge production is considered social. Hilary Kornblith defines a naturalistic approach to epistemology as consisting in the belief that the question of how we ought to arrive at our beliefs cannot be answered independently from the question of how we do arrive at our beliefs; that is to say, “descriptive questions about belief acquisition have an important bearing on normative questions about   15  Although Longino’s work focuses on scientific communities, she understands her conception of objectivity to apply to any epistemic community, whether its domain be the natural sciences or some quite different epistemic pursuit, such as art criticism (1990).

230  Heidi Grasswick belief acquisition” (1994: 3). Given this criterion, it is no surprise that many feminist epistemologists who adopt the situated knowing thesis identify as naturalized epistemologists (Nelson 1990; Code 1996; Longino 2002). Earlier I discussed how the intensive description of knowing practices evident in feminist epistemology has given rise to criticisms to the effect that the feminists fail to provide a normative epistemology, engaging instead simply in sociological descriptions of knowledge production. But to a large extent these criticisms simply raise the question of whether or not some forms of naturalized epistemology and feminist epistemology, with their attention to descriptions of our knowing practices and activities, could be capable of providing normative accounts. The fact that their analysis attends to descriptive work is itself insufficient to dismiss them from the realm of normative epistemologies. In fact, the descriptive work of feminist epistemology is important for their normative goals of offering guidance for agents who must negotiate through present knowledge-seeking practices at the same time as envisioning how to improve them to better achieve our cognitive goals.16

12.3.2  Contextualizing Epistemic Norms The feminist thesis of situated knowing also has implications concerning the scope of any epistemic norms generated. A situated knower, as opposed to an abstract knower, engages in inquiry of a particular kind, in a specific context, and from a particular location. If our normative questions are centered on how situated knowers can know well, those questions will require contextually specific answers, with attention to the particularities of our epistemic goals in the given context. On a situated approach, there is no reason to suppose, or more strongly require, a set of universal norms for how to inquire well across all contexts since it is recognized that any such set would be unable to do all the work required to generate epistemic recommendations in particular contexts. In this regard, the situated approach differs markedly from the traditional framework that seeks just such a universal set generated through the provision of a general and abstract conception of knowledge that would hold across contexts. On a situated approach, there will be different recommendations offered for different contexts and types of inquiry, given the limited nature of any knowledge, and for differently located knowers. For example, according to the situated knowing thesis, gender will be epistemically relevant only in contexts where social relations are structured along the lines of gender. If our society shifts such that it is less systemically organized along the lines of gender, then gender would become a less interesting category epistemically, and recommendations for how to know well would be different in such a context.17   16  Feminist moral epistemologist Kathryn Pyne Addelson has articulated the point in terms of the need for an epistemology “for knowledge makers.” An epistemology for knowledge makers will require intensive description of how knowledge practices are functioning in order to provide guidance for those knowledge makers. In contrast, she criticizes the traditional epistemology of the Anglo-American canon as failing in this task, instead seeming to be a “theory about knowledge itself ” (Addelson 1993: 265).   17  Cassandra Pinnick criticizes feminist standpoint theory exactly along these lines, claiming that it is problematic for the position of standpoint theory that “if feminists achieve political equity, they would

Epistemic Normativity in Feminist Epistemology  231 Additionally, because the situated knowing thesis suggests that knowers are differently situated, it cannot be assumed that the best way to know will be the same across social locations. One’s positionality may make a difference to how one ought to proceed. For example, if one’s social location strongly limits one’s capacity for achieving a certain kind of knowledge from that location, one may need to rely on the testimony of a differently situated person, who has greater access in virtue of their social location. The recommended epistemic norms should reflect this. Furthermore, if knowers are differentiated according to social location, we cannot assume that the same kinds or instances of knowledge will be of equal concern to all variously located people, and certain knowledge practices may serve the epistemic interests of some more than others. As such, feminist epistemologists have found it epistemologically relevant to ask “knowledge for whom?” (Code 1991; Harding 1991). Recognizing that different kinds of knowledge may carry a differential importance for various situated knowers opens the door for normative considerations of what kind of knowledge we ought to be engaged in seeking communally, a point I return to in the next section. Identifying the contextual sensitivity and the limitations of scope that confine any articulation of epistemic norms on a situated approach also helps explain the prominence of work in feminist epistemology that focuses on specific epistemic situations that they view as either particularly important or as neglected in mainstream epistemological discussions. For feminist epistemologists, one of the defining features of these contexts of knowing is the social relations of power in which knowers are implicated and through which they must negotiate their way as they seek to know well. As such, a large part of their work has focused on determining how people can know well within contexts of oppression. Feminists have investigated the specific epistemic challenges that arise in such situations and suggested contextually-specific remedies. For example, feminist virtue epistemologist Nancy Daukas has argued that within contexts of oppression in which members of some groups are unjustly accorded less credibility than they deserve due to social prejudices, virtuous knowers must strive to exhibit a kind of “oppositional agency” (Daukas 2011). This oppositional agency actively resists the social pressures to assign credibility according to social prejudices. A very similar idea can be found in the work of Miranda Fricker. Fricker characterizes situations under oppression where speakers suffer a credibility deficit due to their social identity as cases of “testimonial injustice.” As a corrective to such unjust credibility assignments, she develops the idea of a virtue of “testimonial justice.” Testimonial justice for Fricker is a virtue that requires that agents exercise reflexive critical social awareness in their role as hearers of testimony (2007: 91).

thereby lose any claim to epistemic privilege” (Pinnick 2003: 29). But this fails to recognize how the situated knowing thesis operates in feminist epistemology, by demanding that such analysis of epistemic privilege would have to proceed relative to the social context. Rather than offering a critique, Pinnick has emphasized exactly what should be expected from feminist epistemology.

232  Heidi Grasswick Of course, acknowledging the limited scope of any epistemic norms offered on a situated approach does not yet answer the question of what the appropriate boundaries are within which appropriate epistemic norms hold steady. Not surprisingly, we also find within feminist epistemology explicit discussions concerning how to determine the appropriate boundaries of the epistemic context within which specific normative claims can be made. For example, Alison Wylie has argued that the tenants of feminist standpoint theory according to which it is claimed that some socially situated groups carry a certain epistemic privilege in virtue of their social position needs to be contextualized. According to Wylie, epistemic privilege will hold for certain social groups with regard to some areas of knowledge, but not likely all. This claim diffuses critics of feminist epistemology who demand that for social situation to be considered epistemically relevant, it must be proven to be so across all forms of knowing, with the most favored demand being a demonstration of the relevance of social situation to knowledge of physics, or to plain perceptual knowledge of physical objects. On Wylie’s analysis, it will be a matter of empirical investigation to determine in which contexts we need to attend more closely to social situation than others in considering how to know well (Wylie 2003). It is against the backdrop of an acceptance of contextualized epistemic norms that feminists are able to address such challenges posed by their critics concerning the relevance of their accounts. It is worth noting that the contextualization of epistemic norms feminists engage in suggests similarities with some of the recent developments of contextualism in mainstream contemporary epistemology. The projects share a rejection of the viability of a universal approach to epistemic norms. However, the focus of their endeavors is in fact quite different. Contextualism within mainstream epistemology has been dominated by semantic contextualism, through which epistemologists have attempted to defeat the skeptic by providing analyses that focus on the relevance of context for a determination of the truth of an “S knows that p” claim (see for example Cohen 1986; DeRose 1992; Williams 2001; Neta 2003). Semantic contextualism, though it rejects the invariantism of traditionalists, remains very much focused on the questions that have defined the traditional framework, such as the articulation of necessary and sufficient conditions of knowledge. In this, it differs substantially from the contextualism of a feminist situated approach that focuses on context-sensitive recommendations for how to know well, with less attention paid to the concept of knowledge itself and its attributions, and almost no attention paid to worries regarding skepticism.18 These differences result in surprisingly little overlap in the projects of feminist epistemologists versus the contextualists discussed in mainstream epistemology. Parallels can be drawn in terms of how they both take the pragmatics of our epistemic situations seriously in assessing our epistemic judgments, yet the overall direction of their projects remain quite distinct.   18  For an interesting use of Michael Williams’ contextualist theory of epistemic justification applied to feminist epistemology, see Rolin (2011).

Epistemic Normativity in Feminist Epistemology  233

12.3.3  The Entanglement of the Epistemic and the Ethical As I noted earlier, one of the more pressing sources of the conflict between feminist epistemologists and their critics on the issue of normativity concerns the relationship between the epistemic and the ethical. We are now in a position to see how this conflict emerges. The thesis of situated knowing conflicts with the traditional epistemologists’ demand that the normative dimension of the epistemic realm be isolated from the ethical in a few different though related ways. To begin, the work of feminist epistemologists has consistently argued that as we turn our attention to the activities of knowing, recognizing that knowing is something we do, and more importantly, something that we do within social situations, the ethical dimension to our knowing becomes not only apparent but crucial to include our epistemic analyses (Code 1991). As I have argued, the situated knowing thesis leads to a prioritization of normative questions concerning the activities of knowing and inquiring well rather than focusing on assessments of the formation and status of beliefs. With activities and practices under assessment, ethical questions of responsibility arise more directly than with assessments that are restricted to the formation and status of beliefs. Epistemic assessments of what constitutes the practices and activities of knowing well form a subset of ethical assessments of how to live well and need to be informed by broader understandings of what it is to live well. In this regard, feminist epistemologists have an affinity with the approach of many virtue epistemologists of the responsibilist variety who seek a broader understanding of the role of our epistemic undertakings in our ethical lives overall. For now though, I leave further discussion of feminists’ connections with virtue epistemologists until the next section. The situated knowing thesis also emphasizes that knowers are socially differentiated. Accordingly, we are deeply epistemically interdependent. We must rely on those who are differently situated for insights that we cannot obtain on our own, and which need to be integrated with our own perspectives. As a result, relationships and interactions between knowers form an integral part of epistemic inquiries. Such social interactions necessarily have an ethical dimension to them, and the ethical quality of our relationships with others will in many cases affect our possibilities for epistemic success in the long run. For example, access to certain kinds of knowledge may require lasting relationships of trust. In this regard, some feminists have argued that in contexts of oppression, knowledge-sharing between positions of oppression and privilege can be extremely difficult due to histories of distrust between groups, even in some circumstances requiring ethically sound relationships of friendship between parties in order for such knowledge-sharing to occur (Lugones 2003). Many feminists have also argued that our knowledge practices are damaged by a lack of social justice that permits the perpetuation of biases within our epistemic pursuits. For example, Janet Kourany argues that our society needs improvements in social justice so that sexist and androcentric biases do not continue to weaken our scientific practices (Kourany 2010). For her, without social justice in society as a whole, our institutions of science

234  Heidi Grasswick will be unable to thrive epistemically. Others focus on epistemic problems that arise in our everyday knowing pursuits, such as our simple reliance on the testimony of others. For example, Fricker argues that the offence of testimonial injustice that occurs when members of socially marginalized groups are attributed less credibility than they deserve due to social prejudices is at once epistemic and ethical. It is an injustice to the speaker, at the same time as it is problematic in impeding the abilities of both hearer and speaker to know well (Fricker 2007).19 Finally, I have made the point that the situated knowing thesis implies a contextualization of our analyses of knowledge-seeking endeavors, and this too has implications for understanding the relationship between the epistemic and the ethical. If knowing is always situated and limited, there is no single form of knowing and inquiry that can be held up as the gold standard for epistemic norms concerning recommended practices and activities. Epistemic assessments must be made within particular contexts, and our epistemic goals may differ slightly according to those contexts. Different epistemic standards may be appropriate for achieving different kinds of knowledge. Ethical considerations will come into play in selecting the type of epistemic inquiries we engage in and the kind of knowledge we ought to be pursuing. If epistemic standards do not hold across types of epistemic inquiry, then the ethical considerations that help determine which inquiries need to be undertaken in turn give shape to the epistemic standards themselves. One clear way of describing this link is in terms of the position taken up by many epistemologists and philosophers of science that our epistemic goals are best understood in terms of our striving for significant truths, not just truths per se (Kitcher 1993; Anderson 1995b). If it is significant truths that we are after, assessments need to be made concerning significance: what are the important things we need to know about in our lives generally, and in specific contexts more particularly? Ethical considerations clearly play a role here in determining what kinds of knowing are important for living well. The need to assess significance is exacerbated by the situated knowing thesis: differently situated knowers may not share similar needs or interests relevant to the type of knowledge or inquiry to be undertaken. Certain kinds of knowledge may be of greater concern from the vantage points of some social locations. For example, sometimes the pressing practical needs of a specific community will demand timely solutions based on more cursory understandings of the causal links of a social or health phenomenon rather than prolonged scientific studies. If knowers are differently situated, and significance is included as a criterion for evaluations of epistemic practices, we must consider for whom the knowledge is significant (Grasswick 2010). Ultimately, this will bring

  19  Though Fricker points out that there is an epistemic loss to both parties, she maintains that the injustice itself is done to the speaker, because it is she who is not being taken seriously as a knower. For Fricker (2007), along with other virtue theorists, being taken seriously as a knower is a central component of a meaningful and ethical life.

Epistemic Normativity in Feminist Epistemology  235 into play ethical issues concerning what kind of knowledge our communities ought to pursue, in order to ensure that our knowledge practices support the necessary inquiries for variously situated agents to be able to live well in part through knowing well. For traditional epistemologists, our choice of knowledge pursuits may be an important ethical issue, but is considered irrelevant from the epistemological point of view precisely because epistemological analysis is narrowly conceptualized as only dealing with the status and features of a claim that qualify it as knowledge, with these features most prominently understood in terms of their truth-conduciveness. Ernest Sosa, for example, has distinguished between “theory of knowledge,” that focuses on the nature, conditions and extent of human knowledge, and “intellectual ethics,” a broader inquiry that “concerns evaluation and norms pertinent to intellectual matters generally, with sensitivity to the full span of intellectual values” (Sosa 2007: 89), and he characterizes the history of epistemology as focused on theory of knowledge. However, on a situated approach, the characterization of knowing always being limited and perspectival challenges the viability of any general characterization of knowledge that this separation of the epistemological and the choice of knowledge pursuit depends upon. The feminist project may be better understood as part of the broader project of intellectual ethics, but the project also questions whether the narrowly defined project of “theory of knowledge” can deliver on the kind of normative tasks feminists believe are important, given their situated approach. Understanding that knowledge is always limited and is developed in materially situated contexts means that not all knowledge is possible: decisions to pursue particular lines of knowledge are simultaneously decisions not to pursue others. So far, one could still maintain a separation of an ethical judgment on the decision to pursue a particular line of knowledge from its epistemic assessment. However, when coupled with the feminist contention that epistemic norms concerning our activities of knowledge-seeking are not the same across types of knowledge pursuits, we reach the conclusion that on a situated approach, the choice of knowledge pursuits has both epistemological and ethical dimensions. An understanding of the appropriate epistemic norms to apply cannot be achieved without consideration of the type of knowledge to be pursued. According to a situated approach, the epistemic normative task consisting in recommendations for epistemic activities and practices cannot be completed in isolation from considerations of the ethical. Evidence of how feminists have stressed the need to consider practices of knowledge production in terms of whose knowledge interests are being served—that is, whose significant truths are being prioritized—can be seen in their contributions to the development of epistemologies of ignorance. Epistemologies of ignorance examine the links between social relations of power and areas of ignorance, arguing that many areas of ignorance are not benign gaps in knowledge but rather are socially constructed ignorances and exist as a result of a particular set of power relations at work (Tuana and Sullivan 2006).20 For example, the fact that little was known about  

  In addition to this special issue of Hypatia devoted to feminist epistemologies of ignorance, see also see

20

236  Heidi Grasswick the differences in female, versus male, symptoms of heart disease and the disease’s significance as a cause of death among women was underappreciated until fairly recently is not a coincidence. Rather, it reflects a biomedical field operating within a society of unequal gender power relations (Grasswick 2010). It is evidence that the practices of generating biomedical knowledge were doing a poorer job than once thought in producing pockets of knowledge significant for women. Another example of feminist work that has attended to patterns of knowledge and ignorance is Fricker’s development of the concept of hermeneutical injustice. Fricker argues that in contexts of oppression, society often lacks the hermeneutical resources required to articulate and understand the experiences of oppressed groups. It marks an injustice because such groups lack the necessary hermeneutical resources for understanding experiences that are particularly significant for them, whereas culturally dominant groups are rarely faced with such a conceptual deficit. She uses as an illustration the evolution of the concept “sexual harassment,” explaining how before this concept came into circulation, women (and society as a whole) lacked the hermeneutical resources needed to understand the experiences victims of (what we now conceptualize as) sexual harassment were suffering (Fricker 2007). With respect to the relationship between the ethical and the epistemic then, feminist epistemologists differ from their critics, many who come to the question from within the traditional framework of epistemology. Feminists are not saying that ethical considerations necessarily trump the epistemic, nor are they saying that the two cannot be understood as conceptually distinct. However, they are resisting a narrow interpretation of the epistemic that would bar ethical considerations from playing any role in epistemic assessments. For feminists taking up a situated approach, the normative tasks they take to be most important for epistemology—offering guidance for the practices and activities of knowing within particular social contexts to help achieve epistemic goals—cannot be fulfilled without including some ethical considerations. Feminist epistemologists are questioning our ability to answer epistemically normative questions without reference to the ethical. The traditional attempt to restrict the epistemological to norms concerning truth conducive belief-formation will, according to feminists, be woefully inadequate in providing the kind of normative recommendations that they believe we need to be able to know well from within our social situations.

12.3.4  Attention to the Perspective of Epistemic Agents The final feature of epistemic normativity stemming from a situated approach is an attention to the internal perspective of agents of knowing, similar to what we find in responsibilist forms of virtue epistemology. The situated knowing thesis claims that knowing necessarily occurs from a located perspective, so its primary normative Sullivan and Tuana’s (2007) collection Race and Epistemologies of Ignorance and Proctor and Schiebinger’s (2008) collection Agnotology: The Making and Unmaking of Ignorance.

Epistemic Normativity in Feminist Epistemology  237 questions will be framed in terms of what agents can do to know well from particular situated locations. According to a situated approach, there is no view-from-nowhere outside of these situated locations from which knowledge claims could be independently assessed. The primary normative work for epistemology then becomes how to evaluate and to recommend habits and practices of situated agents that allow them to achieve their cognitive aims, rather than offer assessments of the status of beliefs themselves. This orientation explains why there have been few feminist contributions to the internalist/externalist debates within epistemology. Those who adopt a situated approach and focus normative questions on what knowers do in their inquiries take it as a given that epistemology needs to concern itself with the perspectives of the knowing agents. In this, feminist epistemologists bear similarities to virtue responsibilists who focus their normative attention on the virtuous characteristics of the epistemic agents and their practices.21 Virtue responsibilists are particularly interested in character traits that might be capable of being fostered in agents, with standard examples including conscientiousness, intellectual courage, and open-mindedness (see, for example, Montmarquet 1993; Roberts and Wood 2007) as opposed to the reliabilist virtue epistemologists’ focus on epistemic faculties, such as perception and memory, that may be less under an agent’s control. Feminists adopting a situated approach share the responsibilists’ focus on traits that can be fostered in agents that help them engage in practices of good knowing as part of living well. But given their attention to differences in socially located knowers, feminists are also more likely to emphasize social epistemic virtues such as an epistemic trustworthiness that can be found in an attunement to one’s own and others’ epistemic strengths and weaknesses (Daukas 2011). Such social virtues allow us to do well interacting with other epistemic agents and negotiating through the situated perspectives that feminists have argued form a central feature of our epistemic predicament. Additionally, feminists’ appreciation of the ways in which our knowing practices and institutions are structured by social systems of oppression lead some to consider expanding the idea of epistemic virtues to our social systems and institutions (Anderson 2012). In taking up the perspective of epistemic agents and focusing on what they can do to know well, feminist epistemologists and responsibilist virtue epistemologists alike are committed to a clear sense of agency: human beings are the kind of beings who can actively make decisions that can contribute to their knowledge-seeking endeavors. Some kind of agency is of course required if epistemology is to ultimately have any normative force. For an “ought” to have any power, there must be an ability in the agent that allows the ought to be carried out. In the case of feminist epistemologists who

  21  Code is one feminist epistemologist who has been clearly identified within the responsibilist camp of virtue epistemology. Her virtue approach and focus on epistemic responsibility are evident throughout all of her work, though interestingly it is her early work (1987) that lacks any explicitly feminist dimension that is usually cited in virtue epistemology venues.

238  Heidi Grasswick adopt a situated approach, this agency will be a socially embedded agency. In order to provide recommendations of habits and practices, feminists must maintain a sense of agency through which knowers are able to critically reflect upon their situated perspectives and shape their habits and practices. Much of the work of feminist epistemology focuses on explaining how epistemic agents can do this. For example, Harding’s (1991) work has called on researchers to engage in “strong reflexivity,” through which they consider the socially situated nature of their inquiry and the relationship between the perspective they bring to it and other socially situated perspectives. Though feminist epistemologists who draw on a situated approach all work with a kind of socially embedded agency, there are significant variations on the specifics of their conception of agency. For example, some take the social nature of knowing to suggest a conception of communities as epistemic agents (Nelson 1990, 1993), while others focus on individuals-in-communities (Grasswick 2004)  assessing how individuals can act well as epistemic agents within their communities and given social structures.22 But regardless of these differences, the situated knowing thesis directs feminists to focus their normative work on the perspectives of the agents themselves, specifically located as they are. Finally, though I have offered an explanation of how the situated knowing thesis itself leads feminist epistemologists to focus on the perspective of epistemic agents in their normative work, there is also another reason why feminist epistemologists are drawn to such an approach. The underlying ethico-political projects of feminists contribute to a conception of individual subjects who are shaped by their social situations, but nevertheless are still capable of acting as reflective epistemic agents who interact with each other and the world. Feminist commitments to progressive social change and transformative justice require that they maintain the possibility of human actors generating such change, with ramifications for their epistemological approach. For example, Daukas argues that the agent-centered versions of virtue epistemologies— that is, the virtue theories of the responsibilist form—are particularly useful tools for feminist epistemologists given the need for feminists to maintain a sense of political and epistemic agency of individual subjects (Daukas 2011). As Daukas notes, there is a commitment within feminist epistemology to “engage the point of view of epistemic agency, conceiving of agents in their particularity, as embodied and politically ‘situated’ ” (2011: 48).

12.4  Situating Feminist Epistemology I began this chapter by juxtaposing the clear normative intent of feminist epistemological projects with critiques that argue instead that feminists have failed to provide   22  Someone like Longino (1990, 2002) is harder to place, since she is clear that the norms that apply are norms that a community must adhere to, yet her work also attends to the importance contributions of critical dialogue that come from the individuals within epistemic communities.

Epistemic Normativity in Feminist Epistemology  239 normative epistemologies. My discussion has sought to clarify these disagreements by revealing some of the key differences in how feminists versus those engaged in the traditional framework conceptualize the normative tasks and demands of epistemology. The normative work undertaken in feminist epistemology prioritizes the activities of knowledge-seeking, contextualizes epistemic norms, conceptualizes the epistemic and the ethical as deeply entangled, and attends to the perspective of epistemic agents. These four features contrast with the traditional framework of epistemology that prioritizes normative assessments of the status of beliefs and focuses on the development of an abstract and acontextual analysis of the conditions of knowledge from which universal norms can be derived and within which ethical considerations play no part. It is noteworthy that the four features of normativity in feminist epistemology that I have articulated also help identify the ways in which feminist epistemology shares some insights with several other contemporary movements in epistemology: naturalism, contextualism, and virtue theory. Feminists are not alone in challenging some of the features of the traditionalists’ conception of epistemic normativity. But the conjunction of the four features I identified in feminist epistemology suggests specific directions of inquiry within these recent epistemological movements that feminists have developed and will continue to develop. We can expect that feminists will “do” naturalism in a particular way—a way that is contextual and maintains an attention to epistemic agency while providing recommendations for improved epistemic practices. Similarly we can expect that they will “do” contextualism in a particular way—incorporating ethical dimensions of epistemic contexts into their analyses, and focusing on contextual analyses of the activities and practices of knowing rather than analyzing the concept of knowledge itself. Finally, they will “do” virtue theory in a particular way—with an agent-centered responsibilist approach that emphasizes social virtues and takes seriously the contextual constraints on the exercise of our epistemic virtues. Of course, there are many tasks I have not undertaken in this chapter. I have not offered a full explanation of the source of normativity in epistemology,23 nor have I specifically argued for the advantages of the feminist approach to epistemic normativity over some more traditional approaches, or vice versa. But if discussions between feminist epistemologists and their critics are going to proceed productively, efforts must be made to locate their differences precisely in order to avoid talking at cross purposes. One of those differences concerns understandings of what is expected from the normativity of epistemology.

References Addelson, K. P. (1993). “Knowers/Doers and Their Moral Problems,” in L. Alcoff and E. Potter (Eds.), Feminist Epistemologies. New York: Routledge, 265–94.   23  One must look to recent work on epistemic value for answers here. See for example Haddock, Millar, and Pritchard (2009).

240  Heidi Grasswick Alcoff, L. and E. Potter. (1993). “Introduction:  When Feminisms Intersect Epistemology,” L. Alcoff and E. Potter (Eds.), Feminist Epistemologies. New York: Routledge, 1–14. Anderson, E. (1995a). “Feminist Epistemology:  An Interpretation and a Defense,” Hypatia 10(3): 50–84. Anderson, E. (1995b). “Knowledge, Human Interests, and Objectivity in Feminist Epistemology,” Philosophical Topics 23(2): 27–58. Anderson, E. (2008). “How Not to Criticize Feminist Epistemology: A Review of Scrutinizing Feminist Epistemology,” URL = . Anderson, E. (2011). “Feminist Epistemology and Philosophy of Science,” Stanford Encyclopedia of Philosophy (Spring 2011 edition), E.  N. Zalta (Ed.), URL = . Anderson, E. (2012). “Epistemic Justice as a Virtue of Social Institutions,” Social Epistemology: A Journal of Knoweldge, Culture and Policy 26(2): 163–73. Bar On, B. A. (1993). “Marginality and Epistemic Privilege,” L. Alcoff and E. Potter (Eds.), Feminist Epistemologies. New York: Routledge, 83–100. Bleier, R. (1984). Science and Gender. New York: Pergamon Press. Chisholm, R. (1989). Theory of Knowledge, 3rd edn. Englewood Cliffs, NJ: Prentice-Hall. Code, L. (1987). Epistemic Responsibility. Hanover: University of New England Press. Code, L. (1991). What Can She Know? Feminist Theory and Construction of Knowledge. Ithaca, NY: Cornell University Press. Code, L. (1996). “What is Natural about Epistemology Naturalized?” American Philosophical Quarterly 33(1): 1–22. Code, L. (2006). Ecological Thinking:  The Politics of Epistemic Location. Oxford:  Oxford University Press. Cohen, S. (1986). “Knowledge and Context,” Journal of Philosophy 83: 574–83. Conee, E. (1998). “Normative Epistemology,” in E. Craig (Ed.), Routledge Encyclopedia of Philosophy. London: Routledge, URL = . Daukas, N. (2011). “Altogether Now:  A  Virtue-theoretic Approach to Pluralism in Feminist Epistemology,” in H. Grasswick (Ed.), Feminist Epistemology and Philosophy of Science: Power in Knowledge. Dordrecht: Springer, 45–68. DeRose, K. (1992). “Contextualism and Knowledge Attributions,” Philosophy and ­Phenom­enological Research 52(4): 913–29. Elgin, C. Z. E. (1996). Considered Judgment. Princeton: Princeton University Press. Fehr, C. (2011). “What Is in It for Me? The Benefits of Diversity in Scientific Communities,” in H. Grasswick (Ed.), Feminist Epistemology and Philosophy of Science: Power in Knowledge. Dordrecht: Springer, 133–56. Fricker, M. (2007). Epistemic Injustice:  Power and Ethics in Knowing. Oxford:  Oxford University Press. Goldman, A. I. (1999). Knowledge in a Social World. Oxford: Clarendon Press. Grasswick, H. (2004). “Individuals-in-Communities:  The Search for a Feminist Model of Epistemic Subjects,” Hypatia 19(3): 85–120.

Epistemic Normativity in Feminist Epistemology  241 Grasswick, H. (2006). “Feminist Social Epistemology,” The Stanford Encyclopedia of Phil­osophy, E. N. Zalta, URL = Grasswick, H. (2010). “Scientific and Lay Communities:  Earning Epistemic Trust through Knowledge Sharing,” Synthese 177: 387–409. Grasswick, H. (2011). “Introduction:  Feminist Epistemology and Philosophy of Science in the Twenty-First Century,” in H. Grasswick (Ed.), Feminist Epistemology and Philosophy of Science: Power in Knowledge. Dordrecht: Springer. Haack, S. (1998). Manifesto of a Passionate Moderate. Chicago: University of Chicago. Haddock, A., A. Miller, et al. (Eds.) (2009). Epistemic Value. Oxford: Oxford University Press. Haraway, D. (1988). “Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective,” Feminist Studies 14: 575–99. Haraway, D. (1989). Primate Visions: Gender, Race and Nature in the World of Modern Science. New York: Routledge. Harding, S. (1991). Whose Science? Whose Knowledge? Thinking from Women’s Lives. Ithaca, NY: Cornell University Press. Harding, S. (2008). Sciences from Below: Feminisms, Postcolonialities, and Modernities. Durham: Duke University Press. Harding, S. (2009). “Standpoint Theories: Productively Controversial,” Hypatia 24(4): 192–200. Hartsock, N. (1983). “The Feminist Standpoint:  Developing the Ground for a Specifically Feminist Historical Materialism,” in S. Harding and M. Hintikka (Eds.), Discovering Reality: Feminist Perspectives on Epistemology, Metaphysics, Methodology, and Philosophy of Science. Dordrecht: D. Riedel Publishing, 283–310. Haslanger, S. (1999). “What knowledge Is and What It Ought to Be:  Feminist Values and Normative Epistemology,” Philosophical Perspectives 13: 459–80. Hubbard, R. (1983). “Have Only Men Evolved?” in S. Harding and M. Hintikka (Eds.), Discovering Reality:  Feminist Perspectives on Epistemology, Metaphysics, Methodology, and Philosophy of Science. Dordrecht: D. Reidel, 45–70. Jaggar, A. M. (1983). Feminist Politics and Human Nature. Totowa: Rowman and Allanheld. Kitcher, P. (1991). “Socializing Knowledge,” Journal of Philosophy 88: 675–6. Kitcher, P. (1993). The Advancement of Science:  Science without Legend, Objectivity without Illusions. Oxford: Oxford University Press. Kitcher, P. (1994). “Contrasting Conceptions of Social Epistemology,” in F. F. Schmitt (Ed.), Socializing Epistemology:  The Social Dimensions of Knowledge. Lanham:  Rowman and Littlefield, 111–34. Knowles, J. (2003). Norms, Naturalism and Epistemology: The Case for Science without Norms. New York: Palgrave. Kornblith, H. (1994). “Introduction: What Is Naturalistic Epistemology?” in H. Kornblith (Ed.), Naturalizing Epistemology. Cambridge, MA: MIT Press, 1–14. Kourany, J. A. (2010). Philosophy of Science after Feminism. Oxford: Oxford University Press. Lloyd, L. A. (2005). The Case of the Female Orgasm: Bias in the Science of Evolution. Cambridge MA: Harvard. Longino, H. E. (1990). Science as Social Knowledge: Values and Objectivity in Scientific Inquiry. Princeton, NJ: Princeton University Press.

242  Heidi Grasswick Longino, H. E. (1993). “Subjects, Power, and Knowledge:  Description and Prescription in Feminist Philosophies of Science,” in L. Alcoff and E. Potter (Eds.), Feminist Epistemologies. New York: Routledge, 101–20. Longino, H. E. (1994). “In Search of Feminist Epistemology,” The Monist 77(4): 472–85. Longino, H. E. (1994). “The Fate of Knowledge in Social Theories of Science,” in F. F. Schmitt (Ed.), Socializing Epistemology: The Social Dimensions of Knowledge. Lanham: Rowman and Littlefield, 135–57. Longino, H. E. (1997). “Cognitive and Non-Cognitive Values in Science:  Rethinking the Dichotomy,” in L. H. Nelson and J. Nelson (Eds.), Feminism, Science, and the Philosophy of Science. Dordrecht: Kluwer, 39–58. Longino, H. E. (2002). The Fate of Knowledge. Princeton, NJ: Princeton University Press. Lugones, M. (2003). Pilgrimages/Peregrinajes: Theorizing Coalition against Multiple Oppressions. Lanham: Rowman and Littlefield. Maffie, J. (1996). “Naturalism and the Normativity of Epistemology,” Philosophical Studies 59(3): 333–49. Martin, E. (1991). “The Egg and the Sperm: How Science has Constructed a Romance Based on Stereotypical Male-Female Roles,” Signs 16(3): 485–501. Millar, Alan. (2010). “Knowledge in Recent Epistemology: Some Problems,” in D. Pritchard, A. Millar, and A. Haddock (Eds.), The Nature and Value of Knowledge: Three Investigations. Oxford: Oxford University Press, 97–119. Montmarquet, J. (1993). Epistemic Virtue and Doxastic Responsibility. Lanham, MD: Rowman and Littlefield. Nelson, L. H. (1990). Who Knows: From Quine to a Feminist Empiricism. Philadelphia: Temple University Press. Nelson, L. H. (1993). “Epistemological Communities,” in L. Alcoff and E. Potter (Eds.), Feminist Epistemologies. New York: Routledge, 121–60. Neta, R. (2003). “Skepticism, Contextualism, and Semantic Self-Knowledge,” Philosophy and Phenomenological Research 67(2): 397–411. Pinnick, C. L. (2003). “Feminist Epistemology: Implications for Philosophy of Science,” in C. L. Pinnick, N. Koertge and R. F. Almeder (Eds.), Scrutinizing Feminist Epistemology:  An Examination of Gender in Science. New Brunswick, NJ: Rutgers University Press, 20–30. Proctor, R. N. and L. Schiebinger (Eds.) (2008). Agnotology:  The Making and Unmaking of Ignorance. Stanford, CA: Stanford University Press. Roberts, R. C. and W. J. Wood. (2007). Intellectual Virtues: An Essay in Regulative Epistemology. Oxford: Oxford University Press. Rolin, K. (2011). “Contextualism in Feminist Epistemology and Philosophy of Science,” in H. Grasswick (Ed.), Feminist Epistemology and Philosophy of Science: Power in Knowledge. Dordrecht: Springer, 25–44. Rooney, P. (2012). “What is Distinctive about Feminist Epistemology at 25?” in S. L. Crasnow and A. M. Superson (Eds.), Out from the Shadows: Analytical Feminist contributions to Traditional Philosophy. Oxford: Oxford University Press, 239–376. Schmitt, F. F. (1994). “Socializing Epistemology:  An Introduction,” in F. F. Schmitt (Ed.), Socializing Epistemology:  The Social Dimensions of Knowledge. Lanham:  Rowman and Littlefield, 1–27.

Epistemic Normativity in Feminist Epistemology  243 Siegfried, C. H. (1996). Pragmatism and Feminism:  Reweaving the Social Fabric. Chicago: University of Chicago Press. Sosa, E. (2007). A Virtue Epistemology: Apt Belief and Reflective Knowledge. Oxford: Clarendon Press. Steup, Matthias. (2012). “Epistemology,” The Stanford Encyclopedia of Philosophy (Fall 2012 Edition), Edward N. Zalta (Ed.), URL = . Sullivan, S. and N. Tuana. (2007). Race and Epistemologies of Ignorance. Albany, NY:  State University of New York Press. Tuana, N. and S. Sullivan. (2006). “Introduction:  Feminist Epistemologies of Ignorance,” Hypatia 21(3): 1–3. Williams, M. (2001). Problems of Knowledge:  A  Critical Introduction to Epistemology. Oxford: Oxford University Press. Wylie, A. (2003). “Why Standpoint Matters,” in R. Figueroa and S. Harding (Eds.), Science and Other Cultures: Issues in Philosophies of Science and Technology. New York: Routledge, 26–48.

13 The Commonwealth of Epistemic Ends Catherine Z. Elgin

Neither the deliverances of the Oracle at Delphi nor those of Grandpa’s trick knee are remotely credible. But some sense perceptions, experiences, introspective insights, and/or a priori intuitions apparently carry significant epistemic weight. Many epistemologists hold that some or all of them constitute our epistemic core. Core deliverances, if there are such things, (1) are intrinsically credible in isolation if anything is, and/or (2) interweave into fabrics of commitments that are intrinsically credible if anything is. If the items in the core are basic, nothing epistemologically more fundamental either can or need underwrite them. Not everyone considers all the items on my list basic. Some epistemologists deny that any deliverances are basic, holding rather that justification requires even core deliverances to cohere. Some include deliverances of testimony in the core. However these details work out, items on the list are standardly taken as clear, central cases of epistemically privileged justifiers. No non-core items are deemed necessary to secure either the justification for such deliverances or their capacity to confer justification. Although some hold that core deliverances depend on one another for credibility, it is widely believed that they do not depend on anything outside the core. This is all highly schematic. It glosses over oceans of epistemological disagreement. I present matters so sketchily to highlight and to problematize a pair of widely shared assumptions. First, epistemic individualism: it is states of an individual epistemic agent that constitute her epistemic core. Her perceptions, her experiences, her introspective insights, her a priori intuitions, and/or testimony that she is privy to are supposed to justify her beliefs. Second, attunement: the core deliverances that justify her beliefs do so because they properly attune her to their objects. Perceptions as of red justify because they properly attune her to red things. A priori intuitions about natural numbers justify because they properly attune her to natural numbers. Unless a belief is about other people’s attitudes or actions, it can be wholly justified regardless of what others (apart, perhaps, from those who provide testimony) say or think or do. For it

The Commonwealth of Epistemic Ends  245 is the believer’s relation to the belief ’s content that secures justification. One need not be a foundationalist to take such a position. Commitment to epistemic individualism and proper attunement are endemic in epistemology. Nevertheless, I will argue that these commitments are mistaken. Epistemic success requires that agents be properly attuned not just to the objects of knowledge but also to one another.

13.1  Orwellian Worries George Orwell’s 1984 calls epistemological individualism into question. Perhaps separate bits of core knowledge are justified in isolation. But 1984 provides reason to believe that they are sources of justification only if the epistemic climate is favorable. To an extent, we already knew that. Descartes’s malevolent demon personifies the possibility that conditions are not congenial to knowledge. The malevolent demon is a supernatural complication. Its natural counterpart is the possibility that the salient regularities that obtain in this spatiotemporal region of the cosmos might not be the real laws of nature. Given that possibility, we rely on familiar regularities at our peril. Susceptibility to skeptical challenges discloses vulnerabilities in our epistemic condition. Traditional challenges reveal that epistemic success is possible only if epistemic circumstances are naturally and supernaturally congenial. Then the regularities we take to be laws of nature at least approximate the genuine laws of nature. Our minds are not being manipulated by malevolent demons, nor our brains by mad scientists, and so on. Perhaps, as some think, we can demonstrate that circumstances are congenial; perhaps not. But minimally, they must be congenial for agents to be epistemically successful. Read as an epistemological thought experiment, 1984 discloses a further constraint:  only if epistemic circumstances are socio-politically congenial is epistemic success possible. For Winston Smith to know, or to reasonably believe, that stones are hard,1 he must stand not only in the right relation to stones, but also in the right relation to his fellows. If this is so, epistemic justification depends on social conditions. Deliverances—even core deliverances—are sources of justification only if neither they nor agreement among them is a product of coercion. This holds whether they themselves are separately justified (as foundationalists claim), or their justification emerges from coherence among them (as coherentists claim), or their justification derives from their reliability (as reliabilists claim). For deliverances to yield justification, epistemic agents must be suitably related, not just to the objects of their beliefs, but to one another. The epistemological problems exemplified in the novel are stratified. Issues that seem relatively benign at one level turn out to be problematic at a deeper level. But the deeper we delve, the more extensive and implausible the Party’s interference is.   1  Orwell (1961: 81).

246  Catherine Z. Elgin Eventually we may reach a level where the machinations appear too extreme to disclose anything of epistemological interest. At the limit Orwell’s scenario may be an epistemological reductio. If we balk before reaching the limit it is not. But even if we dismiss the most extreme machinations, the story discloses that uncongenial socio-political conditions can create an epistemologically toxic environment, resulting in something like Vogel’s (1990) semi-skepticism. Epistemic success rests on socio-political underpinnings. Minimally, certain socio-political threats must fail to be realized; maximally, certain socio-political supports must be in place. 1984 describes not just a political dystopia, but an epistemological dystopia. The Party’s machinations corrupt all standard epistemic resources. The Party regularly revises the historical record to align it with the current political agenda and destroys all previous records. Since, the Party declares, Oceania is now at war with Eurasia and has long been at war with Eurasia, all records claiming that Eurasia was recently an ally are mistaken. That being so, they must be corrected and the erroneous accounts erased. The Party leaves no trace of its tampering. So although citizens know that the record can be changed, they have no way to tell whether, in any particular case, the record has been changed, how it has been changed, or how often it has been changed. It is pointless to consult what purport to be historical documents in order to find out what happened, for there is ample reason to doubt that they have any bearing on the facts. News media publish whatever the Party dictates, with no attempt to verify and indeed no means of verifying the information they convey. Was the quarterly output 145 million pair of boots (as previously predicted) or 60 million pair (as the current, ‘corrected’ version reads)? Or were no boots produced at all? In all probability, no one knows. Yet the corrected version is promulgated as fact.2 Within a day, all mention of the original prediction will be expunged from every record. Although so-called ‘written records’ exist, they have, and are widely recognized to have, no stable relation to the truth. They just state what the Party currently wants everyone to think, and that can change abruptly. With history regularly rewritten, and the news media untrustworthy, denizens of Oceania have no documentary evidence against which to check their memories or their current opinions. Spies are everywhere. Deviating from the Party line is punishable by torture, death, and vaporization, where all evidence that a person ever existed is expunged. So it would be suicidal to attempt to check one’s memories or perceptions of current events against those of one’s compatriots. Since most would probably parrot the Party line anyway, the responses of one’s peers would be presumptively unreliable. Given the pressure to conform and the dearth of objective evidence to the contrary, denizens of Oceania might well accede to what the Party tells them without even considering whether it is true. Of course, they might all, like Winston Smith, be seething with epistemological resentment. Without courting death, Winston has no way to find out.

 

2

  Orwell (1961: 41).

The Commonwealth of Epistemic Ends  247 But his compatriots’ overt behavior indicates that their doxastic default position is to acquiesce to whatever the Party tells them. Arguably then, they do not make assertions. Rather than expressing beliefs about what they take to be the case, their utterances articulate the Party line on a given topic. Let us say then that they typically avow, rather than assert. Intersubjective accord thus indicates not that a contention is true or warranted, but that it is politically expedient. Sense experiences are equally problematic. When what Winston apparently sees before his very eyes conflicts with what the Party reports, he has no way to determine whether he is dreaming, hallucinating, or actually perceiving. Did he really see Jones, Aaronson, and Rutland after the Party reported their execution, or did he merely imagine that he saw them? How could he tell? If this were just a single suspect sighting, it would be no more epistemically problematic than Ken’s taking himself to have seen an ivory gull in Gloucester. That the bird he saw was an ivory gull is unlikely given the species’ normal range. Since no other birder was around, he had no way to confirm his sighting. So Ken may never know whether he actually saw the rare gull he took himself to see. But Winston’s situation is different from Ken’s in at least two respects. First, even if he had been surrounded by other observers, he could not have confirmed his sighting; for no one with a healthy sense of self-preservation would dare to admit to seeing the traitors so long as the Party says that they are dead. Second, and more significantly, Winston’s predicament is utterly general. He can glean no genuinely intersubjective confirmation or disconfirmation for anything he takes himself to see. Nor can anyone else. When it comes to reporting observations in Oceania, the relevant difference is not between seeing and not seeing, or seeming to see and not seeming to see, but between what one is supposed to see and what one is supposed not to see. If Winston says that he sees what he is supposed to see, his fellows will corroborate his reports; otherwise they will not—regardless of what passes across anyone’s visual field. In real life, very young children take all their seemingly perceptual deliverances at face value. Initially, they do not realize that dreams are non-veridical. They come to distinguish seeing from dreaming by learning which of their seemingly visual representations admit of intersubjective support.3 Denizens of Oceania lack the reinforcement required to learn to distinguish genuine cases of seeing from impressions that do not reliably correlate with their surroundings. They have no reason to trust their senses. My assessment of Winston’s prospects for perceptual knowledge might appear unduly pessimistic. Perhaps through self-monitoring he could discover that his perceptual deliverances are reliable while his dreams are not. Coherence and consistency would be his guide. Then, without regard to what anyone else says, he would be in a

 

  Williams (2002: 135–40).

3

248  Catherine Z. Elgin position to know what he sees, hears, or whatever. On the face of it, this seems plausible. But things are not so straightforward. A person has no hope of establishing for himself that perception per se is reliable for the simple reason that it is not. We learn from experience that peripheral vision is not as reliable as focal vision; that the apparent shape of the distant tower is not its real shape; that if we do not know how far away an unfamiliar object is, we cannot tell its size just by looking at it; and if we are sufficiently nearsighted or astigmatic, that deliverances of uncorrected vision are not remotely reliable. Nor are the deliverances of the other senses any more secure. So the objection must be that, on his own, an individual not only can learn to distinguish deliverances of dreams and hallucinations from perceptual deliverances but can also learn to distinguish unreliable perceptual deliverances from reliable perceptual deliverances. Let us concede for the moment that a perceiver can tell whether he is a consistent judge of color. The items that looked red to him yesterday look red to him again today, variations in the apparent colors of enduring objects can be correlated with variations in the light source, and so on. One visual deliverance supports another. Although this might seem sufficient to enable him to draw the relevant distinctions, it is not. Some items that consistently appear red to him might be green. People discover that they are color-blind by learning that other people discriminate shades that they cannot tell apart. If an agent cannot appeal to other people’s color perceptions to check his own, he has no way to tell whether his socks match. For all he knows, he might be color-blind. The same holds for the other secondary qualities. On our own, we do not know what we are missing. A perceiver fares better with respect to primary qualities. He can check his visual deliverances against his tactile deliverances. He can correlate the way things look with the way things feel. If they align, he has, on his own, evidence of reliability. But suppose he discovers a systematic misalignment. Objects that feel circular look oval. That is, the perceiver’s tactile deliverances are as of every point on a disk’s periphery being equidistant from a central point, whereas his visual deliverances are as of the disk being longer in one direction than in another. How could he figure out which, if either, is right? The obvious (and correct) answer is to measure the disk along different axes. If all the measurements yield the same answer, it is a circle; otherwise, it is not.4 Suppose his measurements reveal that the disk is a circle. Suppose he performs the measurements on numerous occasions on different objects and discovers that the misalignment is consistent. Things that feel circular to him generally look elliptical. Then he should conclude that either his visual perception of shape is defective or circles generally present the visual appearance of being elongated in one dimension. Without consulting others, he has no way to tell which is the case. He can find out through measurements which objects are circular; but by himself, he cannot discover how circles should look.

 

  This assumes that he can trust his measurements. For now, let us simply grant that.

4

The Commonwealth of Epistemic Ends  249 For there is no a priori reason why things that are equidistant from a given point should look equidistant from a given point.5 Of course, most of our perceptual deliverances are not just deliverances as of primary or secondary qualities. They are deliverances as of people, animals, plants, and other objects located in and moving about in space. The crucial question is whether such deliverances are trustworthy. To the extent that the reason to accept them does not turn on the agent’s reliability with respect to secondary qualities, perhaps they are. If so, a life-long Robinson Crusoe, who was shipwrecked at birth and reared by wolves, would be in a position to place considerable trust in his mutually supportive perceptual deliverances. The problem comes with establishing criteria for being the same thing—the same object, quality, relation or whatever. If the criterion were: x and y are deliverances as of p just in case x presentations and y presentations are identical then, at least assuming we are not prey to an analog of the private language argument, Life-long Crusoe would have no problem. But deliverances as of the same thing—be it a primary quality, a material object, an organism, or a person—vary. Things present different appearances in different lights, from different perspectives, against different backgrounds, and so forth. Many things change their appearances over time. Arguably an isolated individual could establish workable, stable criteria, but doing so is not straightforward. Nevertheless, nothing I have said so far demonstrates that an isolated individual could not do this. For now, let us leave it at that. Winston’s predicament, however, is worse than Life-long Crusoe’s. It is not just that he gets no positive reinforcement for his take on things; rather, he faces widespread peer disagreement. In the circumstances where he takes himself to see Jones, Aaronson, and Rutland, his similarly situated compatriots avow that they do not. Where he thinks he remembers that Oceania was recently allied with Eurasia, his compatriots avow the contrary. Recent discussion characterizes peer disagreement as disagreement that arises when equally able, similarly situated epistemic agents disagree about a topic. Epistemic peers are equally intelligent, equally well educated vis à vis the topic, equally well positioned to judge, and so forth. Conciliationists6 maintain that if Bob is Anne’s epistemic peer, her realization that he disagrees with her undermines her justification for the claim on which they disagree. She has, they maintain, no reason to think she is in a better position to judge the issue than Bob, so she should suspend judgment or at least lower her credence. The steadfast7 maintain that Bob’s disagreement does not provide Anne with such an epistemic obligation. Anne, they maintain, should judge that even though Bob is generally her epistemic peer, in the case where they disagree, Bob must be mistaken. Bob’s opinion then does not give Anne any reason to change her views.    

  This is a variant on the Molyneux problem.   See Conee (2010).

5

7

 

  See Christensen (2007).

6

250  Catherine Z. Elgin If Winston’s situation involves peer disagreement, some of his compatriots are his epistemic peers. We have no reason to think otherwise. They have the same education, the same sorts of experiences and the same evidence. They are equally well situated to judge visual appearances, news reports, and so forth. There is no indication that Winston is vastly more (or less) intelligent than his compatriots. The only apparent difference is his willingness to question the Party line—a willingness which his compatriots apparently lack. Since differences in skeptical or iconoclastic propensities do not disqualify people from being epistemic peers, at least as far as he can tell (and as far as we can tell), some of his compatriots qualify as his peers. Winston’s situation is more complicated than the ones epistemologists typically discuss. For he cannot tell whether his peers actually disagree with him. Although they may be parroting the Party line, they sometimes avow what they take to be a truth supported by evidence. He has no way to tell whether this is one of those times. If the disagreement is spurious, he obviously should ignore his interlocutors and hold fast to his belief. If it is genuine, and the steadfast position is correct, he should again remain firm. But if the disagreement is genuine and the conciliators are right, he should suspend judgment or lower his credence. What should he do when he does not know whether the disagreement is genuine? We standardly take intersubjective agreement to sustain or to enhance justification. The greater the intersubjective agreement that p, the more likely it is that p. In Oceania, the situation is nearly reversed. The more people agree that p, the more likely it is that the Party wants them to take it that p; and the more likely it is that the Party wants them to take it that p, the less reason they have to believe that p is true.8 In what we take to be normal epistemic circumstances, the steadfast appear intellectually arrogant. If Anne and Bob are epistemic peers, their situation seems symmetrical. It appears as reasonable for Anne to think that she has made a mistake as it is for her to think that Bob has. But given the pervasiveness of the Party’s interference, Winston’s situation seems different. Perhaps he should assume that his interlocutors are either disingenuous or misguided. He not only knows what he sees, he also recognizes that a well honed sense of self-preservation leads his peers to make their avowals on the basis of political expedience rather than warrant or truth. So, unlike Anne, he has positive reason to disregard them. Whatever may be the case elsewhere, in Oceania, it pays to be steadfast. But steadfastness comes at a considerable price. It plunges Winston into arrant dogmatism. The Party does not always deceive; it does so intermittently. Sometimes ordinary citizens of Oceania avow what they actually see or remember. In some cases of disagreement, Winston’s peers are apt to be correct. But, according to the steadfast, whenever Winston finds himself disagreeing with a peer, he should judge that he is right. Moreover, the Party does not restrict its deception to political issues. To reinforce its thought control, it sometimes deceives about mundane matters. So Winston

 

  I am grateful to Jonathan Adler for this point.

8

The Commonwealth of Epistemic Ends  251 cannot restrict his dogmatism to certain topics. Every time he disagrees with anyone about anything, he should assume he is right. He could then never discover that he is color-blind. Having no reason to take himself to be a worse judge of colors than Neville, he should not let Neville’s apparent disagreement undermine his conviction that his socks match. Nor, evidently, does it matter how many others apparently agree with Neville. Winston should think that even if people are normally good judges of color, in this case, they are wrong or disingenuous. Neither the fact of apparent disagreement nor the number of those who apparently disagree with him should carry any weight. Since the Party may have its thumb on the scale, this position is not unreasonable. Maybe Winston has little reason to consider his peers wrong, but he has plenty of reason to suspect that they are disingenuous. Winston need not, and perhaps cannot, recognize anyone as his epistemic superior. So in the face of apparent disagreement, he should always take himself to be right. His compatriots might conceivably be sources of new information, although its reliability would be highly doubtful. But they can never give him reason to revise his opinions. Dogmatism is one worry; epistemic insecurity is another. The Party’s machinations strip away valuable checks on a person’s beliefs. If Winston is reliable in some domain, his take on things in that domain is likely to be correct. But if he cannot tell that he is reliable, this avails him little. He may be steadfast, dogmatic, and complacent about his take on things. But steadfastness sacrifices resources he needs to recognize and correct errors—at least those that do not reveal themselves through obvious inconsistency. He would be epistemically better off if he could adopt a higher-order perspective and recognize when and to what extent he is reliable. By himself, he is in no position to discover systematic or compensating errors. If he considers himself infallible, he is wrong; but if he considers himself fallible, he should be epistemically insecure. Winston has no check on his memory or current sensations, hence no basis for distinguishing between remembering and seeming to remember, or between seeing and seeming to see. So introspection is undermined as well. If one cannot distinguish between remembering and seeming to remember, seeing and seeming to see, and so forth, one cannot reliably identify one’s own mental states. Is a given representation a memory or merely a fantasy? Is it a belief or merely wishful (or fearful) thinking? Does it reflect what is politically expedient or what is likely the case? Such states do not come already labeled. Memory and seeing involve being causally connected to the world in a law-governed way. Fantasy, wishful thinking, and hallucination have a different etiology. If Luke remembers having seen the Eiffel Tower, then his current mental representation as of his having seen it is suitably connected to his actually having seen it. If he imagines that he saw it, his current mental representation as of having seen it lacks that causal connection. If it is in principle impossible to determine whether the requisite causal connection obtains, the distinction between the real and the apparent wanes. Strictly from the inside, there seems no way to learn to tell the difference. This is not to say that every apparent memory must be checkable. But enough of them must be checkable to stabilize the distinction between the two sorts of representations.

252  Catherine Z. Elgin As Wittgenstein says, “An ‘inner process’ stands in need of outward criteria.”9 Strictly from the inside, there seems to be no way to differentiate the real from the merely apparent; and in Oceania, there are no external sources of confirmation. One might object that Oceanians have the same sorts of beliefs about mundane matters as anyone else; they are just more circumspect about voicing them. Perhaps. But this assumption is not obviously correct. Oceanians might internalize the Party’s strictures to such an extent that their attitudes are more reflective of political expedience than of evidence or truth. Beliefs should be sensitive to changes in the world in the sense that, as Tamar Gendler says, “when we gain new all-things-considered evidence—either as a result of a change in our evidential relation to the world, or as a result of a change in the (wider) world itself—the norms of belief require that our beliefs change accordingly.”10 An unsuicidal citizen of Oceania is sensitive to changes in the wider world. Her take on things is responsive to the way the world is, and it changes as the world changes. But the changes she is (most) sensitive to are not changes in the evidence as to whether p is the case; they are changes in the evidence that the Party wants people to take p to be the case. Possibly her belief-forming mechanism has a political override. She is then sensitive to evidence that bears on the truth of the content of her attitude unless that sensitivity clashes with her sense of the political expedient. Possibly, because her sense of the politically expedient so dominates, she fails to form beliefs, instead forming a hybrid attitude that interweaves doxastic and prudential elements. This may be one place where Orwell’s scenario reveals itself to be a reductio, if human beings could not survive without a bedrock of mundane beliefs that are impervious to political meddling. But if a basic function of belief is to promote survival by giving us a way of recognizing and avoiding danger, and if the greatest danger comes from bucking the Party, it would not be unreasonable to expect people’s attitudes to be more attuned to the Party’s mandates than to other facts. The Party’s thought control is so great that it even convinces Winston that 2+2=5. At this point, adherents of the a priori might insist that Orwell has gone too far. Unfortunately, it is not obvious that he has. Winston is tortured into believing that 2+2=5. He does not take to it easily. According to BonJour, “In the most basic cases [a priori] reasons result from direct or immediate insight into the truth, indeed, the necessary truth, of the relevant claim.”11 Presumably “2+2=4” is a basic case. If so, thinkers typically have a direct or immediate insight into its necessary truth. Nevertheless, it is not implausible that through a sufficiently intense and painful process of operant conditioning, a person could be brought to block the direct and immediate insight that 2+2=4 and substitute the (less painful) impression that 2+2=5. The mere fact that 2+2 could not equal 5 does not insure that someone could not be conditioned to think it does. Nor does it insure that someone could not be brought to have a direct and immediate impression that 2+2=5. No doubt one could maintain that Winston and    

  Wittgenstein (1953: #580). See also, Wilfrid Sellars (1963: 127–96).   Gendler (2008: 565).   11  BonJour (2005: 99).

9

10

The Commonwealth of Epistemic Ends  253 other torture victims lost their moorings to the extent that they no longer know what “2+2=” means. So they do not mean what we mean by “2+2=x” If ‘insight’ is a success term, then they lack insight. Nevertheless, Winston has what he takes to be a direct and immediate impression that 2+2=5. In Oceania, seemingly a priori deliverances are unreliable too.12 Torture is memorable. So we may suppose that Winston remembers being tortured into taking 2+2 to equal 5. If so, he has reason to reject his direct and immediate deliverance. But more subtle forms of mind control—perhaps hypnosis, propaganda, or delicate forms of operant conditioning—might have the same result. The Party uses a variety of methods that are evidently widely effective. There is no reason to believe that every interference with a priori deliverances leaves a phenomenologically salient mark. In all the cases I have discussed, things have been so arranged that there can be no independent confirmation of one’s judgments. It is impossible or ineffective, for Winston to ask anyone, “Did you see that?” “Do you remember that?” “Does your calculation agree with mine?” “Didn’t yesterday’s newspaper say something different?” “Weren’t we taught the contrary in school?” It is fruitless for him to consult contemporary news media or historical records, for they are bound to toe the current Party line. Winston, and presumably all other even mildly reflective members of the society, are aware of what the Party is doing. Although the Party purports to be telling the truth, this is a blatant pretense, which no one who is paying attention could believe. Nor, does it matter. Citizens behave as the Party wants them to, regardless. They realize that they cannot trust the media, or their senses, or their compatriots, or their calculations. But their higher-order awareness of the corruption of their epistemic resources does little to improve their epistemic lot. Possibly their higher-order awareness enables them to frequently suspend judgment rather than harbor false beliefs. But they know very little. For the recognition that the Party regularly promulgates falsehoods does not enable them to determine what, in particular, is true. Some of the contentions a citizen of Oceania accepts are no doubt accurate. Sometimes, she actually sees what she takes herself to see. Sometimes, she remembers events that actually occurred as she remembers them. News media occasionally report something that in fact occurred. Some of her calculations are correct. Moreover, some coherent constellations of considerations she accepts are presumably accurate as well. If she believes the members of such a constellation she has locally justified, locally reliable true beliefs. But because such islands of accuracy are surrounded by a vast sea of unjustified contentions, they are not trustworthy. Given the overarching epistemic circumstances, luck plays too great a role in their being true and in their being justified.   12  One might think that closely related beliefs, such as the belief that 1+1+1+1=4 suffice to indicate to Winston that he is wrong. So systematicity considerations would tell against Orwell. But this need not be the case. All that the conflicts can reveal to Winston is that he’s going wrong somewhere. He might, nevertheless, be incapable of figuring out which of his beliefs is incorrect. Remember what it is like to not understand how to do a math problem. You may fully realize that something has gone wrong somewhere without having a clue which of the claims is incorrect.

254  Catherine Z. Elgin Ordinarily, we assume that the coherence and mutual supportiveness of a suitable number and range of deliverances of sensation, introspection, a priori insight, inference and testimony suffice for justification. It would, we think, be a miracle if all these sources pointed to the same falsehood. Orwell’s lesson is that this need not be so. Convergence on the same false or unwarranted conclusion would be a miracle only if the sources were suitably independent of one another. But if mutual support is generated perniciously, the independence assumption is undermined. Minimally, to afford justification, deliverances must not products of collusion or coercion. More strongly, we may also need reason to believe that they are not. A vast Orwellian conspiracy is not required to undermine the epistemic standing of familiar deliverances. More subtle arrangements that silence certain voices or deflate their credibility can produce the same effect on a more modest scale.13 By omitting or downplaying the significance of particular perspectives, prejudice and stubbornness can skew matters to the point where intersubjective agreement does not supply justification. If naysayers are silenced, then the fact that everyone who speaks agrees is not a good reason to think that a contention is true. If the credibility of naysayers’ testimony is deflated, then the fact that the balance of what is taken as evidence weighs heavily in favor of a hypothesis is at best a weak reason to believe it. The negative epistemic consequences do not just undermine the opinions of those who have been silenced. They affect the entire epistemic community. Without access to the discredited opinions, community members do not know what they are missing. Moreover, the skewing of the evidence that results from deflating credibility undermines the trustworthiness of what evidence they have, for it artificially inflates the credibility of the voices that are heard. Such a situation is epistemically unjust. 1984 represents an extreme case. But epistemic injustice is not unusual. Here is an example: In the 1940s and early 1950s Barbara McClintock published a series of papers contending that she had discovered sequences of genetic material that change position on the chromosome of corn. She was not taken seriously. As a result, she said, “I stopped publishing detailed reports . . . when I realized, and acutely, the extent of disinterest and lack of confidence in the conclusions I was drawing from my studies.”14 Historians of science disagree about whether sexism figured in the scientific community’s disregard of McClintock’s findings.15 Whatever the reason, she was effectively silenced. Her discovery was not taken up until the 1960s when François Jacob and Jacques Monod discovered that the same transposition occurs in bacteria. Plainly the understanding of genetics was retarded by the failure to give McClintock’s discoveries their due. Even if it is not Orwellian in motivation or scope, such epistemic injustice deprives the community of inquiry of data and skews the evidence it has.

  13  See Fricker (2007) and McGowan (2009).   14 Barbara McClintock, Letter to J.  R. S.  Finchman, May 16, 1973, , consulted June 10, 2009.   15  See Keller (1983) and Comfort (2001).

The Commonwealth of Epistemic Ends  255 At a minimum, these reflections on 1984 reveal that epistemology needs a non-interference requirement. Only if beliefs and belief generating mechanisms are free of Orwellian, or quasi-Orwellian interference can their products be epistemically warranted. Arguably, they show something more. If my point about the differentiation of representations and the fixation of belief is correct, then to be capable of having beliefs at all requires the support of a non-coercive community, not merely the absence of interference by a coercive one.

13.2  The Epistemological is Political Habermas (2001) maintains that political well-being requires uncoerced conversation. As I read it, 1984 reveals that epistemic well-being does too. Agreement among free and equal inquirers enhances the epistemic standing of a claim; coerced agreement does not. If inquirers are free, they can take up any perspective they like, and examine the issue from that perspective. If they are equal, all inquirers have an equal opportunity and an equal right to venture hypotheses, to raise objections and counter-hypotheses, and provide reasons for them. Under such circumstances, there are no political impediments to discovering what is the case. But if intersubjective agreement results from coercion, collusion, credibility inflation or deflation, none of this holds. If inquirers cannot examine an issue as they see fit, if they cannot raise objections or gain a hearing for them, or if their views are given undue weight, agreement provides small reason to think that a conclusion is true or that it stands up to serious testing. “Free and equal” here is a political requirement on a community of inquiry; it is not a claim that all are equally knowledgeable or equally intelligent. Some hypotheses are plainly untenable; some perspectives are obviously skewed; some methods are demonstrably unsound. These are quickly and rightly dismissed. Nevertheless, to block the Orwellian threat, the opportunity to venture a hypothesis, and the right to have it assessed on its merits must be real. As Mill insists, “The beliefs we have the most warrant for have no safeguard to rest on but a standing invitation to the whole world to prove them unfounded.”16 The epistemic value of uncoerced conversation lies, I suggest, not so much in the particular agreements that it generates as in its propensity to uncover and correct errors. It might seem that all that is needed is a non-interference requirement. For an agent to be in a position to know a fact or to understand a constellation of facts, she must stand in a suitable relation to those facts. The Orwellian scenario is a reminder that other people can undermine her capacity to stand in such a relation. That scenario is enough to discredit the assumptions about individualism and attunement that I mentioned at the outset. I  think, however, that the epistemic community plays a more  

  Mill (1978: 20).

16

256  Catherine Z. Elgin substantive role. It constitutes epistemic norms and specifies the conditions for their satisfaction. With relatively few exceptions, facts are independent of anyone’s beliefs about them. We do not need intersubjective agreement (nor would such agreement help) to make beliefs true. Moreover, the reliability of our modes of access to the facts may be beyond our control. For the purposes of argument, let us suppose that this is so.17 Will’s visual perception yields a deliverance as of a black dog. He straightway believes that a black dog is in front of him. He is passive with respect to this belief in that it is a product of involuntary belief-forming mechanisms. He is, in Kantian terms, heteronomous. Just as Kant’s heteronomous subject acts on whatever inclinations she happens to have, a heteronomous doxastic subject like Will forms beliefs on the basis of whatever deliverances he happens to have. This does not entail that he believes the content of every deliverance, but it is not up to him which ones he believes. Rather than saying that he forms beliefs, it would be more accurate to say that belief contents just strike him as so. If asked why he believes there is a black dog in front of him, he could honestly reply, “That’s just the kind of guy I am.” The heteronomous doxastic subject is under the sway of whatever belief-forming mechanism is in effect. If the mechanism is reliable, his belief is apt to be true; if not, it is not. If the mechanism is reliable, his doxastic position is objectively secure. But from a subjective perspective, he seems unacceptably vulnerable. Since his belief contents just strike him as correct, he is in no position to criticize them or to reflectively endorse them. One might argue that such a subject has resources that I did not recognize. He has second-order beliefs. He could reflectively endorse his first-order beliefs, and be right to do so, if his mechanism for forming second-order beliefs was reliable. Although true, this is unhelpful. If he is heteronomous with respect to his second-order beliefs— if, that is, his belief that a first-order belief is credible just strikes him—the problem recurs. To attempt to solve it by appeal to third-order beliefs which would enable him to endorse his second-order beliefs which would enable him to endorse his first-order beliefs, sets off a disastrous regress. Is there any way to vindicate second-order considerations without embarking on a regress? Given that I’ve posed the problem in Kantian terms, it is no surprise that I extract an answer from Kant. One formulation of the Categorical Imperative is that those maxims are acceptable that an agent can endorse as a legislating member of a realm of ends. These maxims are not just laws that the members of the realm of ends are subject to, they are laws that they make themselves subject to. According to Kant, in the moral realm, legislators enact the laws that bind them. I suggest that the same holds in the epistemic realm. What gives certain second-order claims their epistemic authority is that they express standards, rules, or principles that epistemic agents can   17  This is a simplification, since we can learn to refine our sensibilities and thereby gain epistemic access to aspects of things that were originally inaccessible. Moreover, learning enables us to recognize things to which we were once oblivious.

The Commonwealth of Epistemic Ends  257 on reflection endorse. Thinking of themselves as reasonable and rational, they are prepared to accept those second-order considerations as specifying constraints on what they ought to accept. The suggestion then is that what blocks the regress is that the second-order endorsement is a product of agency. Epistemic subjects are epistemic agents; they take their beliefs, practices and so forth to be answerable to certain norms because they think that cognitively acceptable beliefs, practices, and so forth ought to be answerable to those norms. They thus make the epistemic laws that bind them. And because they are agents, they can both start and end a justificatory path.18 Why should epistemic agents be construed as joint legislators rather than as autocrats? Why shouldn’t each agent decide by herself and only for herself what epistemic principles merit her purely personal reflective endorsement? The answer is this: if an agent reflectively endorses an epistemic principle, she considers it reasonable that her cognitively serious actions, such as inferring and asserting, accord with that principle. She repudiates the gambler’s fallacy because she recognizes that in committing the fallacy she makes herself vulnerable to Dutch Books. She endorses modus ponens because she recognizes that it is truth preserving. Such an agent has no reason to think that as an epistemic agent she should be subject to principles that similarly situated epistemic agents are not subject to. She does not, for example, think that the desirability of avoiding Dutch Books stems from a personal predilection. Because she takes the principles she reflectively endorses to be reasonable and rational in the epistemic circumstances, she thinks they should be binding on similarly situated epistemic agents. But she recognizes that epistemic agents should be subject only to principles that they consider worthy of reflectively endorsement. By her own lights then, only such principles as she can justify to the similarly situated epistemic agents—those who constitute her community of inquiry—merit her reflective endorsement. Because the principles that govern their epistemic practices must be ones that its members can justify to one another, the community of inquiry serves as a stay against idiosyncrasy or bias in reflective endorsement. It not only fails to interfere, it actively fosters epistemic success. Does it follow that Life-long Crusoe is incapable of epistemic success? Possibly. He evidently lacks the resources to rule out certain sources of error. One is confirmation bias. Someone given to confirmation bias weighs evidence that supports his convictions more heavily than evidence that undermines them. If Life-long Crusoe does so, he may have no way to discover his error, particularly if the bias is slight. Then many, if not all, of his beliefs are fated to be unjustified. Suppose, however, that he is not prey to confirmation bias. Then he is, we may assume, as reliable as the rest of us. It might seem therefore that he is just as capable of knowing, understanding, and being justified as we are. But we have a resource that he lacks. We assess our beliefs in light of the standards

 

  I am indebted to Jonathan Adler for this point.

18

258  Catherine Z. Elgin that the community of inquiry has designed to (among other things) filter out confirmation bias. So we have a reason to think that beliefs that satisfy the standards are not the result of confirmation bias. Life-long Crusoe has no such reason. His grounds for his belief are thus epistemically impoverished as compared to ours. Satisfying the standards of a community of free and equal epistemic agents enhances the epistemic standing of a belief. Whether we should conclude that Life-long Crusoe is incapable of knowledge, or some other epistemic success, is not clear. That depends on where thresholds are set. If first order reliability suffices, and Life-long Crusoe is not prey to confirmation bias, or kindred epistemic failings, he knows. If epistemic success requires having reason to believe one has avoided the pitfalls, he does not. But even if he knows, his situation is epistemically precarious. It is, from his perspective, just by luck that he has managed to avoid the pitfalls.19 An epistemic agent can be mistaken about which principles the members of her community of inquiry can justify to one another. In reflectively endorsing p, she commits herself to p’s satisfying standards that are justifiable to the relevant community of inquiry. If it does not, her endorsement of p is an error. By her own lights, she ought not accept it. This raises a more serious worry. What if the community standards are wrong? A community of inquiry sets and reflectively endorses its epistemic standards, and its members justify their inferences and conclusions to themselves and one another by appeal to those standards. Because the community has ample evidence of human fallibility, the standards and criteria of application they reflectively endorse are not considered fixed or final. They are subject to rejection or refinement if they do not serve the community’s evolving epistemic ends.20 The claim of current standards to epistemic authority lies in their having thus far withstood the standing invitation to prove them

  19  Thanks to Alvin Goldman and Sanford Goldberg for pressing me to address this issue.   20  An example may bring this out. “Inanimate objects are identical when their parts are identical” is a principle that many philosophers consider correct. Being a universal claim, it should hold for particles of a viscous fluid. However, if it holds, then “F=ma” does not. Molecules in a viscous fluid move at different rates. In prototypical applications of “F=ma,” forces act on objects like billiard balls that have sharp boundaries. But in viscous fluids, the “forces” on the “object” are the effects on momentum of molecules moving in and out of that “object.” So preserving “F=ma” requires continually redefining what constitutes a particle, letting different molecules comprise it at different times. Although physicists concede that “F=ma” does not hold at the quantum level or at relativistic speeds and distances, it is an extremely valuable law for characterizing the behavior of middle-sized items in this neighborhood of the cosmos. These include the viscous fluids flowing around here. Fluid mechanics thus characterizes its particles so as to comport with the law. Rather than insisting that all component molecules of a particle be the same from one instant to the next, they let the individual molecules come and go, but keep the average enclosed mass constant. (See Wilson 2006: 158–59.) “F=ma,” evidently, is so central a law of physics that scientists are willing to make drastic revisions in the criteria for the identity of a fluid particle over time in order to preserve it. In this case, the tension is acute. One way or another, a major revision in antecedently plausible principles is needed. Either scientists must revise a very reasonable metaphysical commitment about the identity of an object over time, or revise a fundamental law of physics. The recognition of the clash leads to a revision in the principles epistemic agents are willing to accept on reflection.

The Commonwealth of Epistemic Ends  259 unfounded. This is no guarantee, but it affords some reason to countenance them. Nevertheless, one thinks, the principles might be misguided. Being unrefuted is not the same as being irrefutable. Maybe we would do better then to insist that some other feature—perhaps reliability or bearing a suitable relation to foundational claims—is constitutive of epistemic acceptability. Is there any reason to think that the sort of reflective endorsement I have described is anything more than indicative of acceptability? In answering this, we should remember that we are epistemic agents. Epistemic resources are things we work with, not just things we admire or credit. We use them to apply, extend, and improve upon what we take ourselves to know and understand. The considerations that have withstood Mill’s challenge are considerations that epistemic agents consider as sound a basis for reasoning and epistemically responsible action as any they have. Maybe they are not really (in some sense of “really”) a sound basis for reasoning and action. Perhaps an agent’s beliefs are incorrect, or her methods are biased. Perhaps her community shares her errors and biases. This could be so. Although her system has gone through a fairly rigorous process of testing and correction, there is not guarantee that all flaws have been eliminated. Still, what is an agent to do? Either she assesses her deliverances on the basis of standards she reflectively endorses or she does something else. If she forgoes assessment entirely, she is heteronomous. She thinks and acts on whatever strikes her, subjecting her deliverances to no critical filter. Should her beliefs turn out to be true, that is, from her perspective, a lucky accident; she has nothing she can call a reason for them. If she deliberates and acts on the basis of standards she does not reflectively endorse, her epistemic situation is equally bleak. Even if satisfying alien standards is objectively more truth conducive than satisfying her own standards, she has no reason to think so. By her own lights, the fact that a belief that satisfies those standards is true is still a lucky accident. To assess her beliefs on the basis of alien standards thus would be epistemically irresponsible. She has no reason to trust them. Still, to be complacent about whatever standards she finds herself with would be irresponsible as well. The mere fact that the standards are hers is no reason to credit them. But if her standards have survived serious, sustained testing and correction by a community of free and equal epistemic agents—if, despite their best efforts, those agents have not been able to prove them unfounded—relying on those standards in the current epistemic circumstances is not irresponsible. Fallible and potentially flawed as they are, the standards are at least as good as any available alternative.21

  21  I am grateful to Jonathan Adler and Jonathan Matheson for helpful comments on an earlier draft of this chapter.

260  Catherine Z. Elgin

References BonJour, Laurence. (2005). “In Defense of the A Priori,” in Matthian Steup and Ernest Sosa (Eds.), Contemporary Debates in Epistemology. Malden: Blackwell, 98–104. Christensen, David. (2007). “Epistemology of Disagreement: The Good News,” Philosophical Review 116: 187–217. Comfort, Nathan C. (2001). The Tangled Field: Barbara McClintock’s Search for the Patterns of Genetic Control. Cambridge, MA: Harvard University Press. Conee, Earl. (2010). “Rational Disagreement Defended,” in Richard Feldman and Ted Warfield (Eds.), Disagreement. Oxford: Oxford University Press, 69–90. Fricker, Miranda. (2007). Epistemic Injustice. Oxford: Oxford University Press. Gendler, Tamar. (2008). “Alief in Action (and Reaction),” Mind and Language 23: 552–85. Habermas, Jurgen. (2001). “Truth and Society: The Discursive Redemption of Factual Claims to Validity,” in On the Pragmatics of Social Interaction. Cambridge, MA: MIT Press, 85–104. Keller, Evelyn Fox. (1983). A Feeling for the Organism. New York: W. H. Freeman. McClintock, Barbara. (1973). Letter to J. R. S. Finchman, May 16, 1973, URL = , consulted June 10, 2009. McGowan, Mary Kate. (2009). “On Silencing and Sexual Refusal,” Journal of Political Philosophy 17: 487–94. Mill, John Stuart. (1978). On Liberty, Indianapolis: Hackett. Orwell, George. (1961). 1984. New York: New American Library. Sellars, Wilfrid. (1956). “Empiricism and the Philosophy of Mind,” in H. Feigl and M. Scriven (Eds.), Minnesota Studies in the Philosophy of Science, Vol. 1. Minneapolis, MN: University of Minnesota Press, 253–329. Vogel, Jonathan. (1990). “Are There Counterexamples to the Closure Principle?” in Michael D. Roth and Glen Ross (Eds.) Doubting: Contemporary Perspectives on Skepticism. Dordrecht: Kluwer, 13–27. Williams, Bernard. (2002). Truth and Truthfulness. Princeton: Princeton University Press. Wilson, Mark. (2006). Wandering Significance. Oxford: Clarendon Press. Wittgenstein, Ludwig. (1953). Philosophical Investigations. Oxford: Blackwell.

14 Assertion and the Ethics of Belief * Sanford Goldberg

This chapter aims to establish (i) that there is such a thing as an “ethics of assertion,” and (ii) that, given the connection between assertion and belief, the “ethics of assertion” can be used to provide motivation for some (interpersonal) constraints bearing on what might go under the label “the ethics of belief.”

14.1  Introduction to the Topic Investigations regarding the “ethics of belief ” typically speak about one’s duty to believe in line with one’s evidence. From where does this duty arise? And what is the nature of this duty? Is it a duty to oneself? If it is a duty to oneself—and many will wonder how it could be otherwise—in virtue of what does one have such a duty? In this chapter, I aim to shift the ground of the discussion. It is not that I think that the foregoing perspective on the ethics of belief is misguided. Rather, it is that I think that it is possible to see how at least some constraints bearing on the normative dimension of our doxastic lives arise from the duties and responsibilities that we bear to one another as participants in that cooperative activity we call “conversation.” It will be my aim here to focus on one sort of speech act—assertion—and to argue that the norm that governs it gives rise to certain responsibilities, both on the part of the speaker and on the part of the hearer. These responsibilities reflect the mutual expectations that speakers and hearers are entitled to have, in virtue of their mutual awareness of the speaker’s having made an assertion. I will then go on to appeal to the connection between assertion and belief, together with our informational duties and   *  With thanks to the members of the audience at the A. David Klein Philosophy Symposium entitled “The Ethics of Belief ” (University of North Florida, October 2011), where a version of this chapter was presented. Special thanks to Mark Alznauer, Heather Battaly, Paul Carelli, E.J. Coffman, Ryan Davis, Mark Dechesne, Kyla Ebels Duggan, Rich Feldman, Alvin Goldman, Heidi Grasswick, Mitch Haney, Tom Kelly, Arie Kruglanski, Jennifer Lackey, Jon Matheson, Ted Poston, Baron Reed, Ernie Sosa, Sarah Wright, and Rico Vitz for conversations on these topics. (They are not responsible for any of the bad argumentation or false claims in what follows!)

262  Sanford Goldberg responsibilities to others, in order to argue that given these facts about the practice of assertion, we can derive some (admittedly minimal) constraints that we might collect under the label “the ethics of belief.” Since the constraints I will be deriving are somewhat minimal, a comment is in order regarding why these constraints should be of interest. I submit that the manner in which I purport to derive them—from the norm governing the speech act of assertion—is noteworthy. For it suggests that at least some of the duties and responsibilities that go under the label of the “ethics of belief ” are duties and responsibilities that one owes, not (or at least not merely) to oneself, but (also?) to others. Given that the phenomenon of belief is often regarded as a personal (individualistic) affair, I regard it as an interesting result that at least some of the “ethics of belief ” is motivated by interpersonal considerations—what we owe to one another as fellow communicators in a world in which we depend on others for much of what we take ourselves to know. My discussion of these matters will be shaped by three assumptions regarding the speech act of assertion. These assumptions are widely (though not universally) endorsed in the literature on assertion. First, as a type of speech act, assertion answers to an epistemic norm of some sort or other. Second, this feature of the speech act of assertion is common knowledge to all competent speakers of a language. And third, competent speakers of a language can readily and reliably discern when an observed speech act (involving a sentence in a language they understand) is an assertion. I will argue that these three assumptions provide the basis for what I will call an “ethics of assertion.” In order to simplify matters, I will formulate these three assumptions in one thesis. Let us designate this the thesis asserting “Common Knowledge regarding the Norm of Assertion,” (CKNA) to the following effect: CKNA It is common knowledge that assertion—a readily-discernible speech-act type—is answerable to an epistemic norm of some sort or other. In what follows I do not want to assume anything about what that norm requires. This is an issue that has generated a good deal of controversy in the literature. Some say that it requires knowledge, others that it requires adequate evidence, still others that it requires that the hearer occupy some other epistemic position regarding the proposition asserted.1 I don’t think I have to enter into this controversy in order to make the points I wish to make. For whatever you think the norm of assertion is—whether it is   1  What I am assuming here, in connection with CKNA is a widely-endorsed, but not entirely uncontroversial, set of three claims: (1) as a speech act, assertion is governed by a norm; (2) this norm is epistemic, in the sense that the standard it imposes on appropriate acts of this kind is an epistemic standard; and (3) something in the vicinity of these first two claims are common knowledge among competent speakers. Some reject (1) (see Pagin 2011 for a discussion); some accept (1) but reject (2) (see Weiner 2005, and perhaps Bach and Harnisch 1979). Among those who accept either or both of (1) and (2), something like (3) usually comes along for the ride (see Goldberg 2011 for an explanation). Still, (1)–(3) form a triad of claims that are accepted by a good majority of those contemporary philosophers who write on assertion. For the variety of approaches that have been taken regarding the nature of assertion, see MacFarlane (2011).

Assertion and the Ethics of Belief  263 as demanding as knowledge, or something that requires only an adequate degree of evidence—you will agree with the following (at least insofar as you agree that it has an epistemic norm in the first place): in asserting something, the speaker performs an act regarding which it is common knowledge that her act was proper (warranted) only if she had the relevant epistemic authority. It is this piece of common knowledge, I want to argue, that gives rise to what I will call the “ethics of assertion.” We should be able to agree on this much even if we disagree over what the precise content of the norm is.

14.2  The Ethics of Assertion, Part I If CKNA is true, we would predict that the practice of assertion is rich with mutual expectations between speakers and hearers. These expectations are grounded in common knowledge of the fact that assertion has an epistemic norm. In this section I want to develop this point, which I think accurately reflects our experience as both producers and consumers of assertions. After doing so I will go on to argue (in the second half of this section and the whole of the next section) that the mutual expectations themselves reflect what we might call the “ethics of assertion.” Let me begin with the mutual expectations that arise in connection with assertion. Suppose that CKNA is true. Then in asserting something, the speaker performs an act regarding which it is common knowledge that her act was proper (warranted) only if she had the relevant epistemic authority. We might then regard the act of asserting that p as conveying that the speaker does in fact possess the relevant epistemic authority with respect to the truth of the proposition that p.2 Now others have remarked that a common reaction elicited by assertion is to query how the speaker knows, or what her evidence is.3 Employing CKNA, we can describe such a case as one in which the hearer is requesting the speaker to vindicate the conveyed claim of relevant epistemic authority.4 The fact that such a response is ordinary highlights one class of expectations in play when assertions are made: hearers expect speakers who make an assertion to recognize that they are under some obligation to vindicate their relevant epistemic authority, if called upon to do so. But we might also note that speakers typically anticipate that hearers will have such expectations. This is why it comes as no surprise to a speaker if, having made an assertion, she elicits one of these responses in a hearer. What is more, in such a situation, the speaker herself will regard herself as under some obligation to respond: it would be curious indeed for a speaker to assert something, to be queried regarding how she knows, and to respond with a shrug and then simply proceed as if nothing were remiss. (Stranger still if the speaker responded by manifesting her impression that the hearer’s query was inappropriate!) This is not our practice.   2  By “conveying” I have in mind something like “communicating, as part of the conventional significance of the illocutionary force of the speech act.” Clearly, what is conveyed in this sense is not part of the content asserted.   3  The “How do you know?” reaction is described in Williamson (2000).   4  As shorthand I will sometimes speak of the speaker “vindicating her relevant epistemic authority.”

264  Sanford Goldberg And hearers know this. Again, this is merely to say that these expectations that speakers and hearers have of one another are mutually known. We can describe these mutual expectations using the language of responsibility. When a speaker makes an assertion, she does something regarding which it is common knowledge that what she did is proper only if she had the relevant epistemic authority. But in that case the hearer is entitled to believe that the speaker acknowledges that she has the responsibility to possess the relevant epistemic authority. The propriety of this talk of entitlements and responsibilities has as its source the norm (or rule) governing assertion. Consider in this respect the following argument from Angus Ross: It is a quite general feature of rule-governed life that the responsibility for ensuring that one’s actions conform to the rules lies primarily with oneself and that others are in consequence entitled to assume, in the absence of definite reasons for supposing otherwise, that one’s actions do so conform. Thus where the rules are such that one may perform a certain action only if a certain condition obtains, . . . then to perform the action is to entitle witnesses to assume that the corresponding condition obtains. If that assumption proves false and others act upon it with unfortunate consequences, at least part of the responsibility will lie with oneself for having entitled them to make that assumption. (Ross 1986: 77–8)

Ross’s conclusion on this score5 is in the vicinity of the claim I wish to make, though it is a bit stronger. His conclusion is that if it is common knowledge that doing X is proper only if such-and-such is the case, then if subject S does X, others are entitled to assume that such-and-such is the case, with the result that they can blame S if it should turn out that such-and-such is not the case. Applied to the case of assertion, Ross’s claim would be this: if you assert that p, others are entitled to assume (in the absence of definite reasons to the contrary) that the assertion was warranted, with the result that they can blame you if it should turn out that it was not warranted. My claim is slightly different: if you assert that p, others are entitled to assume that you acknowledge that you are responsible for having the relevant warranting authority. In one clear sense, my claim is weaker than Ross’s. That a speaker acknowledges having a certain responsibility does not imply that she has fulfilled that responsibility. Consequently, an entitlement to assume the former does not by itself entitle one to assume the latter, or (arguably) even to assume that the latter holds absent reasons to think otherwise. As we will see, however, my weaker claim, to the effect that S’s asserting that p entitles an audience to assume that S acknowledges being responsible for having the relevant warranting authority, is not without consequence. Before getting to this, though, it is important to be clear about what makes it appropriate to speak of responsibility in connection with assertion. Assertion—at least in its paradigmatic instances—is a public act. Its public nature is seen in the facts, first, that assertions are (typically) made to an audience, but second, that the distinctive

  5 Interestingly, Ross makes the foregoing remark in connection with the rule-governed nature of language use.

Assertion and the Ethics of Belief  265 significance of acts of this type—their aptness for spreading knowledge, but also their susceptibility to being exploited in lies or other forms of insincerity—reflects what is common knowledge regarding the norm governing acts of this type. The responsibilities that arise in connection with assertion arise from this distinctive significance. Some have tried to get at this distinctiveness by likening the act of asserting something to the act of promising someone something.6 The idea that asserting is akin to promising (in some relevant respect(s)) has much to say for it.7 As I see matters, the core similarity is this: the making of an assertion, like the making of a promise, generates a certain kind of entitlement for the relevant audience to form a certain expectation. Take promising first. To promise to X is (among other things) to entitle the person (or persons) to whom one made the promise to expect that one will X.8 If one then fails to X, so that the audience’s expectation is false, the audience can hold one responsible for the unhappy state of affairs. Now take asserting. In a similar way, to assert that p is to entitle the relevant audience to expect that one has the relevant epistemic authority with respect to the proposition that p. However—to revisit the point I made in the previous paragraph—to say that H is entitled to expect S to have the relevant epistemic authority need not be seen as saying or implying that H is justified in believing that S has the relevant epistemic authority.9 The entitlement here is rather an entitlement to hold S responsible for having that authority. The core of the parallel between promising and asserting, then, comes to this: in both cases the speaker (in performing the speech act she did) authorizes the hearer to form a certain expectation; this expectation reflects the mutually-recognized, interpersonal nature of the expectations generated by speech acts of that kind (promise or assertion); and if this expectation is violated (its content is false), the audience can hold the speaker responsible for that unhappy state of affairs. Just as a person who promises “owes it” to the addressee to do what she promised, so too a person who asserts “owes it” to those who observed the assertion to have the relevant epistemic authority.10 I have just defended the idea that it is appropriate to speak of responsibilities in connection with a speaker’s making an assertion, and correlatively of a hearer’s entitlement to hold a speaker responsible. Still, we are not quite at the point where we can speak of assertion’s generating distinctly moral responsibilities. To see why this is, we   6  See, e.g., Angus Ross (1986: 79), Gary Watson (2004), Ted Hinchman (2005), David Owens (2006), and Richard Moran (2007).   7  This said, I think the analogy is likely to mislead in important respects. See footnote 12.   8  It is also to incur the obligation to X.   9  Whether the entitlement to assume that S has the authority is sufficient for H to be justified in believing that S has the authority is a further question – one at the heart of much discussion in the epistemology of testimony. This point is missed by some who contribute to the literature; see Goldberg (2011) for a discussion.   10  The similarity between promising and asserting, while real, can be misleading. As I see matters, one important difference is this: a promise is made to a particular addressee; an assertion, however, need not be addressed to anyone in particular, but instead can be for public consumption. Some (such as Hinchman 2006) have used this to suggest that it is not the broad class of assertion, but rather the special case of assertion that constitutes a telling, that should be the focus of the discussion in the case of testimony. I disagree; see again Goldberg (2011) for discussion.

266  Sanford Goldberg will need to go beyond the similarities between asserting and promising, and enter into the particular content of the norm of assertion itself. My point here can be made in terms of the knowledge norm of assertion, so assume (for the purpose of illustration only) that knowledge is the norm of assertion. It would seem that there are ways of failing to know that p which are such that, were one to fail to know that p in one of these ways, then, while one’s assertion that p would be unwarranted (for failing to satisfy the knowledge norm), one would be blameless in so doing. (And by “blameless” here I mean to include any sense relevant to one’s moral duties to others.) Consider for example Gettier cases: there, one believes that p, one’s belief is both true and justified, yet one fails to know that p, owing to some sort of knowledge-undermining epistemic luck (not reflecting any irresponsibility on one’s own part). So it would seem that if knowledge is the norm of assertion, then not all unwarranted assertions are cases in which the speaker deserves moral blame—with the result that if the speaker can be held responsible in these cases, the responsibility at issue is not moral responsibility. The foregoing illustration supports the conditional claim that if the norm of assertion is knowledge (or indeed any sort of “externalist” epistemic standing), then not all cases of unwarranted assertion are cases in which the speaker deserves moral blame. But even if knowledge or some other externalist epistemic standing is the norm of assertion, two things can be said about the connection between the norm of assertion and the ethics of assertion. First, it remains the case that the speaker “lets the hearer down,” even if the speaker is not to be blamed for doing so. (“Letting the hearer down” would then not be, or at least not always be, a moral failing.) And second, we can still bring in the moral dimension here, albeit in a qualified way. For at a minimum we can say this: in any case in which the speaker “lets the hearer down,” insofar as this “letting down” reflects a failure of the speaker to have lived up to her moral responsibilities, she deserves moral blame. What is more, given CKNA (and the mutual knowledge that it engenders), a hearer is entitled to regard one who asserts as having done all that can reasonably be expected of her to ensure that her assertion satisfied assertion’s norm. In short, even on the assumption that not all unwarranted assertion is morally blameworthy assertion, we still have discerned a significant connection between the norm of assertion and the ethics of assertion. The picture would be this: having asserted that p, it is S’s responsibility to have the relevant epistemic authority, whatever it is; assuming that such authority amounts to knowledge or some other externalist epistemic standing, then S may fail to live up to her assertion-generated responsibility without being morally blameworthy; and whether a case of unwarranted assertion involves moral blame is determined by whether S did all that could reasonably be expected of her to ensure that she had the relevant authority. From a moral perspective, then, what the speaker owes to the hearer is this: she must have done all that can reasonably be expected of her to ensure that she had the relevant epistemic authority. I regard it as an interesting and substantive question what we can reasonably expect of each other in this regard. I suspect that matter is very complicated; in particular, I suspect that an adequate account will have to accommodate facts

Assertion and the Ethics of Belief  267 regarding the various sorts of contexts in which we make assertions, and the various interpersonal relations we have to those to whom we address our assertions. However, rather than trying to provide such an account, I want instead to move on to consider another part of the ethics of assertion. I do so out of the conviction that what a speaker owes to a hearer is only part of the content of the ethics of assertion. The other part has to do with the ethical responsibilities on the part of the hearer: what the hearer owes to the speaker, as it were. This part has generated much less discussion in the philosophical literature on assertion;11 but I think it is an important part of our story.

14.3  The Ethics of Assertion, Part II I want to suggest that, if CKNA is true—it is mutually known by competent language users that assertion is answerable to some sort of epistemic norm12—then the hearer has certain responsibilities as well. To give a sense of the matter here, consider the following case. In the course of a joint activity between various people who are coordinating with one another, a speaker (S) tells a hearer (H) that p. The speech act passes in silence: H neither confirms that he has accepted what S has told him, nor says anything to indicate that he has rejected (or otherwise ignored) what she has told him. Soon thereafter, S acts on the assumption that H accepted that p—only to find out that H does not accept that p. The question is: if S’s acting on the contrary assumption lead to difficulties (for S or for the joint activity), does H bear any responsibility for these difficulties? Does S enjoy anything like an entitlement to assume that when her assertions pass in silence they have been accepted—so that H had a responsibility to indicate rejection? Is this case relevantly analogous to the responsibility S would bear if H had accepted S’s statement under conditions in which S did not in fact have the relevant epistemic authority?13 These questions get us in the ballpark of the sort of issue I want to raise in this section, but they do not get at the most general form of that issue. For one thing, the case above is one in which S and H are engaged in a joint activity; and it might be thought

  11  Much less, but not no attention. Recent work in this ballpark includes Ted Hinchman (2005), Miranda Fricker (2007), Richard Moran (2007), and Jeremy Wanderer (2011). A brief but (characteristically) highly suggestive and insightful discussion can be found in Anscombe (1979).   12  For ease of exposition I have dropped the explicit reference to the reliable discernibility of assertions.   13 Lest one think that this is a crazy notion, consider this, from Stalnaker’s classic (1978) paper “Assertion”: “To make an assertion is to reduce the context set in a particular way, provided that there are no objections from the other participants in the conversation. . . . [T]‌he essential effect of an assertion is to change the presuppositions of the participants in the conversation by adding the content of what is asserted to what is presupposed. This effect is avoided only if the assertion is rejected” (reprinted in Stalnaker 1999, 86; italics added). Although it is not entirely clear from this whether Stalnaker would agree that all parties to the conversation are entitled to assume that an assertion has been accepted unless there is some public indication otherwise, such a reading is compatible with this passage. (It would be the proper reading if we assume that hearers who object to or otherwise reject an assertion have a duty to make this fact known.)

268  Sanford Goldberg that this makes the case sufficiently special so as to preclude any general conclusions regarding hearers’ responsibilities in the face of assertion as such. For another thing, even if there are responsibilities that hearers bear to speakers on observing them make an assertion, it is not obvious that these responsibilities take the form: either accept what the speaker says or make clear that you do not do so. Again, we would like a more general way to get at the sorts of responsibilities a hearer owes a speaker on an occasion of assertion. Still, I think that this case is instructive. I submit that it gets at a sort of disrespect H exhibits towards S were H simply to disregard S’s assertion that p as having any bearing on the question whether p. This can be illustrated as follows. Suppose that S and H are in a conversation in the course of which S asserts that p. Soon thereafter a question arises which turns on whether p. If in response H reacts in such a way as to make clear he has no opinion regarding whether p, where this attitude does not rest on his having any reasons to be dubious of S’s assertion to that effect, it seems that H has not given S’s assertion its due. Simply put, in asserting that p S presents p as true in such a way as to convey that she is relevantly epistemically authoritative; and in failing to regard S’s assertion as having any bearing on whether p, without having any reason to question whether S actually does have the relevant epistemic authority,14 H offends against S’s implicit claim to have relevant epistemic authority. In order to make clear that H is under some obligation here, and (more specifically) that it is appropriate to talk of the responsibilities that H in his role as a hearer has to S in her role as a speaker, I want to develop the hypothesis that simple disregard of an assertion involves a kind of disrespect. I assume that to disrespect someone is to harm them in some morally relevant way; the question I want to raise concerns the nature of the disrespect involved in a case in which a hearer does not accord another speaker’s assertion the status it deserves. Here I assume that simply disregarding an assertion is a paradigmatic way of not according the assertion the status it deserves. The first matter to be addressed, then, concerns the status that an assertion deserves to be accorded (from a moral point of view). Here I find it helpful to begin with some remarks by several authors who have written about things in this vicinity. I begin first with a terse but highly suggestive remark by Elizabeth Anscombe (1979), who attests to the moral nature of the wrong that is done to a hearer whose assertion is not believed: It is an insult and may be an injury not to be believed. At least it is an insult if one is oneself made aware of the refusal, and it may be an injury if others are. (150)

  14  Such a reason might be a reason to question whether p (say, H has evidence that he thinks S lacks, supporting the hypothesis that ~p); or it might be a reason to question whether S’s evidence is good evidence for p (say, H has evidence that he thinks S lacks, which suggests that S’s evidence for p is defeated, or weak, or misleading, or . . .); or it might be a reason to question whether S was sincere in her assertion that p. These alternatives need not be exhaustive.

Assertion and the Ethics of Belief  269 Glossing Anscombe’s point in a more recent paper, Richard Moran puts matters this way (and here he is expressing a view which he appears to endorse): [T]‌he offense remains even when the speaker’s audience takes his having made the statement to count as evidence for its truth, just as . . . he may take the speaker’s having made [a] promise to make it more probable that he will do the thing in question. The offense lies in his refusing to accept what the speaker freely and explicitly offers him, in favor of privately attending to what the speaker’s action passively reveals, just as someone might refuse an apology while still taking it in this case to be a reliable indication of remorse. What makes sense of such refusals is the fact that acceptance of an assertion or an apology brings with it certain vulnerabilities and responsibilities of its own. Accepting an apology, for instance, brings with it the responsibility to put away one’s resentment, and makes one vulnerable to a particularly bruising possibility of deceit. These risks are avoided by simply taking the apology as more or less good evidence for remorse, and then making of it what one will. (Moran 2007: 301)

So, too, Ted Hinchman notes the “slight” that a hearer exhibits towards a speaker when the speaker’s telling is not accorded its proper status: Imagine A manifestly looks as if he needs to learn the time, so S tells him it’s noon, but A doesn’t regard himself as having thereby acquired any entitlement to believe it’s noon. Imagine not that A regards himself as having acquired an entitlement to believe it’s noon that gets defeated by such background knowledge as that S’s watch tends to run fast, but that A regards himself as not having acquired any entitlement, not even a (now defeated) prima facie entitlement, to believe it’s noon. . . . [Here,] A is failing to acknowledge S—he is, as we say, ‘slighting’ S. . . . [T]‌he explanation of S’s sense of having been slighted . . . [is that] she has tendered an invitation to A to trust her and explicitly been rebuffed. (Hinchman 2005: 565)

Glossing this, Hinchman goes on to say this: We can now account for the nature of A’s abuse when he fails to regard himself as coming to have an entitlement to believe that it’s noon—when he fails, in effect, to take S’s word for it. . . . In telling A that p, S offers him something, an entitlement to believe that p, which she conceives as his for the taking. In recognizing her intention to tell him that p, A satisfies that intention, and S thereby counts as telling him that p. But in refusing to acknowledge the entitlement, he refuses the offer she makes in telling him that p. . . . [W]‌hat’s at stake for S when she tells A that p is his recognition of her as worthy of his trust. (Hinchman 2005: 568)

For her part, Miranda Fricker—the theorist who is most responsible for the recent attention paid to the harms hearers can inflict on speakers by not believing them— casts the matter in terms of the sort of “injustice” that the hearer does to the speaker. Characterizing “testimonial injustice” as involving situations in which “prejudice causes a hearer to give a deflated level of credibility to a speaker’s word” (Fricker 2007: 1), she notes that in such cases: a hearer wrongs a speaker in his capacity as a giver of knowledge, as an informant. . . . [T]‌he primary harm one incurs in being wronged in this way is an intrinsic injustice. Clearly, this harm may go more or less deep in the psychology of the subject, and . . . where it goes deep, it can

270  Sanford Goldberg cramp self-development, so that a person may be, quite literally, prevented from becoming who they are. (Fricker 2007: 5)

Finally, picking up on Fricker’s work, J. Wanderer has more recently spoken of the nature of the “testimonial insult” that a speaker can experience when her testimony is not accorded proper status: I experience testimonial insult when I perceive that your deliberation following my testimony reveals that you did not treat that act as a genuine input to the deliberative process. (Wanderer 2011: 16)

Wanderer goes on to distinguish between two ways that testimony is not treated as a “genuine input into the deliberative process,” with two corresponding “harms” that the hearer does to the speaker. The first obtains when the hearer ignores the testimony, which generates the harm of “depriv[ing] the testifier of an active voice in public discourse” (17); and the second obtains when the hearer rejects the testimony, which does a distinctly “second-personal” sort of harm to the testifier, one which “violates” her “full status as testifier” (19). Here Wanderer remarks that, while Fricker appears to have characterized only the former sort of injustice, what is distinctive of the latter sort is that “the injustice is firmly rooted in a normatively significant second-person relation between two people, in which deliberators do an injustice to testifiers in rejecting them” (19). Several things about these remarks are noteworthy. First, everyone appears to recognize that a hearer H can do harm to a speaker S—H can “insult” S, or cause “offense” to her, or “slight” her, or “violate” her status as a testifier—if H reacts in certain ways to S’s testimony. (I am going to assume that these remarks carry over to S’s assertions more generally.) Second, given the language that these theorists use to describe this infraction, it would appear that this harm is a distinctly moral harm, or at least has a distinctly moral dimension. And third and relatedly, it would appear that at least many, if not all, of the authors recognize that the harm in question is a sort of harm that one particular person does to another particular person. Can we use our guiding assumption, CKNA, to make sense of these aspects of the ethics of assertion? As we saw above, given CKNA assertions can be described as involving a speaker’s presenting a proposition as true in such a way as to convey that she has relevant epistemic authority on the matter at hand.15 In light of this, I submit that an assertion ought to be ascribed a status that is appropriate to its nature as an act in which the speaker has invoked her own epistemic authority. To ignore or dismiss an assertion on insufficient (or no) grounds, then, is to disrespect the speaker as an epistemic subject, by inappropriately repudiating her claim to relevant authoritativeness. We might go further and say that in the envisaged sort of case the speaker is disrespected not merely as an epistemic subject, but also as someone who aims to be offering   15  Indeed, it is for this reason that an assertion that p is, in Wanderer’s apt phrase, “a genuine input to the deliberative process.”

Assertion and the Ethics of Belief  271 help. Since this help pertains to the epistemic needs of the hearer H (at least as these are anticipated by S), and since S’s aim in addressing H is to satisfy those needs, H’s repudiation of S’s assertion is a case of S’s failing to accept H’s role in something that is—or at least which H should regard S as taking to be—a joint epistemic effort. Arguably, the category of epistemic injustice that Miranda Fricker (2007) calls “testimonial injustice” is a special case of this more general category. The more general category is that of being a speaker whose assertions are not accorded proper credence by a hearer; testimonial injustice is the special case in which the unwarranted downgrade in credence reflects the application of a prejudice on the part of the audience. I should say that the harm of which we have been speaking need not exhaust the harm that a hearer H does to a speaker S when H does not accord S’s assertion its due status. In particular, if this failure to accord S’s assertion its due status is itself a public act—one that is observed by others, and recognized by them to be the act of disrespect that it is—then it is quite possible that H thereby harms S’s reputation, her standing in the community. This would be the case when (i) others are aware of H’s ignoring or rejection of S’s assertion (on the basis of no present evidence), and (ii) on the basis of this awareness others conclude that S’s word is not to be trusted. Of course, it will not always be true that others who observe H’s ignoring or rejection of S’s assertion will draw such a conclusion; perhaps they will conclude that H is inappropriately dismissing S. But it can sometimes happen that others draw this sort of conclusion; and when it does, H’s failure to accord S’s assertion its due status harms S (not only by disrespecting S in both her role as epistemic subject and her role as helper in a would-be joint epistemic effort, but also) by affecting her standing as a potential contributor to information-sharing exchanges.16 In this way we have reconstructed the potential two-part nature of the harm described in Anscombe’s terse but suggestive remark: It is an insult and may be an injury not to be believed. At least it is an insult if one is oneself made aware of the refusal, and it may be an injury if others are. (Anscombe 1979: 150)

In sum, I submit that, in addition to the speaker’s having responsibilities—what she owes to the hearer—the hearer, too, has responsibilities—what he owes to the speaker. The principle responsibility is to give to the speaker’s assertion what is its due, and in particular to recognize, and respond appropriately, to the asserter’s implicit claim to have relevant epistemic authority. Precisely what this comes to is a difficult question; again, I suspect that a full accounting will have to take into account the various sorts of context in which assertions are made, the types of relations in which hearers and speakers stand to one another, and so forth. Here my claim here is the minimal one that the ethics of assertion require the hearer not to summarily dismiss the bearing of the speaker’s assertion that p on the question whether p. Failure to live up to this responsibility harms the speaker, first, by disrespecting her status as an epistemic subject in her own right, second, as a participant in what should be seen as a joint epistemic activity,  

  This is the sort of harm explored at length in Fricker (2007).

16

272  Sanford Goldberg and third, by diminishing her reputation among her peers (in those cases in which the hearer’s disrespectful attitude is recognized by others).

14.4  From the Ethics of Assertion to the Ethics of Belief Having developed some claims regarding the ethics of assertion, I now want to move on to introduce some speculations about how my (admittedly somewhat programmatic) remarks about the ethics of assertion might be thought to bear on the ethics of belief. I believe that there are two basic connections between the ethics of assertion and the ethics of belief, corresponding to the two “directions of responsibility” in the assertion case: what the speaker owes to the hearer (section 14.4.1), and what the hearer owes to the speaker (section 14.4.2).

14.4.1  What the Speaker Owes to the Hearer I begin with what the speaker owes to the hearer. Here, the connection between the ethics of assertion and the ethics of belief is underwritten by two claims which, together, serve to link these two domains. The first of the linking claims pertains to the ethics of helping others: giving information to others who need it is a special case of helping others, and as such is governed by relevant ethical norms. The second linking claim pertains to the link between assertion and belief: one’s beliefs ought to be such, that when the need for information arises on a topic on which one has relevant beliefs, one is in a position to be helpful by way of providing the needed information through asserting what one takes to be the case. These two claims, taken together, support the conclusion that the norm of assertion, in the form of CKNA, exerts constraints on the ethics of belief—a conclusion I will be developing below. It should be uncontroversial that others’ needs give us reasons to help them. These reasons are not all-things-considered reasons—they can be defeated. But they are reasons nonetheless. Suppose you see a person holding two large bags of groceries, standing in front of the door, where it is manifest that he is wondering how to get the door open. You are nearby, not in any hurry, with nothing in your hands, and you acknowledge that it wouldn’t be very hard for you to help by opening the door for him. Here, his clear need gives you a reason to help him by opening the door for him; and the fact that you have no other pressing obligations, together with the ease with which you could help him, ensure that this reason is not defeated or overridden by other considerations. You should help him by opening the door. It is very hard to make explicit the principles that govern the generation of these sorts of reasons, and the considerations that can defeat or override them. For my purposes, however, I will assume that this much is uncontroversial: there are occasions on which the fact that another person is in need generates for you a reason to help her; and in situations in which you have such a reason, where it is also the case that you are in a

Assertion and the Ethics of Belief  273 position to help with relative ease, and where there are no other countervailing (defeating or overriding) considerations, provides you with an all-things-considered reason to help her. In putting things in this way, I do not assume that the fact of another’s need always generates a reason for you to help: perhaps there are constraints deriving from the spatial proximity between you and the person in need, or the relationship that holds between the two of you, and so forth. And in generalizing over countervailing considerations, I only assume that there are cases in which these prima facie reasons amount to ultima facie reasons to help. This much, I hope, is uncontroversial. We might describe this as the ethics of helping those in need. One sort of need that can generate reasons of this sort is the need for information. Suppose that a stranger approaches you and asks you for the time. You understand her, have no reason not to oblige her, and take yourself to be relevantly informed (having glimpsed at a clock no more than five minutes ago, and recalling your watch’s observed readout). In this sort of case, her clear need for information gives you a reason to help her by telling her the time; and since you have no other countervailing reasons, you should help her by telling her what time it is. In short: we sometimes have ethical obligations to others that are generated by their needs, and a special case of this obtains when their need is a need for information. Assertion figures in this story as the kind of speech act that is uniquely apt for addressing this need. CKNA enables us to see why this is. To begin, people speak out of all sorts of motives. As a result, a speaker who aims to help another person by providing him with the information he needs is confronted with the problem of acting in such a way as to make this motive clear. In parallel fashion, a hearer who wants his informational needs met is confronted with the problem of determining when another’s speech act aims to satisfy that need. Now both hearer and speaker will recognize that the hearer’s need is a need for true information. And both will recognize as well that the hearer will judge the chances that a speaker is speaking truly by judging the chances that speaker has the relevant epistemic authority in relation to the proposition presented as true. Given these as background, what is needed, if a speaker is to address a hearer’s need for information in a way that the hearer will recognize as aimed at being helpful, is a speech act in which the speaker conveys her relevant epistemic authority regarding the very proposition she presents-as-true in her speech act. In short, what is needed is the speech act of assertion, governed by CKNA. I do not claim that assertion is the only type of speech act through which a speaker can pull off this trick. However, I do claim that the speech act of assertion will typically be the most efficient way to do so. What is more, this fact—that assertion will typically be the most efficient way to do so—is itself a piece of common knowledge. The result is that when a speech act other than assertion is used in these sorts of situations, it can often generate questions in the hearer’s mind—questions that might lead her to wonder whether the speaker meant to communicate something other than what she said. Now I submit that a speaker who, acknowledging her ethical obligations in connection with addressing a hearer’s informational needs, aims to satisfy those needs, ought to do

274  Sanford Goldberg so in a way that avoids unnecessary risks of confusions in the communication. But the risks of confusion that arise when a speaker addresses another’s informational needs through some speech act other than assertion will often be unnecessary risks. This is for the simple reason that any competent speaker has, and will be presumed to have, recourse to the speech act of assertion itself—with the result that the use of a speech act other than assertion can be expected to generate in the hearer’s mind the question noted above. The result is that the speaker who ought to satisfy another’s informational needs on a given occasion will typically be under some pressure to do so by way of making an assertion, rather than through the performance of some other type of speech act. And so we get what we might call a “positive duty” in connection with the “ethics of assertion”: when one has a non-defeated and non-overridden reason to address another’s informational needs, and one regards oneself as having the relevant information in question,17 one ought to do so by way of making the relevant assertion.18 The foregoing considerations provide one basis on which to think of the ethics of assertion. With this as our basis, we can proceed to the ethics of belief by way of the link between assertion and belief. Here my claim is this: one’s beliefs ought to be such, that when another’s need for information becomes manifest (and there are no countervailing considerations), one addresses this need (by way of making an assertion) if and only if one properly takes oneself to have the relevant information. If this is so, I will be arguing, the norm of assertion exerts an indirect constraint on what (from a moral perspective) one ought to believe. In what follows I want to defend two claims, each corresponding to a dimension on which, given the connection between assertion and belief, the ethics of assertion constrains the ethics of belief. The first claim concerns the would-be speaker’s higher-order beliefs, in particular, her beliefs regarding her epistemic position on the matter regarding which the other subject’s need for information has become mutually manifest. To a first approximation, I will be arguing for this conclusion: insofar as a speaker S is a cognitively mature subject who has a prima facie reason to address another’s informational need regarding whether p, then whatever second-order beliefs S has (concerning her own epistemic position regarding whether p) ought to be reasonable. The second claim for which I will be arguing regards S’s first-order beliefs, in particular, her beliefs on the very subject-matter regarding which the other subject’s need for information has become manifest. To a first approximation, I will be arguing that: if (i) S is epistemically rational, (ii) S has a prima facie reason to address H’s informational need regarding whether p, and (iii) there are no countervailing considerations that defeat   17  It may be that one does not have a non-defeated and non-overriding reason to address another’s informational needs unless one reasonably regards oneself as having the relevant information. However, I am not sure whether this is so, so I add this condition at the risk of possible redundancy.   18  One might worry that a speaker has this obligation only if she reasonably regards herself as having the information needed. (Does someone who thinks he knows everything really have a reason, even a merely prima facie reason, to help anyone in need of information?) This may be; if so the reasonability constraint should be added to what I say below. I will not explore this further here.

Assertion and the Ethics of Belief  275 this reason (save perhaps S’s lacking the information sought), S ought to address this need by asserting that p if and only if she is relevantly informed as to whether p. As I develop this line of argument below, it will become clear how the norm of assertion (understood in terms of CKNA) exerts constraints on the ethics of (first-order) belief. Suppose that it is mutually manifest to S and H that H is in need of information regarding whether p, that S is nearby, and that it would not be particularly difficult or costly for S to provide that information to H were she (S) to have it. S should do so (by way of making the relevant assertion) if and only if she is relevantly informed regarding whether p.19 To wit: if S is relevantly informed but does not share this information with H, she fails to help one in need, under conditions in which she was in a position to do so (and where there were no countervailing considerations). If S aims to help H by addressing his need for information regarding whether p, but is not herself relevantly informed, S’s attempt will not succeed at being helpful, and may even be harmful (as when S’s assertion is false and H, believing S, comes to act on the false belief, with bad consequences). As a result, if S is to determine whether she should help H on this occasion, she must determine whether she is relevantly informed. Now let us restrict our attention to those cases in which this determination takes the form of S’s forming higher-order beliefs regarding her state of informedness on the matter at hand. Insofar as these beliefs will determine whether she takes herself to be relevantly informed, and so will determine whether she regards herself as under an obligation to address H’s need for information, ideally these beliefs ought to represent herself as relevantly informed if and only if she is relevantly informed. Of course one’s own state of informedness is not something that is transparent to one: there are cases in which one believes that p on the basis of good (but not infallible) evidence, where it turns out that it is not the case that p; and there cases in which one believes that p on the basis of good (but not infallible) evidence, where it turns out that one’s belief is true, but the truth of the belief is a matter of luck (Gettier cases). Presumably, one is not always, or perhaps even ever, in a position to rule these kinds of scenario out, while at the same time responding in a timely fashion to another’s need for information. With these sorts of possibility in mind, we might say this: insofar as S determines whether she is relevantly informed whether p, and does so by way of forming higher-order beliefs about her state of informedness, she has a moral duty to be reasonable in the higher-order beliefs she forms. To appreciate the basis of this duty, and to see what the demand of “reasonableness” comes to, consider a case in which S forms higher-order beliefs about her informedness, where these higher-order beliefs fly in the face of the evidence she herself has. We can consider two cases. In one she regards herself as relevantly informed when in fact she is not (the “false positive” case); in the other she regards herself as not informed when in fact she is (the “false negative” case).   19  To be relevantly informed is to mentally represent how things are in a way that is both correct and epistemically satisfactory. More on the notion of epistemic satisfactoriness below.

276  Sanford Goldberg Taking the false positive case first, assume that S wrongly regards herself as in a position to be helpful in addressing H’s need for information, where as a result S’s assertion is either false or unwarranted. If H acts on the information S provides, only to find out that S was actually uninformed on the matter, I submit that H is in a position to resent S so long as S’s state of actual uninformedness was something S ought to have discerned at the time. Harkening back to something I argued in section 3, we can speak in terms of what H can reasonably have expected of S: H is entitled to regard S as having done all that can reasonably be expected of her in connection with satisfying the norm of assertion. Given this, H is entitled to suppose that any higher-order beliefs S formed in this connection captured all that H could reasonably have expected of S in this regard. While it is hard to say precisely what this comes to, it is clear how S might come up short. For example, if it was mutually manifest that a particular ~p scenario, α, was “live” in the context in which H was in need of information, then insofar as S (believing herself to be relevantly informed) asserts that p, H is entitled to regard S’s epistemic position as being such as to be able to rule out α. If this is not so, H is entitled to resent S for an unwarranted assertion, and so (by extension) is entitled to resent S for regarding herself as having been in a position to help in the first place. S ought not to have believed herself to be relevantly informed, given the circumstances. Next, taking the false negative condition, assume that S wrongly regards herself as uninformed, under conditions in which in fact she is informed. To make sense of this case, let us imagine that S regards herself as uninformed, not because she fails to recognize that she herself believes that p, but rather because, while she recognizes that she believes that p, she regards this belief as not based on grounds that are sufficiently good to warrant an assertion that p, but where this higher-order belief is false (her grounds are assertion-warranting). Then, even though it is mutually manifest that H needs information regarding whether p, S does not tell H that p. Perhaps S is silent, not saying anything at all. Or perhaps S enters a hedged assertion, as in: “Well, I’m not sure, but I think that there are good reasons to suppose that p.” Although both of these sorts of cases can be handled in the same way, it will be easier to discuss the silence case first. If S is silent, saying nothing on the issue whether p, I submit that H is entitled to resent S so long as S’s higher-order beliefs regarding her own state of informedness failed to capture something H could reasonably have expected of S in this regard. So suppose that it was mutually manifest both that H’s need was acute, and that S was the person best position to weigh in. Under these conditions, H was entitled to regard S’s silence as indicating that there was no relevant information to be had. To the extent that S had (and recognized that she had) evidence bearing on whether p, then, H is entitled to resent her for not speaking up. If this evidence was sufficiently good as to clearly warrant assertion, H is entitled to resent S for not having informed him that p, and so is entitled to resent S for her having formed false higher-order beliefs regarding her own state of (un)informedness. (H to S: “Why didn’t you tell me that p?” S to H: “Because I didn’t know that p.” H to S: “You shouldn’t have thought that you didn’t know. After all, you had excellent evidence in support of p, and you had no relevant

Assertion and the Ethics of Belief  277 evidence against p!”) But what should we say of those cases in which S’s evidence is clearly assertion-warranted, but where S, having false higher-order beliefs on this matter, enters only a hedged assertion? In most cases of this sort, I suspect that S’s hedged assertion discharges whatever obligations she has in connection with being helpful regarding H’s need for information. Still, cases can be imagined in which S’s epistemic position on p was so good, that she ought not to have hedged—she ought to have simply asserted that p straight-out—where her not having flat-out asserted that p generates doubts in H’s mind that S ought not to have generated. I will not discuss these sorts of case further, save by expressing my thought that here, too, S may be guilty of having second-order beliefs she ought not to have had (namely, regarding the strength of her epistemic position on p). Let me try to formulate more carefully the conclusion for which I have been gesturing. The thesis I have been gesturing towards is this (“EHOB” for the “Ethics of Higher-Order Belief ”): EHOB Given two subjects S and H and circumstances in which (a) it is mutually manifest that H is in need of information regarding whether p, (b)  (bracketing whether S has the relevant information) there are no considerations that defeat or override the reason S has to help H in this regard, and (c) S is a cognitively mature agent who has the capacity for higher-order belief, any higher-order beliefs S forms regarding her state of informedness on whether p ought to capture all that H can reasonably expect of S in this regard. My case for this has been as follows. First, when (a)–(c) are true, then she ought to provide H with that information (by making the relevant assertion) iff S is relevantly informed. But to determine whether she is in this position, S must have some sense of her epistemic position on the question whether p. This is a matter of performing a reasonable self-assessment of her state of informedness. And given (c) this will take the form of higher-order beliefs regarding her epistemic position on p. In particular, she ought to determine whether she is in a position to make a warranted assertion on this topic (where this is a matter of satisfying the norm of assertion). And so we see that under conditions (a)–(c), S ought to form reasonable higher-order beliefs in connection with the question whether she is in a position to make a warranted assertion on the question whether p. It is perhaps not particularly controversial that, given that assertions can be made in an attempt to help others in their informational needs, the ethics of assertion might bear on one’s higher-order beliefs concerning whether one is in a position to assert warrantedly. More surprising perhaps is that the ethics of assertion may constrain what one ought to believe on first-order questions. Although I am not in a position to argue for a strong conclusion in this connection, what I can do is argue for a conditional thesis: if a currently popular (although not uncontroversial) assumption about the conditions under which evidence warrants belief is true, then the ethics of assertion constrain the ethics of belief even on first-order belief contents. I will introduce

278  Sanford Goldberg the relevant assumption about epistemic rationality below. With it, I will argue for the following conclusion (“ABL” for “Assertion-Belief-Link”): ABL Given two subjects S and H and circumstances in which (i) it is mutually manifest that H is in need of information regarding whether p, (ii) (bracketing whether S has the relevant information) there are no considerations that defeat or override the reason S has to help H in this regard, and (iii) S is epistemically rational,20 then S ought (morally) to address H’s need for information (regarding whether p) by asserting that p iff S’s evidence warrants the belief that p. Once I establish ABL, I will then go on and use it to show how the ethics of assertion bear on the ethics of belief. In order to establish that ABL holds, I need to establish that if (i)–(iii) hold, then the following Assertion-Belief biconditional holds: AB ≡ S ought (morally) to address H’s need for information (regarding whether p) by asserting that p iff S’s evidence warrants the belief that p.21 In order to establish this I will need to appeal to another biconditional regarding the ethics of assertion. I will call this biconditional the “Ethics of Assertion biconditional”: EA ≡ S ought (morally) to address H’s informational needs (regarding whether p) by asserting that p iff S occupies an epistemic position in which she can rule out all relevant alternatives to [p]‌. (Here, an alternative is “relevant” when it is relevant in the conversational context.) My claim will be that the Ethics of Assertion biconditional, EA, holds whenever (i)–(iii) hold. Once I establish this, I will appeal to EA as part of an argument to the effect that if (i)–(iii) hold, AB holds. To see that EA holds whenever (i)–(iii) hold, assume that (i)–(iii) hold. We can now break our task in two, corresponding to the left-to-right and the right-to-left direction of EA. For the left-to-right direction, assume that the LHS of EA holds: S ought (morally) to address H’s informational needs by asserting that p. Now we can note the following truism about what one ought to do, when one ought to help another person: when one ought to help H in respect R, one ought to do so by way of φ-ing only if φ-ing can be reasonably expected to help H in respect R.22 (If you should help an elderly person cross the street, then you should do so by way of gently guiding the person by the arm only if gently guiding the person by the arm can reasonably be   20  The need for this epistemic rationality condition may not be immediately obvious; it will emerge below.   21  I distribute the “ought” here so that this principle can be read as follows: if S ought (morally) to address H’s informational needs (regarding whether p) by asserting that p, then S’s evidence warrants the belief that p; if it is not the case that S ought (morally) to address H’s informational s (regarding whether p) by asserting that p, then it is not the case that S’s evidence warrants the belief that p.   22  This truism should probably be relativized to circumstances, but I will not bother doing so here.

Assertion and the Ethics of Belief  279 expected to help the person cross the street.) Applied to the present ‘ought’, we get: S ought (morally) to address H’s informational needs by asserting that p, only if S’s asserting that p can reasonably be expected to help address H’s informational needs. But it is also clear, in turn, that S’s asserting that p can reasonably be expected to help address H’s informational needs only if it is also true that she can rule out all relevant alternatives to [p]‌. If this is not so—if S cannot rule out all relevant alternatives to [p]—then her asserting that p in this context cannot reasonably be expected to address H’s informational needs, for the simple reason that in such a case S’s asserting that p would expose H to epistemic risks. In particular, it would expose H to the risk a situation involving the relevant alternative(s) which, by hypothesis, S can’t rule out. Were such a situation to obtain, S’s asserting that p, far from being helpful to H, would be positively harmful to H—for H might well have proceeded, reasonably enough, to assume that this act of S’s authorizes H to regard all relevant alternatives as ruled out. And so we see that if S’s assertion that p is to be helpful, she must be in a position to rule out all relevant not-[p] alternatives. In short, under conditions in which (i)–(iii) hold, if S ought (morally) to address H’s informational needs by asserting that p, then S occupies an epistemic position in which she can rule out all relevant alternatives to [p]. Now, to establish the right-to-left direction of EA, assume that the RHS of EA holds: S occupies an epistemic position in which she can rule out all relevant alternatives to [p]‌. But then, since by (i)–(iii) she is under an obligation to meet the informational needs of H if she has the relevant information, she ought to assert that p. But she does have the relevant information, since she can rule out all relevant not-[p] alternatives. And so we see that the right-to-left direction holds as well, and with this completes the argument that EA holds whenever (i)–(iii) hold. I will now appeal to this fact in order to argue for ABL. In particular, my claim is that if (i)–(iii) hold, then, given a plausible assumption regarding epistemic rationality, AB holds. To do so, let us again assume (i)–(iii). And let us assume the LHS of AB as well: S ought (morally) to address H’s informational needs by asserting that p. Since S ought to assert that p, we get the result (from EA) that S occupies an epistemic position in which she can rule out all relevant alternatives to [p]‌. But to occupy this position is to have an epistemic standing that warrants belief. But in such a case it would be irrational for S herself not to believe that p: having any attitude other than belief that p would fly in the face of the evidence that she has—evidence that enables her to rule out all relevant not-[p] alternatives. So we see that, on the assumption of (i)–(iii), if S ought (morally) to address H’s informational needs by asserting that p, then S’s evidence warrants the belief that p. Now (to get the right-to-left direction of the rule in AB), we assume again that (i)– (iii) hold. However, rather than assume the RHS of AB in order to derive the LHS, I will instead assume that the LHS of AB does not hold, in order to derive that the RHS does not hold either (which is equivalent to what is wanted). So, assume that the LSH of AB

280  Sanford Goldberg does not hold: it is not the case that S ought (morally) to address H’s informational needs by asserting that p. Since this is also the denial of the LHS of EA, the RHS of EA is false as well: it is not the case that S occupies an epistemic position in which she can rule out all relevant alternatives to [p]‌. Now consider the following principle regarding Evidence sufficient to Warrant Belief (“EWB”): EWB If S’s evidence does not rule out all relevant not-[p]‌alternatives, S’s evidence does not warrant the belief that p. Although EWB is not uncontroversial,23 it does seem to have something going for it. After all, the point (or at least a central point) of belief is to guide action (e.g., by providing premises for practical syllogisms), and all else equal it would seem unwise to employ [p]‌in a practical syllogism (or to act on the assumption of p) if one’s evidence does not rule out all relevant not-[p] alternatives. But rather than argue for EWB, I will regard my conclusion as conditional on the assumption of EWB. In short: if EWB is true, then, given (i)-(iii) and the negation of the LHS of AB, we get the negation of the RHS of AB, which is the desired conclusion. Putting our two results together, we get the following: assuming that EWB is true, then in any situation in which (i)–(iii) hold, S ought morally to address H’s informational needs by asserting that p iff S’s evidence warrants the belief that p. But this is just to say that, given ER, the Assertion-Belief biconditional holds whenever (i)–(iii) hold. So concludes my case for the assertion-belief link, ABL. The time has come to show how, given ABL, the ethics of assertion imposes some (weak) constraints on that of belief. Suppose that ABL is true and that conditions (i)– (iii) hold. Then if S’s evidence warrants the belief that p, S ought (morally) to address H’s need for information (regarding whether p) by asserting that p; but if S’s evidence does not warrant the belief that p, then S ought not to address H’s need in this way. This gives us a way of determining, independently of what S actually believes, whether she ought to address H’s informational needs by asserting that p: we simply assess S’s epistemic standing on the question whether p. Now consider two possibilities that might obtain. The first is that we determine, independently of what S actually believes, that she ought to address H’s informational needs by asserting that p. But if S ought to address H’s informational needs by asserting that p, S herself ought to believe that p, for the simple reason that if she doesn’t, then her assertion would be insincere (and she ought not to be insincere). Here is a case in which what S ought to believe is determined by what S ought to assert (in order to address a hearer’s informational need). Next, consider a second possibility: that we determine, independently of what S actually believes, that she ought not address H’s informational needs by asserting that p. In this case, S herself ought not to believe that p. This verdict reflects an assumption   23  Some will hold that lottery propositions (for a sufficiently large lottery) falsify ER: it is not epistemically irresponsible for a subject to believe that her ticket is a loser even when her evidence does not rule out that her ticket is a winner, but instead merely makes it very improbable that she will win.

Assertion and the Ethics of Belief  281 regarding an empirical connection between belief and assertion, whereby it is human nature to have a strong disposition to express one’s beliefs in (assertoric) speech. For this reason, if one is under conditions in which one’s epistemic standing doesn’t warrant the assertion that p, then to believe that p is to unduly risk exposing others to the risk of unwarranted—and so unhelpful—assertion. Thus, one should not believe that p in these circumstances, since one should not unduly risk exposing others to such risks.24 In effect, the foregoing argument appeals to the ethics of assertion, together with links between belief and assertion, in order to argue that one ought to believe only what one would be warranted in asserting. Failure to comply would put one in a situation in which one would not be placed to be helpful, were the occasion to address another’s need for information to arise. Seen from one angle, our result is not that surprising. After all, we systematically depend on one another for information, and since each of us is guided in our information-sharing practices by what we take to be true—that is, by our beliefs—we ought to believe only what would be helpful to provide to those seeking information. Seen from another angle, however, our result is surprising: what one believes is typically thought to be both a “private” affair as well as an affair over which subjects have little control, whereas what one asserts is a public affair in which the speaker is responsible for her actions. How can the norms of the latter generate ethical constraints on the former? There are big questions here that I cannot claim to have addressed. Still, at least this much seems right: assertions have a collection of properties that make them a plausible candidate for grounding some ethical constraints on belief. In particular, assertions are the standard way by which to address others’ need for information, they are the standard way by which linguistic creatures express our beliefs, and human beings have a strong disposition to express what we believe in this way. For this reason, that the norms of assertion constrain the ethics of belief is not much more surprising than is the claim that there are ethical constraints on how we ought to help others who seek information. I can’t pretend to be fully confident that I have articulated a sound route from the norm of assertion to the ethics of belief. I hope instead that what I have presented here is suggestive. In particular, I hope it encourages others to consider the route from the norm of assertion to the ethics of belief, where the connecting link comes in the form of the ethics of addressing others’ (informational) needs. If the foregoing argument merely succeeds in attracting attention to the prospects for an argument of this sort, I would be satisfied—even if the argument as I have presented it fails to convince.

  24  We might add: on the assumption that (ceteris paribus) one acts on what one believes, in believing that p under these circumstances S thereby unduly risks exposing others to liability in any joint effort in which the question whether to act on p is involved. (One ought not to unduly risk exposing others in this way.) I put this point in a footnote, though, since it does not derives from the ethics of assertion.

282  Sanford Goldberg

14.4.2  What the Hearer Owes to the Speaker I turn, finally, to the last issue I will discuss: how what the hearer owes to the speaker constrains the ethics of belief. Happily, here I can be brief. Given CKNA and what is mutually known in a context of an observed assertion, the hearer owes the speaker the respect that is her due—respect that acknowledges her as someone who has invoked her own epistemic authority on the matter at hand. In section 14.3 I framed this in terms of the hearer’s responsibility to ascribe to the assertion its due—to take it as an act in which a speaker presents a proposition as true in such a way as to invoke her own relevant epistemic authority, and to treat it accordingly. To fail to do this is to treat the speaker with disrespect. What does any of this tell us about the ethics of belief? More specifically, how can the mere fact that one must accord another speaker’s assertion its due constrain what one oneself ought to believe? I submit that the obligations here are only indirect, but nevertheless real. They involve the attitudes we as hearers should have towards any observed assertion: they are not to be regarded as mere evidence, but rather as a claim to truth that purports to be backed by another epistemic subject’s relevant authority. Since this might seem more controversial than I intend it to be, I will conclude by stating in programmatic terms the view I mean to be endorsing. One can come to know all sorts of things from another’s assertion. In hearing S assert that it is cold outside H can come to know: that S would like the heat turned up, that S has a cold (her voice is scratchy), that S is anxious (H knows that S only comments on the weather when she is anxious), that S speaks English, and so forth. In none of these cases does H’s knowledge depend on S’s relevant epistemic authority. Indeed, in none of these cases does H even need to assume that S takes responsibility for having that authority. It is when H does assume that—roughly: when H accepts what S asserts, because S so asserted—that H takes on the sort of attitude that is appropriate to proper assertion. My claim, then, is this: when S asserts that p, H cannot simply regard the event as one of S’s producing noise, regarding which H can then ask what sort of evidence S’s having done so provides H. Rather, H must regard S’s assertion for what it is: an act whereby S presents a particular proposition as true, in such a way as to convey that she has the relevant epistemic authority on the matter. I rather doubt that this amounts to a requirement that H explicitly believe something of this sort. (Few of us ever do form such beliefs.) Rather, the obligation is this: one must so regard assertions, that one manifests that one ascribes to them something like the status just described. One can do this without forming an explicit belief. Perhaps it is manifested in how one would defend oneself, were one to refrain from accepting the assertion (one would recognize the need to have reasons for skepticism); perhaps it is clear in how one would defend oneself, were one to accept the assertion (one would recognize one’s dependence on the speaker’s authority). The point is simply that one cannot simply ignore or reject an assertion on no grounds whatsoever. One must

Assertion and the Ethics of Belief  283 guide oneself in testimonial belief-formation so as to manifest one’s sensitivity to other speakers as epistemic subjects in their own right.

References Anscombe, E. (1979). “What is It to Believe Someone?” in C. F. Delaney (Ed.), Rationality and Religious Belief. South Bend, IN: University of Notre Dame Press, 141–51. Bach, K. and Harnisch, R. (1979). Linguistic Communication and Speech Acts. Cambridge, MA: MIT Press. Fricker, M. (2007). Epistemic Injustice:  Power and the Ethics of Knowing. Oxford:  Oxford University Press. Goldberg, S. (2011). “Putting the Norm of Assertion to Work: the Case of Testimony,” in J. Brown and H. Cappelen (Eds.), Assertion: New Essays. Oxford: Oxford University Press, 175–96. Hinchman, T. (2005). “Telling as Inviting to Trust,” Philosophy and Phenomenological Research 70(3): 562–87. Lackey, J. (1999). “Testimonial Knowledge and Transmission,” The Philosophical Quarterly 49: 471–90. Lackey, J. (2007). “Norms of Assertion,” Noûs 41: 594–626. MacFarlane, M. (2011). “What Is Assertion?” in J. Brown and H. Cappelen (Eds.), Assertion: New Essays. Oxford: Oxford University Press, 79–96. Moran, R. (2007). “Getting Told and Being Believed,” in J. Lackey and E. Sosa (Eds.), The Epistemology of Testimony. Oxford: Oxford University Press, 272–306. Owens, D. (2006). “Testimony and Assertion,” Philosophical Studies 130: 105–29. Pagin, P. (2011). “Information and Assertoric Force,” in J. Brown and H. Cappelen (Eds.), Assertion: New Essays. Oxford: Oxford University Press, 97–136. Ross, A. (1987). “Why Do We Believe What We are Told?” Ratio 28(1): 69–88. Stalnaker, R. (1978). “Assertion,” Syntax and Semantics 9: 315–32. (Reprinted in Stalnaker 1999: 78–95.) Stalnaker, R. (1999). Context and Content. Oxford: Oxford University Press. Wanderer, J. (2011). “Addressing Testimonial Injustice:  Being Ignored and Being Rejected,” Philosophical Quarterly 62: 148–69. Watson, G. (2004). “Asserting and Promising,” Philosophical Studies 117(1): 57–77. Weiner, M. (2005). “Must We Know What We Say?” Philosophical Review 114(2): 227–51. Williamson, T. (2000). Knowledge and Its Limits. Oxford: Oxford University Press.

15 Evidence of Evidence is Evidence Richard Feldman

15.1 Introduction It will be useful to begin this discussion by disentangling a few of the many questions epistemologists might ask about rational responses to peer disagreement.1 The one that initially caught my attention is this: Q1) Can people involved in peer disagreement reasonably disagree? Commonplace examples bring out what I find intriguing about this question. In numerous ordinary situations, generally reasonable, well-informed, well-intentioned people disagree. Some topics—politics or religion—can generate heated conflicts. Some topics are largely about matters of taste, as in disagreements about the merits of a movie or a bottle of wine. Some disagreements result from different preferences or attitudes toward risk, as when two people differ over the “best” ways to invest one’s retirement account. In some of these cases, it may be that people don’t disagree, at least not in the sense of holding conflicting doxastic attitudes toward the same proposition. But in some cases they do. That is, in some cases, one generally reasonable, well-informed, well-intentioned person believes a proposition and another person with those same characteristics does not. One question, then, about disagreement is whether both people in such a situation can be reasonable in their own beliefs. That’s what (Q1) is intended to ask. Of course, there are many ways in which (Q1) can be made more precise, largely by specifying what counts as a “peer” and what counts as “reasonable.” In later sections of this chapter, I will say something more about interpretations of these terms. But before doing that, I will consider some other nearby questions. The second question is this: Q2)  How should one rationally respond to learning of a peer disagreement?   1  Some of the discussion in this section draws on Feldman (2006).

Evidence of Evidence is Evidence  285 (Q2) seems to be very closely connected to (Q1). I’ll argue later that there is at least a little distance between them. (Q2) asks what a person should rationally do upon learning that a peer disagrees; it asks how beliefs should be revised in the light of disagreement. As formulated, (Q2) invites the assumption that there is a uniform answer to the question. For example, it suggests that the answer might be something like “One should always stop believing what one previously believed.” Of course, the right answer might not be anything like that. The answer might be that the rational response can vary, depending upon the details of the situation. There is a seemingly closely connected third question: Q3) What should a participant in a peer disagreement think about the rationality of the peer? This latter question has received less attention in the literature as it has developed. But I find it intriguing. In part, this is because there is an appealing answer suggested by the common idea that “reasonable people can disagree.” The idea is that, in at least some common cases, one should think that one’s peer is reasonable in believing as he or she does, but that one’s own belief is reasonable as well. Again, there is no reason to think that there is a simple fully general answer along the lines of “Always think that the peer is reasonable,” but a possibility is that it is at least typically appropriate to have that belief. And this answer to (Q3) seems to imply that the answer to (Q1) must be “yes.” For it suggests that, at least in the cases under discussion, the correct view is that both people are believing reasonably, i.e., they reasonably disagree. And this suggests an answer to (Q2): since, in these cases, both peers are reasonable, there is no need for either peer to modify his or her beliefs as a result of the disagreement. A secondary theme in parts of this chapter is that the connections between these questions are somewhat more complex than one might think. (That is, more complex than I thought they were.) The other theme, and the real focus of the chapter, is that the answers to the questions turn largely on the significance of evidence about evidence. This could also be called “higher-order evidence.”2

15.2 Evidentialism In the discussion that follows, I will assume without argument a general view about what makes a belief reasonable. That view is evidentialism.3 In this section, I will characterize the view and briefly discuss a few key consequences. Roughly, evidentialism is the view that epistemic justification or rationality or reasonableness depends on evidence. Another way to state it is to say that one should follow one’s evidence, that is one should believe in accordance with the evidence one   2  I previously discussed this topic in Feldman (2009). Sections 15.1–4 of this chapter are drawn from that discussion. See also Kelly (2010).   3  This view is developed in Conee and Feldman (2004).

286  Richard Feldman has. Although there are differences between “justification” and “rationality” and “reasonableness” and “should believe” and what I say here glosses over those differences. However, I believe that any such differences do not significantly affect the points I want to make here. In any case, here is a formulation of evidentialism: EVI. A doxastic attitude is justified for a person if and only if the attitude fits the person’s evidence:  When one’s evidence is counterbalanced, suspension of judgment is the justified attitude. When the evidence better supports a proposition than its denial, belief is the justified attitude; when it is the denial that is better supported, disbelief is justified. The evidence a person has is the information the person has to go on in forming beliefs. In the typical case, other justified propositions might capture much of this. In my own view, a person’s evidence ultimately comes down to the person’s experiences (including memories). There are many important details to be worked out about this in order to provide a full theory. One key issue, for example involves the status of stored memories. Do things that one is not currently thinking about, but does have in memory, count as part of one’s evidence? Fortunately, for present purposes, it is not necessary to settle these details. For the most part, in the cases under consideration, the relevant evidence will be relatively explicit. A second key concept in the formulation of evidentialism expressed in EVI is that of evidential “support.” The basic idea here is that a body of evidence supports, or fails to support, a proposition. In the formulation of evidentialism that I prefer, evidential relations are “objective” and necessary. In other words, evidential relations are necessary “quasi-logical” relations among propositions, or perhaps experiences and propositions. On this view, it is not a contingent fact about a body of evidence that it supports a particular proposition. A simple example will illustrate the idea and an important consequence. Suppose I know that my colleague, J, drives a red Toyota and regularly parks in a particular place when she’s on campus. I arrive on campus and see a red Toyota just like hers parked in her spot. I might then conclude that J is on campus and say that my supporting evidence is the fact the appropriate kind of red Toyota is in her parking spot. On my view, this is not exactly right. The fact that a car of that kind is in that spot is not, by itself, supporting evidence for the conclusion that J is on campus. That fact, conjoined with the background evidence that she drives that kind of car and typically parks in that spot when she’s on campus, does support my conclusion. Frequently, on my view, when we state our evidence for a proposition, we in fact cite only a part of that evidence. And the part of the evidence may not be, on its own, supportive of the conclusion. Thus, Parts of a reason (or evidence) for a conclusion are typically not themselves a reason or evidence for that conclusion. Roughly, by itself the simple evidence is not (usually) by itself evidence at all for the conclusion. In conjunction with the relevant additional information, often but not

Evidence of Evidence is Evidence  287 always common knowledge that need not be mentioned in conversation, it is evidence. Or, more carefully, the conjunction of the stated proposition and the background knowledge is evidence. This point will matter in some of the arguments to follow.

15.3  Evidentialism and Disagreement To the extent that there is a general evidentialist answer to questions about what you should do in response to learning of peer disagreement, it is this: follow your evidence. There is no reason to abandon evidentialism simply because there are cases in which seemingly reasonable peers disagree. In fact, evidentialism seems to me to provide exactly the right way to think about disagreement. It instructs us to ask how learning about a disagreeing peer affects one’s evidential situation. It asks us to reflect on what one should think, now that one has this new information about the disagreement. Some rival views about justification, such as reliabilism or proper functionalism, just seem to me to miss the mark. The literature on disagreement contains extensive discussion of principles that say such things as that when confronted with a disagreeing peer, one should always suspend judgment, or split the difference, or stick to one’s guns.4 But I think that if evidentialism is correct, then it is almost surely true that there are no special epistemic principles about disagreement. If there were any such special epistemic principles about what is justified in cases of disagreement, then either they undermine evidentialism (because the justified attitude is not always the one supported by the evidence) or they are implications of evidentialism (because the evidential impact of the evidence one gets from a disagreeing peer always yields the outcome specified by the principle). To take an example, a special principle might say that in cases of peer disagreement, the reasonable thing to do is to suspend judgment about the disputed proposition. If this is not in conflict with evidentialism, then it must be that the evidence obtained in the disagreement always has the result that one’s total evidence is neutral. This brings to the forefront the question highlighted at the end of the introductory section, on the evidential impact of encountering a peer who disagrees. I think that there is a strong case to be made that there are no such general principles, though much depends upon what sorts of principles would count. Also, much depends upon how the term “peer” is understood. I turn next to a discussion of these issues. Under discussion, then, is whether there will be any true principles of the form: If a person, S, learns that a peer disagrees about proposition P, and . . . then it is reasonable for S to . . .

  4  See Feldman (2003), Christensen (2007), Elga (2007), Bergmann (2009), Kelly (2010), and Lackey (2010) for an array of examples.

288  Richard Feldman Specifications of the antecedent of such a principle might add further details about the conditions in which the disagreement arises. The consequent might specify the attitude S should adopt toward P or how S ought to adjust his attitude toward P. Such principles might help in the derivation of answers to the questions with which this chapter began, especially (Q1) and (Q2). For example, one simple specification adds nothing to the antecedent and states in 'the consequent that suspension of judgment is the justified attitude: If a person, S, learns that a peer disagrees about proposition P (i.e., S learns that S believes P and S* disbelieves P and that S* is S’s peer), then it is reasonable for S to suspend judgment about P.5 Given some additional modest assumptions, this principle would imply that suspending judgment is also reasonable for S*. And this would yield the result that in the cases covered, S and S* should have the same attitude, and thus they cannot reasonably disagree. Hence, this yields an answer to (Q1) on one of its interpretations. In thinking about this, it will be helpful to focus on the concept of a “peer.” One can give the concept a kind of philosophical purity by building a lot into it—so much that there are no peers, even oneself over time. For example, you could say that peers must have exactly the same evidence. Given this, plus evidentialism, plus what I’ve called the Uniqueness Thesis,6 you can deduce that two peers should always have exactly the same attitudes. Hence, two peers who discover that they disagree should have exactly the same attitudes. It does not yet follow that they should suspend judgment; it’s possible that their shared evidence supports believing, or disbelieving, the proposition in question. One might think that the evidence of the disagreement has a neutralizing effect, making suspension of judgment the right attitude. Even if this is not right, the result that the peers cannot reasonably disagree remains in place: since they are peers, they have (on the current interpretation) exactly the same evidence, and thus should have the same attitudes. However, elevating the word “peer” in this way robs the discussion of what, for me, was most interesting about the topic of disagreement. No real or realistic disagreements will amount to peer disagreement, in this sense. Even after a long, detailed, and thorough discussion, people will not have exactly the same evidence. There might be background information not shared. Their own experiences will not be the same. Even if they’ve described their experiences to one another, at best each has only a second-hand account of the other’s experiences. Thus, if “peer” is understood in this way, then the topic of “peer disagreement” is not about disagreements that actually occur. The kinds of examples that provoked the discussion for me are not covered by principles like this.

   

5

  I came dangerously close to endorsing this principle in Feldman (2006).   For discussion, see Matheson (2011). I discussed the topic in Feldman (2007).

6

Evidence of Evidence is Evidence  289 It is possible to relax the conditions on peerhood so that people who are alike in the ways that real people who are relevantly similar to one another count as peers. Thus, you might be a peer of someone who has a fairly similar educational and social background and shares a general worldview with you. This leaves open the possibility that peers have different evidence. And this leaves open the possibility that peers, even peers who know that they disagree, can nevertheless have different justified attitudes. Now, if a philosopher is eager to identify some special epistemic principle about disagreement, then that philosopher might add more clauses to the antecedent of the principle in order to produce a truth. Strengthening the antecedent and, perhaps, weakening the consequent might also yield a truth. For example, one might reformulate the principle so that the consequent does not specify that the peers must suspend judgment but only that they must believe (or disbelieve) the proposition with less confidence than they previously did.7 I do not believe that any fully general principles along these lines whose antecedent conditions are actually satisfied in many realistic cases will in fact be true. But that is not something that I can establish, and I will not attempt to do so here. What I do want to emphasize is that anyone who is content to settle for the general evidentialist injunction to follow the evidence will have no need to formulate a special principle about what to do in cases of disagreement. This will eliminate the need to define “peer” precisely. Such philosophers will settle for case-by-case evaluation, with the possibility that there will be clusters of similar cases about which lower level generalizations can be formulated. The question in each case is: what should one think after learning that this person disagrees. It might matter whether the other person is generally better informed or smarter than you, or whether the person is more apt to make mistakes than you. But these are just facts about the person that enter into the calculation. It is worth noting that the views of someone is not a peer, or you do not regard as a peer, can also have an impact on what it is reasonable for you to believe. There will not be any need to have a precise definition of “peer.” I think that there is something of value in thinking about peer disagreement, where “peer” is given an informal, somewhat vague, interpretation. On this interpretation, one’s colleagues are generally one’s peers. People far less informed on a topic are not. But you just need not worry about getting any of this precise. If someone who is, in this sense a peer, says something surprising to me—denies something I thought all of “us” believed—then, I think, that matters. I should take that fact into account. But I do not need some precise definition of “peer” to do that. I think that it is not a problem that there are not precise special principles about what to do in cases of peer disagreement. I think that the same is true with respect to many other kinds of cases. For example, I believe that there are no fully general principles about what one should believe upon learning that a proposition has been reported as

 

  For discussion, Matheson (2009).

7

290  Richard Feldman true by a reliable news source. How one should respond always depends upon one’s other evidence. The generally reliable news source might erroneously report something about which I have direct and conclusive evidence. The report might make absolutely no difference to what I should think. One might try to formulate some principle that is immune to counterexample by modifying the conditions in which one gets the news, but I fail to see the point of the effort. What remains true, as I will discuss in the next section, is that in getting the news report, or in learning of a disagreeing peer, one does get evidence of notable sort. It is possible, I think, to say something about this evidence. And that will provide some help in thinking about the questions identified at the beginning of this chapter.

15.4  Evidence Obtained in Disagreements In actual cases in which one discusses a topic with a peer with whom one disagrees, one can obtain evidence of various kinds. During a substantive discussion of an issue, one might get some ordinary, first-order evidence about the subject matter. Your peer might simply provide some facts, or testify to some facts, about whatever it is at issue. Similarly, the peer might describe novel argument or a new line of thought about the topic of controversy. Of course, this can happen in cases of agreement as well. One might get some new evidence supporting the conclusion one already believes. In addition to this kind of evidence, what one can obtain in cases of disagreement is evidence about what the peer thinks the shared evidence supports. Or, by learning that the peer believes as he or she does, and knowing that the peer tends to follow evidence, one might get evidence to the effect that the peer has some evidence supporting a view contrary to one’s own. Or that the peer has a view about the significance of shared evidence. This is evidence of evidence. And the question that intrigues me has to do with the nature and significance of this kind of evidence, that is, the question is what to make of evidence of evidence. As the title of the chapter suggests, my view is captured in the principle: EEE.  Evidence of evidence is evidence. In the remainder of this chapter, I will expand upon and discuss this principle.8 In Feldman (2006), I appealed to this principle as part of an argument for the idea that one should, in many cases of disagreement, revise one’s view in the direction of one’s disagreeing peer. I still think that something along those lines is true, although I do not want to assert any fully general principles. But before pursuing that issue, it is necessary to clarify EEE.

  8  My thinking about this topic has benefitted greatly from discussion with Brian Barnett, Jon Matheson, and Bill Rowley.

Evidence of Evidence is Evidence  291 There is a wholly trivial reading of the principle according to which it asserts that if something is evidence of (or about) evidence, then it is evidence. That, of course, is true—evidence of anything at all is evidence. EEE was intended to convey something more significant than that. Plainly, the point had something to do with evidential connections or evidential support. One might consider something like this: EEE1.  Evidence of evidence for P is evidence for P. This seems to say that evidence that there is an evidential connection—an evidential support relation—between propositions is evidence for the supported proposition. Perhaps a more perspicuous rendering of this idea is: EEE1*.  If E1 is evidence that E2 supports P, then E1 supports P. But this is surely false.9 Suppose I get some evidence supporting the fact that a certain kind of evidence supports a certain kind of conclusion. For example, I might get evidence, from philosophical argument or from testimony, that standard inductive arguments are in fact good arguments. That is, I get evidence that something like “All examined As are Bs” is good evidence for “All As are Bs.” The evidence supporting the proposition that there is this evidential support relation is plainly not evidence for the proposition “All As are Bs.” The idea behind the slogan was surely something not just about abstract evidential relations among propositions. It was about what happens when someone learns a certain fact about evidence. And the key fact in question is that one’s peer has evidence for a certain conclusion (namely, the denial of what one believes oneself). The idea can be formulated this way: EEE2.  Evidence that a person, S, has evidence supporting P, is evidence supporting P. I think that something close to this is true, but EEE2 is not quite right. It will be easier bring out the problem by formulating EEE2 slightly more perspicuously: EEE2*. If E supports the proposition that someone has evidence supporting P, then E supports P. This is not true. Someone can have conclusive evidence establishing both that another person has evidence supporting a proposition and that the supported proposition is false. In other words, the person has evidence that the other person has misleading evidence for P. If E is this person’s combined evidence, then E supports the conjunction of “Someone has evidence supporting P” and “~P.” Thus, E supports its first conjunct

  9  Brian Barnett got me to recognize these first two interpretations of (EEE), and to see that they were obviously false.

292  Richard Feldman (making the antecedent of EEE2* true), but E does not support P (making the consequent of EEE2* false). But consider any person who has evidence of this sort. Such a person has some evidence to the effect that another person has evidence for P, and some additional evidence against P. Call these two parts of this person’s evidence E1 and E2. It is the conjunctive evidence, E1 and E2 (=E in the example of the previous paragraph), that provides the objection to EEE2*. Notice, however, that the person who does have this conjunction as evidence does have E1 as evidence. And, in my view, this is evidence for P. This provides the basis for a better principle: EEE3. If S has evidence, E1, supporting the proposition that there is someone who has evidence that supports P, then S has some evidence, E2, that supports P. I think that EEE3 is true. There are true instances of it in which E2 is identical with E1. However, crucially, E2 could be a part of E1 or simply distinct from E1. Given EEE3, it is possible to draw a conclusion about the evidence obtained in cases of disagreement. ED. In virtually every case of a disagreement, one gets evidence that one’s peer has evidence for the conclusion the peer believes. That is, when you learn that your peer believes ~P, given what you know about yourself and your peers, you get some evidence that your peer has some evidence for believing as he or she does. That provides an instance in which the antecedent of EEE3 is true. You therefore have some evidence that ~P is true. And, given that the situation is a disagreement, you have some evidence against what you believe, P. Thus, in disagreement cases, given EEE3, a person typically gets evidence against the person’s antecedent belief.10 ED states what evidence one gets in cases of disagreement, but it leaves entirely open just what impact this evidence against one’s antecedent view has on one’s overall evidence. I think the impact can vary greatly, and this helps explain why general principles about how to respond to disagreement are hard to come by. A few examples will illustrate this point. Consider first a typical public issue in which it is well known that there is widespread disagreement. If a person encounters a peer who disagrees, it is not likely to have much impact on her beliefs. It is easy to see why. The evidence obtained from learning that this particular peer disagrees is, ordinarily, not terribly significant. She already knows that people have evidence on their side of the dispute and learns only that this particular person has some evidence of that sort. That is generally not a very significant piece of evidence.   10  The exceptions to this, if there are any, are cases in which you have no background evidence about yourself and your peer that provides reason to think that the peer has some reason to believe as he or she does. Such cases will be unusual, but perhaps they are possible.

Evidence of Evidence is Evidence  293 Contrast that case with David Christensen’s widely discussed Restaurant Case, in which one learns that a peer thinks that the shares of the restaurant bill differs from what one thought oneself.11 This is surprising information. It may well have a much bigger impact on what it is reasonable to believe. In fact, in some circumstances it may well lead to the view that it is equally likely that one’s peer is right as that one is right oneself. Suspension of judgment may then be called for. Finally, consider a case in which one learns that someone with a rather different background than one’s own has a different view about some social practice such as effective ways to bring up one’s children. Differences in such a domain may come as a surprise, and in conjunction with background evidence may provide one with reason to believe that one’s own view is in part socially induced and possibly less well supported than one had thought. Of course, details about the situation may well make a great deal of difference in such cases. This brief catalogue of cases illustrates an important point about evidence. The impact of learning a particular kind of fact, in this case that someone has evidence supporting the denial of what one believes, can vary greatly from case to case. That is simply a fact about how evidence works. If I am right, in virtually all cases of disagreement, one obtains some evidence against one’s prior view. But the impact of that evidence can vary greatly. As I’ll argue, or assert, below, it may have no impact. That is, the new evidence may not require any change whatsoever in one’s beliefs. This supports the idea discussed earlier, that there are no applicable general principles of the sort discussed earlier in connection with (Q1) and (Q2).

15.5  Connections Between the Questions The conclusion of the previous section is that in virtually every case of peer disagreement, where the peers are generally reasonable people who are responsive to evidence, each peer gets evidence against his or her prior view. In this section, I want to examine briefly what light this sheds on the questions about disagreement noted at the beginning of this chapter and on the connections between those questions. Recall the questions: Q1)  Can people involved in peer disagreement reasonably disagree? Q2)  How should one rationally respond to learning of a peer disagreement? Q3) What should a participant in a peer disagreement think about the rationality of the peer? There is a somewhat attractive line of thought suggesting partial answers to these questions. The starting point of this line of thought is that, in at least many typical cases of disagreement, there is an apparent symmetry, or near symmetry between the peers. That is, neither peer has good reason to believe that his or her own evidence is superior  

  Christensen (2007: 193).

11

294  Richard Feldman to the other person’s evidence. Further, after an extensive exchange of information, their bodies of evidence are similar in content. This suggests that, at least for these cases, (a) they can not justifiably believe different things (in response to (Q1)), (b) they should respond by suspending judgment (in response to (Q2)), and (c) they should think that the others are not reasonable if they maintain belief (in response to (Q3)). According to this line of thought, point (a) follows from the strong similarity of their evidence and the Uniqueness Thesis, mentioned earlier. Point (b) follows from (a) and the symmetry of the cases. And point (c) seems to follow from (a). This suggests that these partial answers to the questions are connected to one another. Only someone enamored with evidentialism could be taken in by the last step of this line of thought, the part about (c) and (Q3). What it is reasonable for one to think about the status of another person’s belief depends in large part upon what one reasonably thinks about rational belief. Given the possibility of reasonably believing non-evidentialist views about that, and the possibility of rationally denying the Uniqueness Thesis, there is little connection between (Q3) and the other questions. Suppose a person justifiably believes a view about rationality that is very liberal—sincerity of belief is enough for justification. Such a person would be fully justified in believing that her sincere peer is reasonable in maintaining belief, even if that is in fact incorrect. The evidence of evidence she has obtained does not have the impact one might have expected, given her reasonable but erroneous view about justification. Similar results obtain for anyone with other reasonably held views about rational belief that depart from evidentialism and uniqueness. If one reasonably thinks that it is possible for people to come to different conclusions even if their evidence is similar, then, finding out that a peer has come to a different conclusion on the basis of evidence like one’s own does not provide a reason to think that the peer is less than reasonable. It is thus difficult to get from an answer to (Q1) or (Q2) to an answer to (Q3), in part because what one learns in cases of disagreement is evidence of evidence. How this interacts with what one should think about the rationality of peers depends in large part on one what reasonably thinks about rationality generally. The issues surrounding the connections between (Q1) and (Q2) are complex. I will address the topic only briefly. Suppose that evidentialism and the Uniqueness Thesis are true, and consider again cases in which peers have essentially the same evidence and, according to these view should have the same attitude toward the disputed proposition. It could be that they should both suspend judgment, but it could also be that one of them has misinterpreted the evidence and one peer should have the other peer’s view. The point I want to make about this is simply the following: it is one thing to say that, given a total body of evidence, a person ought to have a certain attitude toward a proposition, and it is another thing to say that upon learning something that results in a person’s having that total body of evidence, the person should change to the attitude supported by that body of evidence. There’s a difference, or at least potentially a difference, between views about what beliefs are justified by a body of evidence and views about how beliefs should be changed in light of new evidence. An example will bring out the possible difference.

Evidence of Evidence is Evidence  295 Suppose an overly optimistic person unreasonably believes that he has the ability to succeed at some task. He has no reason at all to believe this. He then gets a tiny bit of evidence in support of his unjustified belief in his own competence. That belief is still overwhelmingly undermined by his evidence. It’s clear that the belief that he is competent is not justified in this circumstance. We can say that he should not have that belief. But it is another question whether the theory of rational belief change should imply that upon getting this modest amount of supporting evidence for the conclusion that he is competent, he should abandon that belief. One might think abandoning a belief when one gets evidence for it is not a rational response. And if that is correct, then the connection between questions about what is justified or reasonable (as in (Q1)) are not so closely connected to questions about how about reasonable change in belief (as in (Q2)). I do not wish to defend any view about the connection between these kinds of questions. My point here is just to bring out the potential gap between them and to show that this gap may matter in cases of disagreement. To see the latter point, consider again the role of beliefs about rationality in these cases. As in the examples described above, suppose one believes, with at least some reasonable basis, that sincerity of belief is sufficient for rationality or that there can be multiple reasonable responses to evidence. But, again, also assume that these beliefs are false. It may be that in fact a peer in a case of disagreement has an unreasonable belief. But, given these other beliefs, it may be that learning about the peer’s beliefs is not an occasion on which this unreasonable should change. Given the situation, this may not be the kind of evidence that should provoke the change to what is in fact the reasonable belief. In this regard, the peer may be in a situation somewhat like the overly confident person who gets some slight amount of evidence in support of an otherwise unjustified belief. The belief remains unjustified, but this new evidence does abandoning that belief. It thus remains open whether learning about the beliefs of the peer is the occasion on which the person who has an incorrect belief should change. That is, even if the theory of justified belief implies that one peer is not believing as he or she should, the theory of belief change may not imply that learning about this situation should provoke the change to the justified belief. That depends upon how evidence of evidence affects rational belief change.

15.6  Puzzles and Questions In this final section I will consider several objections to EEE3, the final formulation of the evidence of evidence is evidence principle: EEE3. If S has evidence, E1, supporting the proposition that there is someone who has evidence that supports P, then S has evidence, E2, that supports P. Some examples suggest that this principle is highly implausible. By responding to these examples, I hope to defend the view and to provide some additional clarification of the impact of evidence of evidence.

296  Richard Feldman

15.6.1  Objection 1: Redundant Evidence EEE3 implies, implausibly, that by finding more and more peers who believe a proposition, one gets more and more evidence in support of that proposition. Consider a person in a large, crowded, well-lit auditorium. Suppose the person goes around the room asking each person whether he or she thinks that the lights are on. EEE3 implies that each positive response provides additional evidence that the lights were on, and thus that the person gets more and more justified in this belief. But this is a mistake. The person already knew perfectly well that the lights were on, and the additional reports make no difference. Two responses suggest themselves. One might argue that the person does get increased justification with each additional confirmation, though the amount of these increases may become vanishingly small. That view seems wrong to me. Instead, I favor denying an “additivity of evidence” principle: 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. Alternative formulations of the consequent of this principle might be “S’s evidence supports P more strongly than it did before,” or “S should believe P more strongly or with more confidence.” I think AE is false. But the argument in Objection 1 depends on AE. In the example described, the antecedent is true each time the person gets an additional affirmation that the lights are on. But, eventually the consequent is false. At some point, the evidence tops out. This has the result that I must deny that the evidential support relation can be defined in terms of justification increase. But that seems right to me, at least if the connections are necessary in the way I think they are. If there is certainty (maximal justification), then additivity must be false. Notice that it would not make sense to deny that the evidence obtained from surveying each person in the room does (together with one’s background evidence about reliability in such circumstances) provide evidence for the conclusion that the lights are on. This is because, in cases in which the person does see the lights himself, this testimony would provide evidence for that conclusion. Given the necessity of evidence, it must be evidence even when it is redundant. Thus, redundant evidence is evidence, even if it makes no difference in levels of justification.

15.6.2  Objection 2: Evidence from Beliefs about Oneself EEE3 implies that a person’s realization that she herself believes something provides her with evidence that her belief is true. Suppose you consider your evidence, rationally conclude that it supports P, form the belief that P, think some more and notice

Evidence of Evidence is Evidence  297 that you believe P, apply EEE3 and conclude that you now have even more evidence, so believe more strongly, (and so on . . .) Even without the “and so on,” this seems wrong. A variant of the theme: If you keep changing your mind, it seems wrong to say that your evidence keeps changing. But EEE3 apparently implies that it does. In the most worrisome case, suppose that in fact your “objective” evidence is neutral. The fact of your belief or disbelief seems to tip the scales in whichever direction you go. This all seems absurd. This objection is not unlike the first one. To begin, my view does commit me to the view that a person’s knowledge of her belief does provide her with evidence for its truth, when taken in conjunction with common background knowledge of her reliability. This is a consequence of the objectivity and necessity of evidential relations, as noted in section 15.2. In conjunction with background evidence, the fact that the person believes as she does surely can provide another person with evidence for the truth of the content of her belief. But this is a fixed evidential fact. So, it is evidence for the person herself. Similarly, the fact that a person previously believed something can provide that person with evidence. The objection would only be troublesome if the fact that one has this additional evidence boosts the level of justification for the proposition. But, denying AE avoids this outcome. The fact that you believe as you do is (in conjunction with other information) a reason to believe the proposition you believe. It is evidence. And if it is evidence for others, then it is evidence for you. I see no flaw in EEE3 here.

15.6.3  Objection 3: Defeated Evidence Suppose you read an article in a newspaper that says “Yesterday, the newspaper erroneously reported that P. We regret the error.” EEE3 implies that you thereby get evidence for P, given only some innocuous assumptions. The innocuous assumptions are that in reading the newspaper, you get evidence that the newspaper reported that P yesterday and that this, together with your background evidence, supports P. I think that EEE3 gets this case exactly right. When you get evidence, you can at the same time get a defeater for that evidence. The combined evidence does not boost your level of justification for the conclusion. But the part of the evidence that supports that conclusion does continue to support that conclusion. Two considerations support this view: first, if you were asked whether you had any evidence at all in support of P, it is clear that the answer is “Yes”—you have evidence about yesterday’s newspaper report. To ignore this would be wrong. Of course, to fail to mention the defeater for this evidence could be seriously misleading. Second, if you learned that the defeater was mistaken, e.g., that today’s paper retracted the wrong things, you would not gain new evidence for P, but you would (other things equal) become justified in believing P. This is because the evidence you got when you learned of the retraction contained defeated evidence for P and that evidence is no longer defeated.

298  Richard Feldman A striking variant of this objection is instructive.12 Suppose you tell a friend something you know to be a lie. You know that your generally reliable friend believes you. By EEE3 you get evidence for the deceitful conclusion. As I see it, this is correct: you do have evidence for that conclusion. It’s defeated evidence that you have manufactured. But it must be evidence for the conclusion, since it provides your deceived friend with good reason to believe what you said. The point can summed up this way: defeated evidence is evidence.

15.6.4  Objection 4: Transitivity Issues Branden Fitelson argues convincingly that versions of the evidence of evidence is evidence principle along the lines of (EEE1) fail due to a kind of transitivity failure. Something could be evidence for P, but fail to be evidence for something P supports. He claims that a similar problem affects a principle very close to (EEE3).13 The example runs as follows. Suppose that Jim knows that a card, c, is going to be drawn from a standard deck. Jim knows: E1:  c is a black card Jim also knows that another person, John, has looked at the card and knows which one it is. So Jim has evidence supporting the claim that John has evidence for E2:  c is the ace of spades And this supports, indeed it entails, p:  c is an ace Given this setup, Jim has some evidence that John has evidence for E2 and p. Prior to his seeing John look at the card, Jim had no reason to think that John had any evidence for E2. That is, for Jim, the likelihood that John has evidence for E2 goes up when he sees Jim look at the card. Given EEE3, it follows that Jim has evidence for E2 and p. But, Fitelson writes, “Jim does not possess any evidence . . . which supports p.” (87) He notes that Jim’s evidence for p can’t be E1 (or his evidence for E1), since this does not support p. Nor can it be E2, which Jim reasonably disbelieves, since it is so unlikely that this is the card drawn. I think, however, that Jim does have some evidence for p. It is true, of course, that Jim has no better reason to believe p after he sees John draw the card than he had before he saw him draw the card. But this does not imply that he has no evidence for this conclusion, or that he didn’t get any such evidence in conjunction with seeing John draw the card. At some point in the story, either when John drew the card or prior to that, Jim learned that c was a card to be drawn from a standard deck. In learning this, he acquired some reason to believe that it was an ace. After all, even if he knew it was    

  I originally heard this objection in personal communication from Hud Hudson.   Fitelson (2012).

12 13

Evidence of Evidence is Evidence  299 a card, it could have been a card from a different kind of game, say a children’s game with different kinds of cards. Knowing that it was a black card from a standard deck narrowed down the possibilities some. When Jim gets the evidence that John has seen the card, he gets some evidence that John has evidence for E2, and this is redundant evidence for p. Furthermore, he gets similar evidence for the card being each of the other live possibilities. Hence, Jim’s rational beliefs do not change. But, there is in this story, some evidence for p and no counterexample to the final evidence of evidence is evidence principle.

15.7 Conclusion I conclude that, properly understood, evidence of evidence is evidence. However, the proper interpretation of this principle is the relatively modest principle captured in EEE3. Given that in typical situations a person involved in a disagreement has some reason to think the disagreeing peer has some evidence for the disputed proposition, it follows that the person gets some evidence against his or her original view. But this leaves open entirely what impact that evidence might have. Since it is possible for it to be redundant or defeated evidence, it may not imply that the rational status of the person’s beliefs changes at all. Depending upon the other factors in the case, it may have a significant impact, or it may only modestly affect how well the person’s evidence supports the conclusion. This makes it difficult to provide general answers to questions about the rational status of beliefs in cases of peer disagreement.

References Bergmann, Michael. (2009). “Rational Disagreement after Full Disclosure,” Episteme: A Journal of Social Epistemology 6(3): 336–53. Christensen, David. (2007). “Epistemology of Disagreement: The Good News,” Philosophical Review 116: 187–217. Conee, Earl and Feldman, Richard. (2004). Evidentialism. Oxford: Oxford University Press. Elga, A. (2007). “Reflection and Disagreement,” Noûs 41(3): 478–502. Feldman, Richard. (2003). “Plantinga on Exclusivism,” Faith and Philosophy 20, 85–90. Feldman, Richard. (2006). “Epistemological Puzzles about Disagreement,” in Stephen Hetherington (Ed.), Epistemology Futures. Oxford: Oxford University Press, 216–36. Feldman, Richard. (2007). “Reasonable Religious Disagreements,” in Louise Antony (Ed.), Philosophers without God. Oxford: Oxford University Press, 194–214. Feldman, Richard. (2009). “Evidentialism, Higher-Order Evidence, and Disagreement,” Episteme 6: 294–312. Fitelson, Branden. (2012). “Evidence of Evidence is not (Necessarily) Evidence,” Analysis 72: 85–8. Kelly, Thomas. (2010). “Peer Disagreement and Higher Order Evidence,” in Richard Feldman and Thomas A. Warfield (Eds.), Disagreement. Oxford, Oxford University Press, 111–74.

300  Richard Feldman Lackey, Jennifer. (2010). “What Should We Do When We Disagree?” in Tamar Szabo Gendler and John Hawthorne (Eds.), Oxford Studies in Epistemology. Oxford: Oxford University Press. Matheson, Jonathan. (2009). “Conciliatory Views of Disagreement and Higher-Order Evidence,” Episteme: A Journal of Social Philosophy 6(3): 269–79. Matheson, Jonathan. (2011). “The Case for Rational Uniqueness,” Logos & Episteme:  An International Journal of Epistemology 2(3): 359–73.

16 Believers as Thermometers Thomas Kelly

16.1  Disagreement and the Thermometer Model of Belief Revision Case 1: The Steadfast Juror You and I are conscientious and attentive members of a jury charged with determining whether the accused is guilty. Neither of us has any reason to think that he or she enjoys some advantage over the other when it comes to answering questions of this kind. Indeed, what evidence we have suggests that we are likely to be more or less equally reliable when it comes to making such judgments. Because of this, if we had been asked in advance of the trial which one of us is more likely to be wrong in the event of a disagreement, we would have agreed that we were equally likely to be the one who is wrong. 1 Nevertheless, you and I arrive at different views in response to the evidence presented during the trial: while I find myself quite confident that the accused is guilty, you find yourself equally confident that he is innocent. Suppose that, upon learning that I am convinced that the accused is guilty, you stick to your guns and steadfastly maintain your original view that he is innocent. It’s not that you take yourself to have discovered some independent reason to discount my judgment over the course of the trial—as would be the case if, say, you had noticed that I was distracted at some crucial moment, or if I had confided in you that I was suffering from a headache that was interfering with my ability to think clearly. Rather, you retain your belief that the accused is innocent because you take this to be a perfectly reasonable thing to think, given the evidence that is currently available to us.2   1  Cf. Elga’s (2007) account of what it is to treat someone as an epistemic peer in his sense.   2  It is an interesting question, although not one that I will pursue here, whether taking your own view to be a perfectly reasonable response to the evidence commits you to thinking that it is the perfectly reasonable response, and that anyone who responds differently responds in a way that is less than perfectly reasonable. Granted that your view that the accused is innocent commits you to thinking that my view that he is guilty is false, are you similarly committed to thinking that my view is unreasonable, given that we have the same evidence to go on? For a recent exchange on this question, see White (2013) and Kelly (2013a).

302 Thomas Kelly Question: could this steadfast response be a reasonable one on your part? Many philosophers are skeptical that it could. Call a view Conciliationist just in case it entails that, in cases like that of The Steadfast Juror, it would be unreasonable to steadfastly maintain your original view in the face of my conflicting opinion.3 A guiding thought behind Conciliationism is that, when the case is set up right, a fully rational person who is aware of the relevant symmetries and lacks independent reason to discount the opinion of the other person will treat her own original opinion and the original opinion of the other person evenhandedly—and that such evenhanded treatment is incompatible with steadfastly maintaining her own original view. Thus, given that initially I am confident that the accused is guilty while you are equally confident that he is not, you should suspend judgment about the issue upon learning my opinion. (I should do the same.) As Richard Feldman puts it: [C]‌onsider those cases in which the reasonable thing to think is that another person, every bit as sensible, serious, and careful as oneself, has reviewed the same information as oneself and has come to a contrary conclusion to one’s own . . . An honest description of the situation acknowledges its symmetry . . . .In those cases, I think, the skeptical conclusion is the reasonable one: it is not the case that both points of view are reasonable, and it is not the case that one’s own point of view is somehow privileged. Rather, suspension of judgment is called for. (2006: 235)4

Although tempting, I believe that Conciliationism is incorrect. The correct thing to say about The Steadfast Juror, I think, is this: whether it’s reasonable for you to continue to believe that the suspect is innocent even after learning that I think otherwise is not something that can be determined, given only the facts about the fiction provided. What are some circumstances in which maintaining your view might very well be reasonable? Suppose that the original evidence with which we are presented in court strongly supports the view that the suspect is innocent. Your original belief is a rational response to what was then our total evidence; mine was not. (Against a general background of competence, I commit a performance error.) After you learn that I think that the accused is guilty, your total evidence has changed: it is now on the whole less supportive of the view that he is innocent than it was previously. It’s thus reasonable for you to reduce your confidence to at least some degree. Still, the total evidence available to you then might very well make it more likely that the suspect is innocent than that

  3  For views that are at least in the near neighborhood of the one described here, see Feldman (2003, 2006, 2007), Elga (2007), Christensen (2007, 2011), Bogardus (2009), Matheson (2009), Kornblith (2010), and Cohen (2013), among others. There are interesting and important differences among these philosophers that I will not discuss here. In what follows, I will use the term “Conciliationism” to pick out an epistemological view of a familiar general type without attempting to follow any particular epistemologist in detail.   4  Although Feldman’s early writings on the topic of disagreement provide paradigm statements of the position that I call “Conciliationism” they do not accurately reflect his latest views. Indeed, on the basis of Feldman (2009) and recent conversations, I believe that there is relatively little (and possibly no) difference between his current views and the one that I defend under the heading “The Total Evidence View.”

Believers as Thermometers  303 he is guilty, to the point that it’s reasonable for you to believe that he is guilty. In any case, there is certainly no guarantee that the uniquely reasonable response on your part is to suspend judgment, or retreat to a state of agnosticism. Elsewhere I’ve criticized Conciliationism at some length and offered what I take to be a superior alternative, the Total Evidence View (Kelly 2005, 2010, 2013b). However, Conciliationism might seem simply inescapable, given a highly intuitive picture of how we should take the views of others into account. I’ll call this picture the thermometer model. According to the thermometer model, the views at which people independently arrive are like the readings of more or less accurate thermometers. When information about the reliability of a particular thermometer is available, that determines the extent to which we should credit its readings. Similarly, when information about the reliability of another person is available, then that determines the extent to which we should defer to that person, or credit his opinions. When spelled out more fully, the thermometer model seems to deliver substantive epistemological conclusions. In particular, applied to questions about disagreement, it seems to support Conciliationism over alternative views that are more hospitable to steadfastness. Consider, for example, the following kind of case. Case 2: Conflicting Thermometers I form my beliefs about the ambient temperature in some room by consulting my thermometer. (To avoid complications, we can suppose that I have no other access to the temperature of the room. Perhaps I’m in an adjacent room, looking at my thermometer through a window, and I know that the temperature of the room that I’m in is not reliably correlated with the temperature of the room in which the thermometer is located.) I have good reason to trust my thermometer—I know that it’s worked well in the past—so the beliefs that I form in this way are perfectly reasonable. But suppose that, on a given occasion, we discover that the reading returned by my thermometer is inconsistent with the reading returned by your thermometer, which is also clearly visible in the adjacent room. Of course, if I had evidence that my thermometer is generally more reliable than yours, or that yours is more likely to be malfunctioning on this particular occasion, then it would be reasonable for me to give significantly more credence to what my thermometer says than to what yours says. But in fact, I have no reason to think anything like this: prior to this episode, what evidence I had suggested that our thermometers were equally reliable.5 In these circumstances, I should be agnostic about whose thermometer is correct, and divide my credence evenly between their readings, at least until further relevant evidence becomes available. That is, it would be unreasonable for me to steadfastly maintain my original view about the temperature after learning what your thermometer indicates. Certainly, it wouldn’t be defensible to favor the reading of my   5  Of course, I now have the following reason to think that your thermometer is wrong on this particular occasion: the fact that my thermometer says something else. But by the same token, the fact that your thermometer reads as it does gives me a reason to think that my thermometer is wrong on this occasion. So the symmetry is still intact.

304 Thomas Kelly thermometer simply because that’s what my thermometer says, or because that’s what I justifiably believed before I learned about your thermometer. Moreover, the same seems to be true even if your thermometer really is malfunctioning on this particular occasion, and mine is functioning perfectly. So long as I have no independent evidence that that’s what’s taking place (e.g., a third thermometer that agrees with mine), the mere fact that my thermometer is the one that’s functioning properly on this occasion doesn’t justify my giving more credence to what it reports than to what yours reports. I think that this is the right thing to say in the case of thermometers. But once we accept the Equal Weight View of Thermometers, isn’t the Equal Weight View of Peer Disagreement inescapable? I expect that you and I will agree about whether the accused is guilty, but it turns out that we disagree. If I had evidence that I’m more reliable about this type of question, or some independent reason to think that I’m the one who got it right on this particular occasion, then I would have reason to favor my initial view over yours. But ex hypothesi, I don’t have any such thing. Indeed, worse still: what evidence I have suggests that you’re just as reliable as I am. Surely, in these circumstances, it would be unreasonable to favor my opinion over yours, given the relevant symmetries. I should divide my credence equally between your original opinion and my original opinion. But this is just what the Conciliationist thinks is true. The question then, is whether accepting what seems to be the correct answer in the case of the conflicting thermometers compels us to give the Conciliationist answer in the jury example, and (more generally) in paradigmatic cases of “peer disagreement.” A Conciliationist who appeals to the thermometer model will claim that the answer to this question is “Yes.” Indeed, someone might press the case by changing the details of the jury example to increase its resemblance to the thermometer example, in ways that (one might think) could not possibly introduce any normative difference. Thus, perhaps in the jury case the question that’s before you and I is not whether the accused is guilty or innocent, but the temperature of the room in which the trial is taking place. One way in which we could be peers with respect to questions of this latter kind is the following: I always arrive at my views about the ambient temperature by deferring to my thermometer, while you always arrive at your views by deferring to your thermometer, and our thermometers are (in general) equally reliable. Now suppose that we simply get rid of the thermometers, and each of us relies on his or her own “sense” of how hot the room is. No doubt, if our thermometers are generally reliable, then getting rid of them will hurt our reliability in absolute terms, but that is perfectly consistent with our being equally good at making such unaided judgments. And so long as our comparative reliability with respect to the relevant class of judgments is unchanged, then it seems like throwing away the thermometers should make no difference at all: if we were rationally required to suspend judgment in response to discovering conflict among our thermometer-aided judgments, then it seems like the same is true when we discover conflict among our thermometer-independent judgments about the temperature. But then why shouldn’t our judgments about whether the accused is innocent or guilty be in the same boat? In this way, someone might claim that reflection on the

Believers as Thermometers  305 thermometer example reveals that the Conciliationist’s approach to disagreement is compelling. So shouldn’t we be Conciliationists after all? I don’t think so. Why not? I want to approach this question indirectly. Here is the strategy that I will pursue in what follows. I will describe a particular version of Conciliationism, which I will call “Strong Conciliationism,” and distinguish it from some weaker versions of the view. Strong Conciliationism is supported by the thermometer model, but there are compelling reasons to think that it’s false. Perhaps for that reason, Strong Conciliationism tends to be unpopular among actual, flesh-and-blood Conciliationists and their fellow travelers. On the other hand, weaker forms of Conciliationism escape some of the more debilitating problems that afflict their stronger cousin, but they are not supported by the thermometer model. In fact, proponents of the weaker versions of Conciliationism have just as much reason to reject the thermometer model as anti-Conciliationists do. So the burden of the rest of the chapter is to show the following: there is no version of Conciliationism that is both plausible and supported by the thermometer model. Where that leaves us is something that I will address briefly at the very end of the chapter.

16.2  Conciliationism: Some Varieties For heuristic reasons, the epistemology of disagreement literature often focuses on relatively idealized, two person cases of disagreement like the jury case described above. Indeed, in attempting to zero in on the type of view that I am concerned to oppose, I characterized as conciliationist any view that entails that it would be unreasonable to stick to one’s guns in cases like The Steadfast Juror. On reflection, however, it seems clear that any plausible and well-motivated version of Conciliationism will have implications not only for two person cases of disagreement, but also for many other cases in which questions arise about how to adjust one’s views in response to learning the opinions of others. These other cases will include cases of disagreement involving more than two individuals, as well as cases that are not cases of disagreement at all. Suppose, for example, that we amend the jury case so that it is a case of agreement rather than disagreement (while leaving the other details the same). In this variant, you and I both independently arrive at the view that the accused is innocent; we then learn what the other thinks. Given that the Conciliationist holds that my arriving at a contrary view would have rendered it unreasonable for you to steadfastly maintain your original opinion when we disagree, the Conciliationist should say that my opinion counts as evidence in favor of your view when we agree, so that when you learn that we arrived at the same view you should increase your confidence in that view (assuming that you were not already maximally confident that it is true). Of course, there is nothing logically inconsistent about an epistemological view according to which disagreement, but not agreement (in the sense of independent convergence) is potentially of epistemic significance. But no clear-headed Conciliationist will find such a view attractive.

306 Thomas Kelly Similarly, the clear-headed Conciliationist will see the kinds of considerations that lead her to deem the steadfast juror unreasonable to have implications for cases of disagreement involving more than two individuals. For example, consider another variant on the original jury case; in this variant, all twelve members of the jury independently arrive at a view about whether the accused is guilty. (Let’s suppose, unrealistically, that each of the jurors stand in the same “peerhood” relations to the others that you and I stand in to one another in the original version of the case.) If it turns out that eleven of the jurors arrive at the conclusion that the accused is innocent, while the twelfth arrives at the opposite conclusion, then the Conciliationist should insist that the twelfth juror should be quite confident that he was the one who made the mistake, and that the accused is actually innocent.6 And this is so because the sorts of considerations (relating to evenhandedness, etc.) that motivated the thought that suspending judgment is the appropriate response in the original jury case can similarly be used to motivate the thought that an even more concessive response is in order when one finds oneself greatly outnumbered among individuals each of whom is no less qualified to render a judgment than oneself. Thus, the Conciliatonist approach to cases like The Steadfast Juror, if theoretically well motivated, suggests natural extensions to other cases in which questions arise about how to take the views others into account. Suppose that n peers independently arrive at an opinion about whether p. The peers then learn what the others think. Consider the state of opinion that would result from giving equal weight to the opinions of the n peers. Let’s call this the conciliated opinion. Any Conciliationist then, should think that the following normative claim is true: Upon learning the distribution of peer opinion, it would be unreasonable for a peer to maintain some opinion that diverges from the conciliated opinion. Notice that here and in what follows, I’m simply using the term “conciliated opinion” as a tag for the output of a certain aggregation function: namely, whatever aggregation function the Conciliationist would have us employ in order to capture the intuitive notion of giving “equal weight” to the original opinions of the peers. In fact, cashing out this intuitive notion proves surprisingly difficult.7 But in any case, I think that it is at least clear what the Conciliationist will say about two very special cases: (1) If the n peers are perfectly divided about whether p, then agnosticism/suspension of judgment about whether p is the conciliated opinion. For example, if 50% of the peers believe p, and 50% of the peers are equally confident that not-p, then the conciliated opinion is agnosticism about p. (I take it that it would be completely bizarre, and contrary to the spirit of the view, for a Conciliationist   6  To be clear, it is not only the Conciliationist who can say this; for example, it is open to a proponent of the Total Evidence View to embrace this verdict as well.   7  For some of the problems afflicting some of the more obvious proposals, see especially Jehle and Fitelson (2009).

Believers as Thermometers  307 to say that some other attitude towards p might be the conciliated opinion in those circumstances.) It also seems safe to assume the following: (2) If the n peers are unanimous about whether p, then the opinion on which they have independently converged is the conciliated opinion. For example, if all of the peers independently arrive at the view that p is false, then the view that p is false is the conciliated opinion. (Again, I take it that it would be bizarre for the Conciliationist to say anything else about this very special case.) So we have at least this much of a specification of the aggregation function. What Conciliationists should say beyond that is something that I will leave up to them. Consider again the claim that: Upon learning the distribution of peer opinion, it would be unreasonable for a peer to maintain some opinion that diverges from the conciliated opinion. This claim says, in effect, that not diverging from the conciliated opinion is a necessary condition for one’s later opinion to be fully rational; a perfectly rational believer would not hold some opinion that is out of step with the conciliated opinion. Let’s call this claim Weak Conciliationism. We should distinguish this view from a significantly stronger one: Strong Conciliationism: Upon learning the distribution of peer opinion, it would be unreasonable for a peer to maintain some opinion that diverges from the conciliated opinion. Moreover, if she holds the conciliated opinion in response to learning about the distribution of peer opinion, then her opinion about p is perfectly reasonable. That is, a peer’s holding the conciliated opinion is both sufficient as well as necessary for her opinion to be fully reasonable or justified. Clearly, Strong Conciliationism is stronger than Weak Conciliationism, inasmuch as it entails, but is not entailed by, Weak Conciliationism. Because of this, it’s natural to suspect that Strong Conciliationism might be vulnerable to objections that don’t arise for Weak Conciliationism. In fact, I think that there are compelling objections to Strong Conciliationism that don’t arise for Weak Conciliationism. Let me rehearse two such objections here.8 Consider first: Case 3: Implausibly Easy Bootstrapping A possible world is populated by two peers, each of whom is keenly interested in the question of whether God exists. Each is generally reliable (although not infallible) in responding to evidence, and knows this about himself and the other person. Moreover, both peers are practicing Conciliationists. As it happens, the evidence available in their world that bears on the question of whether God exists strongly suggests that  

  I first presented these cases in Kelly (2010: 125–31).

8

308 Thomas Kelly God does not exist. However, each of the peers badly misjudges this evidence and independently arrives at the view that God exists. At time t1 then, each of the peers irrationally believes that God exists. At time t2, the two peers meet and compare notes. Because they accept Conciliationism, they continue to believe that God exists. Given that this is a case in which all of the peers independently converge on the view that God exists, that view is the conciliated opinion (by (1) above). If holding the conciliated opinion is a sufficient condition9 for the reasonableness of one’s belief, then the peers’ belief in God is perfectly reasonable after they compare notes, despite the poor job that each of them did in responding to their original evidence. More generally, Strong Conciliationism has the consequence that views for which there is originally no good evidence or reason to think true can bootstrap their way into being rationally held whenever irrationally overconfident peers encounter one another and confirm each other’s previously baseless opinions. But that seems extremely implausible. I think that the fact that Strong Conciliationism has this consequence is a good reason to reject it. But one might not be convinced. After all, it’s plausible that there are at least some cases in which independent convergence on a given view makes it reasonable to hold that view, even if the reasoning that led to that view is bad reasoning.10 (Suppose that a roomful of competent mathematicians perform some complicated calculation and each of them independently arrives at the same answer. Plausibly, at the moment they learn that they have all arrived at that answer, it is reasonable for them to believe that that’s the correct answer—even if in fact it’s the wrong answer, and (miraculously) each of the mathematicians made exactly the same mistake, or some series of mistakes that led to the same answer that everyone else arrived at.) For this reason, it’s worth pushing the case against Strong Conciliationism a bit further. Recall that the Strong Conciliationist maintains that holding the conciliated opinion is sufficient for believing reasonably, and that the conciliated opinion is the opinion that results from giving equal weight to each of the original opinions of the n peers. So when there are 100 peers, the conciliated opinion is the result of giving equal weight to each of 100 opinions, when there are 36 peers, the conciliated opinion is the result of giving equal weight to the 36 opinions, and so on. Consider now the special case in which n=1. That is, consider Case 4: Case 4: Even Easier, and More Implausible Bootstrapping: Single Person Cases A  possible world is inhabited by a single individual. The individual is generally reliable (although not infallible) in responding to evidence, and knows this about himself. As it happens, the evidence available to him that bears on the question of whether God exists strongly suggests that God does not exist. However, he badly   9  At least, provided that one holds the conciliated opinion because it’s the conciliated opinion, i.e., the fact that one holds the conciliated opinion is a reflection of one’s awareness of the distribution of opinion. In what follows, I will assume that we are talking about cases in which this condition is satisfied.   10  For further discussion of this point, see Kelly (2011).

Believers as Thermometers  309 misjudges the evidence and arrives at the view that God exists. At time t1 then, he irrationally believes that God exists. At time t2, he reflects on the fact that belief in God is the conciliated opinion. If Strong Conciliationism is true, then the individual’s belief that God exists is perfectly reasonable at time t1. But that seems clearly wrong. I first raised these kinds of bootstrapping concerns in a paper entitled “Peer Disagreement and Higher Order Evidence.” In that paper, I  failed to distinguish between what I  am now calling “Strong” and “Weak” Conciliationism, and wrote indiscriminately of “The Equal Weight View.” Since then, I’ve heard responses to these bootstrapping objections from a large number of philosophers, a group with varying degrees of sympathy for Conciliationism. Some people found these objections compelling, but others did not. Here I want to say something about the latter group. Of course, one possible response to these kinds of objections is to simply bite the bullet and insist that the relevant kind of bootstrapping does produce fully reasonable belief. But generally speaking, that has not been the response of those who deny that the bootstrapping considerations are compelling objections against Conciliationism. The more common response—both in print and in conversation—has been that such bootstrapping objections are decisive objections against some possible view, but that that view is not a charitable reconstruction of what actual Conciliationists have had in mind. After all, the bootstrapping objections are only effective against views on which adopting the conciliated opinion is sufficient for reasonable believing. It’s implausible to think that someone can end up with a perfectly reasonable belief simply by responding to the distribution of peer opinion in the way that the Conciliationist recommends, even if he has responded to all his other evidence irrationally. But why should the Conciliationist be saddled with that view? That is, why should the Conciliationist be understood as embracing Strong Conciliationism, as opposed to Weak Conciliationism? An example of a Conciliationist who wants nothing to do with Strong Conciliationism is David Christensen (2011). As he puts it: Conciliationism tells us what the proper response is to one particular kind of evidence. Thus the Equal Weight Conciliationist is committed to holding, in Kelly’s cases, that the agents have taken correct account of a particular bit of evidence—the evidence provided by their peer’s [opinion]. But having taken correct account of one bit of evidence cannot be equivalent to having beliefs that are . . . rational, all things considered. If one starts out by botching things epistemically, and then takes correct account of one bit of evidence, it’s unlikely that one will end up with fully rational beliefs . . . If Conciliationism is understood in the right way, then, it is not committed to deeming the post-compromise beliefs . . . automatically rational. (2011: 4)

Notice that, because Strong Conciliationism entails that holding the conciliated opinion is sufficient for reasonable believing, but Christensen denies this, the version of Conciliationism that he favors is actually incompatible with Strong Conciliationism.

310 Thomas Kelly Because I believe that there are decisive objections to Strong Conciliationism, I think that this is the right thing for the Conciliationist to say. However, what I now want to argue is the following: The thermometer model does not support views like Christensen’s. On the contrary, it tells against them. This is because what the thermometer model supports is Strong Conciliationism, as opposed to any weaker version of the view. In order to see this, consider the following analogue to Case 3 above: Case 5: Malfunctioning but Convergent Thermometers I arrive at my views about the temperature in the adjacent room by consulting my thermometer, which is clearly visible through a window. At time t1, the thermometer indicates that the temperature of the room is 75 degrees, and I believe accordingly. At time t2—just to be sure—I consult your thermometer. It also indicates that the temperature of the room is 75. Because this is what both thermometers indicate, I continue to hold that belief. In fact, however, both of the thermometers happen to be malfunctioning badly on this occasion, albeit in exactly the same way: the actual temperature of the room is 45 degrees. What’s the correct thing to say about Case 5? Consider first the question of how it’s reasonable for me to respond to the discovery that your thermometer indicates that the temperature is 75 degrees, given that I already believe this. Here it seems clear that I’m rationally required to continue believing that the temperature is 75 degrees. (Indeed, I should become even more confident that this is true, assuming that I was not already completely certain.) Given that I already believe that the temperature is 75 degrees, it would clearly be unreasonable to respond to this additional piece of confirming evidence by taking up some other view about the temperature. But here’s the key point: it’s not simply that I’m rationally required to hold the belief that it’s 75 degrees at the later time, in a way that leaves open whether that belief is all things considered reasonable or not. Rather, given that my only access to the temperature of the other room is via the two thermometers, and both of the thermometers are indicating that that temperature is 75 degrees, my belief to that effect is fully reasonable and justified. That is, when the malfunctioning thermometers converge in their readings, my believing as the two thermometers indicate is a sufficient condition for the reasonableness of the relevant belief. In short, if we apply the thermometer model to Case 3 above, the view that it favors is Strong Conciliationism. Similarly, compare Case 4 with Case 6: The Malfunctioning Thermometer (Single Thermometer Case) My usually reliable thermometer indicates that temperature in the adjacent room is 75 degrees, so I believe accordingly. Unbeknownst to me, my thermometer is malfunctioning badly on this particular occasion: the actual temperature of the room is 45 degrees. I have no access to your thermometer. In these circumstances, my belief that the temperature is 75 degrees is fully reasonable or justified, notwithstanding its falsity. So if we take the thermometer analogy

Believers as Thermometers  311 seriously, we should conclude that the individual in Case 4 is fully justified in believing that God exists, despite the fact that evidence available in that world overwhelmingly supports the opposite conclusion. But that’s absurd.

16.3  Conclusion: Beyond the Thermometer Model? Conciliationist views about disagreement are often motivated by analogies involving inanimate measuring devices. We are invited to think of our opinions as like the readings of thermometers, and to respond to conflicts among the former as we would to conflicts among the latter. However, when the thermometer model is applied consistently rather than opportunistically, unpalatable consequences result, consequences that few if any Conciliationists would embrace. If what has been said so far is substantially correct, then Conciliationists are faced with a choice. First, a Conciliationist might embrace the thermometer model and bite the bullet with respect to the kind of bootstrapping cases discussed here. Alternatively, the Conciliationist might embrace what seem to be the intuitively correct verdicts about the bootstrapping cases while eschewing the thermometer model. If Conciliationists pursue the second option—as I believe they should—then a common project for Conciliationists and anti-Conciliationists emerges:  the project of exploring the limitations of the thermometer model, or where exactly the model breaks down. Consider again The Steadfast Juror and Conflicting Thermometer cases. Above, I suggested that the following is a pressing question for any anti-Conciliationist: why doesn’t giving what seems to be the clearly correct answer about the conflicting thermometer case—namely, that when our thermometers conflict, we are rationally required to suspend judgment about the temperature—commit one to the view that the steadfast juror is also rationally required to suspend judgment about the innocence of the accused? Here is my own answer to this question: in the steadfast juror case, you and I end up with more evidence than we do in the conflicting thermometer case. In the conflicting thermometer case, we end up with two pieces evidence about the temperature in the adjacent room, viz. the fact that my generally reliable thermometer indicates that the temperature is such-and-such, and the fact that your generally reliable thermometer indicates that it is something else. Plausibly, the uniquely reasonable response to these two pieces of evidence is to suspend judgment about the actual temperature. On the other hand, in the steadfast juror case, the evidence we have to go on includes, but is not limited to, the analogous facts about our opinions, viz. the fact that you judged that the accused is innocent, and the fact that I judged that he is guilty. But in addition, our total evidence also includes all of the evidence that was presented to us in court. And there is simply no guarantee that the uniquely reasonable thing to do is to suspend judgment given all of the available evidence. (As suggested above, this can depend on such things as whether the evidence originally presented to us in

312 Thomas Kelly court really does support the proposition that the accused is innocent, and if so, how strongly it supports that proposition.) If we left things here, one might conclude that the thermometer model survives intact: the epistemic significance of our opinions is just like the epistemic significance of thermometer readings, it’s just that in some cases but not in others, one has additional evidence to go on. However, there is an important respect in which our beliefs differ from thermometer readings, a respect that seems particularly relevant to the epistemology of disagreement. On the one hand, when I arrive at a belief about the temperature by relying on a thermometer that is in fact malfunctioning, that belief is, notwithstanding its falsity, fully reasonable or justified, so long as I have no reason to suspect that the thermometer is malfunctioning. On the other hand, when the malfunctioning is in my thinking itself—as when I take up the belief that the accused is guilty despite the fact that the evidence presented to us suggests that he is innocent— then my belief is not fully reasonable or justified, even if I have no independent reason to think that anything is amiss. Indeed, my belief is rationally defective even if I am in general extremely reliable when it comes to the relevant kind of reasoning, know this about myself, and treat it as a reason to trust the judgment at which I have arrived in the present instance.11 Consider then your position, upon discovering that we disagree in each of the two cases. In the thermometer case, you are confronted with someone whose conflicting opinion about the temperature is itself perfectly reasonable. In the jury case, however, you are confronted with someone whose conflicting opinion about the guilt or innocence of the accused is unreasonable. But it is independently plausible that, whatever skeptical pressure is created by disagreement, that pressure is weaker when the opinions that contradict one’s own are unreasonable opinions (even if those who hold them are generally reasonable people). A theorist who wants to assimilate the jury case to the thermometer case in effect argues as follows: “Look, we all agree about what to say about cases involving actual thermometers. But now, just internalize the mechanism: move the thermometer inside the head, so to speak.” However, internalizing a malfunctioning mechanism that leads to a particular belief is not in general an innocent step, inasmuch as doing so will often affect the rationality of that belief (cf. Kelly 2008: 629–30). We should be wary of such maneuvers.12

  11  Imagine, for example, that I commit some grotesque violation of the Principal Principle: despite knowing that I possess only a single ticket in a large and fair lottery, I am far more optimistic about my chances of winning than the odds warrant. Here, my level of confidence is unreasonable even if I am generally extremely reliable about such things (suppose that this is a one time slip), know that I am, and remind myself of this in the context of wondering whether my high level of confidence is actually justified.   12  This chapter was drawn from material presented to audiences at UCLA, the University of Massachusetts (Amherst), Vassar College, the University of North Florida, the “Arctic Epistemology Conference” in

Believers as Thermometers  313

References Bogardus, Tomas. (2009). “A Vindication of the Equal-Weight View,” Episteme 6(3): 324–35. Christensen, David. (2007). “Epistemology of Disagreement: The Good News,” The Philosophical Review 116(2): 187–217. Christensen, David. (2011). “Disagreement, Question-Begging and Epistemic Self-Criticism,” Philosophers Imprint 11(6): 1–22. Cohen, Stewart. (2013). “A Defense of the (Almost) Equal Weight View,” in Jennifer Lackey and David Christensen (Eds.), The Epistemology of Disagreement: New Essays. Oxford: Oxford University Press, 98–120. Elga, Adam. (2007). “Reflection and Disagreement,” Noûs 41(3): 478–502. Feldman, Richard. (2003). Epistemology. Upper Saddle River, NJ: Prentice Hall. Feldman, Richard. (2006). “Epistemological Puzzles About Disagreement,” in Stephen Hetherington (Ed.), Epistemology Futures. Oxford: Oxford University Press, 216–36. Feldman, Richard. (2007). “Reasonable Religious Disagreements,” in Louise Antony (Ed.), Philosophers without Gods:  Meditations on Atheism and the Secular Life. Oxford:  Oxford University Press, 194–214. Feldman, Richard. (2009). “Evidentialism, Higher-Order Evidence, and Disagreement,” Episteme 6(3): 294–312. Jehle, David and Fitelson, Branden. (2009). “What is the ‘Equal Weight View’?” Episteme 6(3): 280–93. Kelly, Thomas. (2005). “The Epistemic Significance of Disagreement,” in John Hawthorne and Tamar Gendler (Eds.), Oxford Studies in Epistemology, Vol. 1. Oxford: Oxford University Press, 167–96. Kelly, Thomas. (2008). “Disagreement, Dogmatism, and Belief Polarization,” Journal of Philosophy 105(10): 611–33. Kelly, Thomas. (2010). “Peer Disagreement and Higher Order Evidence,” in Richard Feldman and Ted Warfield (Eds.), Disagreement. Oxford: Oxford University Press, 111–74. Kelly, Thomas. (2011). “Consensus Gentium: Reflections on the ‘Common Consent’ Argument for the Existence of God,” in Kelly Clark and Raymond Van Arragon (Eds.), Evidence and Religious Belief. Oxford: Oxford University Press, 135–56. Kelly, Thomas. (2013a). “Evidence Can Be Permissive,” in Matthias Steup, John Turri, and Ernest Sosa (Eds.), Contemporary Debates in Epistemology, 2nd ed. Malden, MA: Wiley-Blackwell, 298–311. Kelly, Thomas. (2013b). “Disagreement and the Burdens of Judgment,” in David Christensen and Jennifer Lackey (Eds.), The Epistemology of Disagreement: New Essays. Oxford: Oxford University Press, 31–53. Kornblith, Hilary. (2010). “Belief in the Face of Controversy,” in Richard Feldman and Ted Warfield (Eds.), Disagreement. Oxford: Oxford University Press, 29–52. Matheson, Jonathan. (2009). “Conciliatory Views of Disagreement and Higher-Order Evidence,” Episteme: A Journal of Social Epistemology 6(3): 269–79.

Lofoten, Norway, and a meeting of my graduate seminar at Princeton University. I am grateful to those audiences for their feedback.

314 Thomas Kelly White, Roger (2005). “Epistemic Permissiveness,” in J. Hawthorne (Ed.), Philosophical Perspectives, Vol. 19, Epistemology, Malden, MA: Blackwell Publishing, 445–59. White, R. (2013) “Evidence Cannot be Permissive,” in Matthias Steup, John Turri, and Ernest Sosa (Eds.), Contemporary Debates in Epistemology, 2nd edn. Malden, MA: Wiley-Blackwell, 312–23.

17 Disagreement: Idealized and Everyday Jonathan Matheson

I’ve never met an epistemic peer I didn’t like. I’ve never met an epistemic peer I’ve disagreed with either. Epistemic peers are hard to come by. I think that I’ve encountered exactly one—myself. By no means is this situation unique to me. I imagine the same is true of you—that you are the only epistemic peer of yourself that you have ever encountered, and that you have never found yourself to disagree with yourself, at least not at a single time. I have changed my mind about various things, but my earlier selves are not epistemic peers of my current self, just as your earlier selves are not epistemic peers of your current self. Often my current self has thought about the matter more than my earlier self, has attained more evidence on the matter, and so forth. Other times my current self has forgotten relevant bits of information or how they fit together. Given this situation, it might seem puzzling that almost all of the literature on the epistemic significance of disagreement concerns disagreement with an epistemic peer.1 While the issue of the epistemic significance of disagreement is often motivated by pointing out the widespread and persistent disagreement that we encounter regarding a number of topics in morality, politics, religion, and philosophy in general, the discussion often quickly turns to a particular kind of discovered disagreement—an idealized disagreement with an epistemic peer. However, this is a kind of disagreement that we simply don’t encounter. So, while the question of what we should believe in cases of idealized disagreement is intrinsically interesting, we should want more from an account of the epistemology of disagreement. We should want an account that tells us what we should believe in the everyday cases of disagreement that we are all too familiar with. At this stage in the debate, the epistemic impact of the disagreements that we are aware of is far from obvious.

  For a notable exception, and someone who raises this problem as well, see Nathan King (2011).

1

316  Jonathan Matheson In this essay I want to go some way towards rectifying this matter. I will begin by briefly discussing cases of idealized peer disagreement. First, I will set up what an evidentially symmetric disagreement looks like. I will then briefly explain and motivate one popular account of the epistemic significance of disagreement, the Equal Weight View, and examine its claims about cases of idealized peer disagreement. I will then proceed to extend my conclusions to be made about cases of idealized disagreement to the more everyday cases of disagreement. In doing so it will become apparent both why the focus in the literature has been on the idealized cases of disagreement and why this is a good thing.

17.1  Idealized Disagreements Let’s begin by looking at evidentially symmetric cases of disagreement. While much of the literature on disagreement has focused on evidentially symmetrical disagreements, such disagreements have often been mischaracterized, resulting in a case where the evidence is not symmetrical after all. As I am understanding them, an evidentially symmetric case of disagreement between S1 and S2 meets each of the following conditions: ( i) S1 and S2 are justified in believing that they are epistemic peers at t. (ii) S1 and S2 are justified in believing that they have adopted competitor doxastic attitudes, D1 and D2, toward a proposition p at t. (iii) S1 and S2 are each as justified in believing that S2 has adopted D2 toward p at t as that S1 has adopted D1 toward p at t. (iv) S1 and S2 are justified in believing that there are no circumstantial factors that make it the case that either S1 or S2 is more likely to be correct about p at t. Some clarifications are in order. So understood, an evidentially symmetrical case of disagreement need not actually be a case of disagreement at all, it must simply be a case where the parties are reasonable in taking there to be a particular kind of disagreement. That is, it may be that S1 and S2 in fact have adopted the same doxastic attitude toward p. One’s interlocutor can mislead, and one’s evidence can mislead, so cases where one is rational in taking there to be a disagreement need not be cases where there is in fact a disagreement. All four conditions can be met even when S1 and S2 have in fact adopted the same doxastic attitude toward p. What’s important is that one’s evidence supports that there is a disagreement. Conditions (i) through (iv) each claim that the parties to the disagreement are justified in believing a particular proposition. Meeting these conditions does not entail that either of them actually believes these propositions, just that they are propositions that it is reasonable for them to believe at that time. So, the focus in evidentially symmetrical disagreements is on propositional justification. Condition (i) requires the parties to be justified in believing that they are epistemic peers. While there have been slightly different takes on what constitutes an epistemic peer, the central component is that epistemic peers are equally likely to be correct—that

Disagreement: Idealized and Everyday  317 they are in an equally good epistemic position regarding the disputed proposition. We can subdivide factors which contribute to the likelihood of being correct as follows: (a) Faculty factors: S1 and S2 are equally intelligent, equal in terms of reasoning ability, perceptual ability, etc. (b) Evidential factors: S1 and S2 have the same, or equally good, evidence. (c) Virtue factors: S1 and S2 are equally intellectually virtuous (open-minded, intellectually courageous, etc.). One quick note is in order regarding “equally good” evidence. Two bodies of evidence are equally good with respect to a proposition when they each provide the same degree of justification for that proposition. Richard Feldman’s (2006) Dean on the Quad case is an example of a disagreement where the parties have distinct but equally good bodies of evidence. In this case, two individuals look across the quad. One sees the Dean, and the other does not. Both disagreeing parties have distinct perceptual bodies of evidence since they have distinct perceptual experiences. While these are distinct bodies of evidence, they each support their possessor’s conclusion equally well since each party has equally good reasons to believe that they are perceptually adept. Another example of distinct but equally good bodies of evidence is given by David Christensen (2007). Christensen imagines a case where the two disagreeing parties had each conducted their own poll on the issue at hand with populations of the same size, equally representative, etc. Upon discovering the disagreement, each party knows that the other has a distinct body of evidence, but one that is just as good as her own. Regarding (ii), competitor doxastic attitudes are rival alternative doxastic attitudes. A single body of evidence cannot support each of two competitor doxastic attitudes.2 So, a competitor doxastic attitude to D1 is a distinct doxastic attitude that is not simply related in a nesting way to D1. If we adopt a tripartite doxastic taxonomy, then believing, disbelieving, and suspending judgment are all competitor doxastic attitudes. On a more fine-grained doxastic taxonomy believing to degree 0.7 and believing to degree 0.2 would be competitor doxastic attitudes. It might be that not all distinct doxastic attitudes are competitors. For instance, it may be that all-out believing, and believing to degree 0.7 are not in fact competitor doxastic attitudes, but different descriptions of the same doxastic attitude on two distinct levels of appraisal. I include ‘competitor’ to not rule out this possibility. Condition (iii) is a condition of an evidentially symmetrical disagreement that is often ignored. This condition requires that S1 have just as good of reason to believe that S2 has adopted D2 toward p at t, as S1 does regarding the claim that S1 has adopted D1 toward p at t, and the same for S2. Typically we have introspective evidence of our own doxastic attitudes and we have testimonial evidence of the doxastic attitudes of others. To meet condition (iii) in such cases, the introspective evidence that S1 has that   2  Here I am assuming the Uniqueness Thesis. For a richer defense of the Uniqueness Thesis, see Roger White (2005) and Matheson (2011).

318  Jonathan Matheson S1 believes p must be just as good (provide just as much justification for believing) as the testimonial evidence that S1 has that S2 disbelieves p. If this condition is not met, then S1 will have at least some reason to favor either S1 or S2’s conclusion (depending on where the better reasons point). For there to be the perfect evidential symmetry between S1 and S2, meeting condition (iii) is crucial. Finally condition (iv) is included to make sure that neither party’s evidence gives him or her any reason to think that on this particular occasion either of the parties is more likely to be correct on the matter, even if in general they are equally good at such matters and are equally informed on this issue. Epistemic peers need not be equally likely to be correct regarding every proposition on every occasion. Condition (iv) is there to rule out cases where one of the parties has good reason to think that one of the parties has been compromised (S1 or S2 is drunk, overly sleepy, biased on this topic, etc.), as well as cases where one of the parties has good reason to think that one of the parties is particularly adept on this occasion (had a peppermint, is a morning person, etc.).

17.2  The Equal Weight View Suppose that there is a two party case that meets (i) through (iv). The question at the heart of most of the literature on the epistemic significance of disagreement concerns what S1 and S2 are justified in believing at t’, the time immediately following the discovered disagreement at t. Perhaps the most discussed answer to this question is the Equal Weight View. According to an Equal Weight View, having encountered an evidentially symmetrical case of disagreement S1 and S2 ought to give each other’s conclusion equal weight as his or her own. According to Equal Weight Views, in such an idealized case of peer disagreement, no one peer opinion counts for more or less than any other peer opinion. While this seems plausible, this answer naturally invites the question, what does it mean to give a conclusion “equal weight”? Equal Weight Views have often been thought to entail splitting the difference between the disparate doxastic attitudes—that is, adopting the doxastic attitude halfway between the two competitor doxastic attitudes.3 Let’s suppose that D1, S1’s doxastic attitude toward p at t, is believing to degree 0.7, and that D2, S2’s doxastic attitude toward p at t, is believing to degree 0.3. A split the difference view has it that if conditions (i) through (iv) are met, then S1 and S2 should each believe p to degree 0.5 at t. Believing p to degree 0.5 is the doxastic attitude which splits the difference between D1 and D2. Holding an Equal Weight View, however, does not entail holding a “split the difference” view. For one thing, it is consistent with giving another’s conclusion equal weight that one gives neither peer conclusion any weight. Such a view about the epistemic  

  See Christensen (2007) and Kelly (2010).

3

Disagreement: Idealized and Everyday  319 significance of disagreement would be an Equal Weight View, but not a split the difference view. In addition, splitting the difference only appears to be a plausible way of giving another’s conclusion equal weight when the two parties to the disagreement are of a certain sort. If S1 and S2 are both known to be unreliable in forming beliefs about p-like matters, then splitting the difference does not appear to be an epistemically appropriate response even when conditions (i)–(iv) are met. If neither party is likely to be right about the disputed matter, then splitting the difference would not be a rational response to the discovery of the disagreement. Rather, it is only in cases where S1 and S2 are not only justified in believing that they are epistemic peers, but are also justified in believing that they are reliable in forming beliefs about p-like matters, that splitting the difference is called for. Indeed, the literature on the epistemology of disagreement has focused on this very type of case (although typically without explicitly noting it). It has been disagreement between rational individuals—reliable evidence evaluators—that has been of principle interest in the literature, not philosophical disagreements between unreliable children (even if the children are epistemic peers). Not all ways of being equals have the same epistemic significance. The discovery of a two-party evidentially idealized peer disagreement between individuals that are reliable in forming beliefs about matters like p does seem to call for both parties to split the difference. If both parties are justified in believing that they are equally likely to be correct about p on this occasion and yet have adopted competitor doxastic attitudes toward p, then to adopt a doxastic attitude toward p other than the one which splits the difference would be to show preference or bias toward one of the competing attitudes. So, splitting the difference seems to be the appropriate way to give each peer conclusion equal weight in such circumstances. When understanding Equal Weight Views as split the difference views, we should therefore add the following condition to our conditions for evidentially symmetrical disagreements: (v) S1 and S2 are justified in believing that they are reliable evaluators of the evidence regarding p-like matters. To meet this condition, S1 and S2 must be such that it is reasonable for them to think that they are good at judging the evidence regarding matters like p and subsequently forming the doxastic attitude toward such propositions which fits their evidence. Let’s then call a case that meets conditions (i)–(v) a case of rationally idealized peer disagreement (an RIPD). Notice that this label does not imply that both parties are justified in having adopted their respective doxastic attitudes at t, but only that it is an apparent disagreement between individuals that are a certain sort of epistemic peers—rational peers—individuals who are likely to evaluate the relevant type of evidence correctly. In what follows I will be taking Equal Weight Views in a two-person RIPD to call for splitting the difference between the two competitor doxastic attitudes. While

320  Jonathan Matheson this isn’t the only way to give peer opinions equal weight, this seems to be the view that defenders of Equal Weight Views have had in mind in these cases.4 So, in what follows I will be understanding the Equal Weight View (EWV) as follows: EWV: In a two-person RIPD, S1 and S2 should split the difference and each adopt the doxastic attitude toward p that is midway between D1 and D2. The following case given by Thomas Kelly (2010) can help us unpack and apply EWV: Case 4: Despite having access to the same substantial body of evidence E, you and I arrive at very different opinions about some hypothesis H: while I am quite confident that H is true, you are quite confident that it is false. Indeed, at time t0, immediately before encountering one another, my credence for H stands at.8 while your credence stands at.2. At time t1, you and I meet and compare notes. How, if at all should we revise our respective beliefs? (Kelly 2010: 122)

Understanding this case as a two-party RIPD, you and I should each assign a credence of.5 to H at t1. A credence of.5 is the attitude which splits the difference between us. This verdict is plausible since in meeting the RIPD conditions, the evidence is symmetrically balanced between you and I having adopted the correct doxastic attitude toward p given our evidence. Although we are both likely to be right about H, according to our evidence, neither of us is more likely to be right than the other. Given conditions (i)–(v) it is reasonable to believe that there exists an evidential symmetry between us. It is as though I have gone through the reasoning twice having come up with a different conclusion on each occasion (one being given by me, and the other being given by you). Having met (i)–(v) it appears that the only difference lies in who has adopted each competing opinion, but this factor seems to be an epistemically inert piece of information.5 So, given this scenario, it seems plausible that we should meet each other halfway and split the difference. Any other alternative answer appears to be either arbitrary or prejudiced.6

17.3  Everyday Disagreements While EWV appears to be a plausible account of how idealized cases of peer disagreement should be handled, it is reasonable to want more from an account of the epistemic significance of disagreement. While the question of how idealized cases of disagreement should be handled is intrinsically interesting, the epistemic significance of disagreement extends beyond these idealized cases. Conditions (i)–(v) are rarely (if ever) met, yet it seems that the disagreements that we are aware of should still have some

  4  See Feldman (2006), Christensen (2007), and Elga (2007).   5  One other difference which has been cited by some to be relevant is which attitude is supported by our shared first-order evidence. See Kelly (2005, 2010) for defenses of this claim, and Matheson (2009) for a sound argument for the conclusion that this difference does not have an epistemic effect.   6  In what follows I will be taking it that EWV gives the correct verdicts about RIPDs, however much of the argument that follows does not hang on this.

Disagreement: Idealized and Everyday  321 effect on what we believe about disputed matters. The people we find ourselves disagreeing with are simply never exactly as intelligent as us, exactly as informed, exactly as open-minded, etc. Epistemic peerhood simply does not obtain.7 That said, can we learn anything about how to rationally handle everyday cases of disagreement from how idealized cases should be handled? Suppose EWV is correct about RIPDs. What could we learn from this about how we should respond to the everyday disagreements that we encounter? In what follows I want to examine what consequences stripping off the various idealizations of two-party RIPDs would have for what the parties of the disagreement should believe having encountered it. Fortunately, what we can learn from two-party RIPDs can teach us something about how to respond everyday disagreements. So, what I hope to offer is a way of moving from verdicts about idealized cases of disagreement to verdicts about everyday disagreements. Recall that a RIDP met the following conditions: ( i) S1 and S2 are justified in believing that they are epistemic peers at t. (ii) S1 and S2 are justified in believing that they have adopted competitor doxastic attitudes, D1 and D2, toward a proposition p at t. (iii) S1 and S2 are each as justified in believing that S2 has adopted D2 toward p at t as that S1 has adopted D1 toward p at t. (iv) S1 and S2 are justified in believing that there are no circumstantial factors that make it the case that either S1 or S2 is more likely to be correct about p at t. (v) S1 and S2 are justified in believing that they are reliable evaluators of the evidence regarding p-like matters. I want to now consider the justificatory effects of stripping away these idealizations. Condition (ii) is not an idealized condition at all. It is met in our everyday cases of disagreement. We are often justified in believing that other people disagree with us about various propositions. Meeting (ii) only requires that we are reasonable in believing that there is a disagreement. Similarly, condition (v) is not overly idealized. Most of us are justified in believing this most of the time. We are often justified in believing that we are reliable evaluators of the evidence, and that we are pretty good at evaluating the evidence regarding the matter under dispute. Recall, that this condition is only in place to ignore cases where the party of a disagreement is justified in believing that he is likely to be mistaken on the matter. That is not a feature shared by many of the everyday disagreements that we are interested in. The idealizations, therefore, lie in (i), (iii), and (iv). Let’s begin with condition (i). Condition (i) can be met even when S1 and S2 are not in fact epistemic peers—it only requires that they are justified in believing that they are. That said, one’s evidence rarely (if ever) supports that another individual is one’s

 

  For a more detailed argument for this conclusion, see King (2011).

7

322  Jonathan Matheson epistemic peer. For S1’s evidence to support that S2 is his peer, it must support that they are equals in their evidence, faculties, and intellectual virtues. Let’s first consider equality in faculties and the effects of stripping away the idealization here. If S1 is justified in believing that he is superior to S2 in terms of reasoning ability or intelligence, then this will give S1 some reason to favor S1’s conclusion regarding the disputed proposition—this will give S1 a reason to make more modest concessions to S2 (to not give S2’s conclusion equal weight). The degree to which S1 is justified in believing that he is superior to S2 in terms of reasoning ability and intelligence will typically affect how much conciliation is to be made. If S1 is justified in believing that he is only slightly superior to S2 in this way, then everything else being equal, S1 will not have to give S2’s conclusion equal weight, but he will still have to make quite significant conciliation toward S2. Everything else being equal, the more superior in terms of reasoning ability and intelligence that S1 is justified in believing that he is, the less conciliation will be required of him from the discovery of the disagreement with S2. So, being justified in believing oneself superior to one’s interlocutor in terms of reasoning ability and intelligence is a reason to make less conciliation. One might worry at this point that if one has any epistemic reason to prefer one’s own conclusion she should not make any doxastic conciliation. Support for this thought might come from the Uniqueness Thesis.8 If the Uniqueness Thesis is true, a body of evidence justifies at most one competitor attitude toward a proposition given. Applied to our case, if the evidence supports only one doxastic attitude, and our subject has a little more reason to think that he adopted that attitude, then it might seem that no conciliation would be called for. If only one of the doxastic attitudes toward p is justified by the evidence, and S1 has some reason to think that it is his over his interlocutor’s, then why should S1 make any conciliation at all? After all, neither party adopted an attitude between the two, so there might seem to be little reason to make conciliation and adopt one of those attitudes. While there is some pull to this theoretical worry, when it is applied to cases it can be seen to give the wrong verdicts. Suppose that you have two thermometers in the same room, and that the room has only one temperature at a time. Suppose further that one of the thermometers is slightly more reliable than the other, though both thermometers are reliable. Let’s say that you find out that the slightly more reliable thermometer read “80” at t and then later discover that the slightly less reliable thermometer read “74” at t. In such a case it seems that you shouldn’t simply stick with believing that the temperature in the room was 80 at t with the same level of confidence, even if that is what the more reliable thermometer says. You have acquired some powerful evidence that the temperature in the room is not 80 degrees—evidence from the other reliable (though slightly less so) thermometer. Though this other evidence is not as strong as the evidence that it is 80 degrees in the room, it is evidence that makes a difference as   8  For a more detailed discussion and defense of the Uniqueness Thesis, see White (2005), Feldman (2006), and Matheson (2011). For criticism, see Ballantyne and Coffman (2011).

Disagreement: Idealized and Everyday  323 to how justified you are in believing that it is 80 degrees in the room, and thus which attitude you should adopt toward that proposition. So, even when S1 is justified in believing that he is more intelligent than S2, S1 might still in the end be epistemically required to make quite significant doxastic conciliation toward S2. Further, while having reason to believe that you are intellectually superior gives you a reason to make less conciliation, reasons can be defeated. This bit of evidence concerning one’s intellectual superiority is only one piece of the evidential puzzle, and this bit of evidence may be supplemented with other bits of evidence which create inequalities between S1 and S2 in the other direction and which in turn require more conciliation of S1. In addition, some disagreements may make the impact of inequality along intellectual lines quite insubstantial. If the disagreement pertains to the color of ordinary sized objects in good light, the day of the week, the location of the nearest gas station, etc., then even if S1 is justified in believing that she is in general more intelligent or that she has superior reasoning abilities, this may not factor in at all, or at least not very significantly, regarding what she should believe in this case having discovered the disagreement. Such an asymmetry may not make S1 any more likely to be right on the disputed matter, and so long as that is the case, then there is no reason for S1 to privilege her own conclusion even if she is justified in believing that she is more intelligent. A physicist may be no more likely to be correct about the location of the nearest Starbucks than a 5th grader, even though the physicist is more intelligent. If S1 is still justified in believing that S2 is just as likely to be correct about such matters, despite the inequality in terms of intellectual faculties, then the discovered disagreement will have the same epistemic effect in the non-idealized case as it would have in the idealized case. The second factor in epistemic peerhood is evidential equality—that S1 and S2 have the same, or equally good, evidence. It is plausible that no two individuals ever actually have the same body of evidence, and it is only slightly less idealized for them to have distinct but equally good bodies of evidence. What are we to say about the epistemic significance of disagreement when this evidential equality condition is not met? Cases in which the parties of the disagreement differ only with respect to evidence which does not affect their justification for believing the disputed proposition (whether in quantity or quality), will make the discovery of the disagreement have precisely the same epistemological effect as cases where there is equality in these matters (holding all the other factors fixed, that is). So long as S1 is justified in believing that S2’s evidence regarding the disputed proposition p is just as good as S1’s, then evidential differences (in either quantity or quality) unrelated to the justification of p will not affect how S1 ought to respond to the discovery of the disagreement. Using the same example, though the physicist has a lot of evidence that the 5th grader lacks about physics, if they have equally good evidence about the location of the Starbucks, this evidential difference won’t affect what is epistemically required of the disputants. It is also important to remember here that the evidence pertaining to one’s justification for believing p does not simply include evidence directly pertaining to p, but also any

324  Jonathan Matheson relevant higher-order evidence such as evidence about the nature of evidence and the significance of disagreements.9 Differences in the evidence possessed by the parties that are relevant to the justification of the disputed proposition, however, can make for conciliatory differences between the idealized case and the everyday case. If S1 is justified in believing that he has more evidence on the issue (i.e., all the evidence S2 has on the matter, plus some additional evidence), then he will have reason to make more modest concessions. In such a scenario, S1 will have reason to think that he is more likely to be correct on the matter, and thus has reason not to give S2’s conclusion equal weight, and perhaps even reason to give S2’s conclusion no weight at all. We can imagine here a case where S1 has in addition to all of S2’s evidence an undercutting defeater which takes away all of the justification provided for p by S2’s total evidence. So, even small differences in evidence may call for quite large deviations in terms of how much conciliation is called for from the idealized case. Alternatively, if S1 is justified in believing that S2’s evidential position toward p is superior to S1’s (in quantity or quality), then S1 will have reason to make even more conciliation to S2 than in the idealized case. In such a scenario S1 will have reason to believe that S2 is more likely than S1 to be correct on the matter, and thus has reason not to give S2’s conclusion equal weight, but to give it even more weight than her own. Finally, the third condition of epistemic peerhood is equality in terms of epistemic virtues. To be epistemic peers, S1 and S2 must be equally intellectually virtuous—they must be equally open-minded, equally intellectually courageous, equally careful, and so forth. By now it should be clear how stripping away the idealizations will affect the significance of the disagreement. The more S1 has reason to believe that she is superior in terms of intellectual virtue, the less reason she has to make conciliation to S2, and the less reason she has to give S2’s conclusion equal weight. The more S1 has reason to believe that S2 is superior in terms of intellectual virtue, the more reason S1 has to give even more deference to S2. As with evidential factors, it may be that some intellectual virtues are such that their possession does not figure into the likeliness of being correct regarding some matters. Perhaps if the disagreement pertains to the color of ordinary sized objects in good light, the day of the week, the location of the nearest gas station, and so forth, being aware that one’s interlocutor differs in terms of her intellectual courage will not affect how likely she is to be correct on the matter. So, some differences in intellectual virtue will seemingly make for no difference at all regarding the epistemic significance of disagreement. Let’s turn now to condition (iii). Condition (iii) is idealized in that it is typically not met. Typically an individual will have better reason to believe that she believes what she does than she will have regarding what another believes. Plausibly, this is because introspection can provide better reasons than testimony typically does. Propositions

 

  For more on higher-order evidence, see Matheson (2009) and Kelly (2010).

9

Disagreement: Idealized and Everyday  325 that people believe from introspection are typically more justified than propositions that people believe from testimony.10 Eliminating this evidential symmetry between S1 and S2 will have epistemic effects. If S1 is more justified in believing that S1 has adopted D1 toward p at t1 than S1 is justified in believing that S2 has adopted D2 toward p at t1 (as S1 typically will be), then S1 will have some reason to favor his own conclusion. In such a scenario S1 has some reason to privilege D1 as a response to the evidence regarding p over D2.11 So here too, when the idealization is stripped away, S1 will typically have a reason to favor the doxastic attitude that S1 is more justified in believing is actually held by one of the parties.12 Like above, however, how such a reason will interact with the other asymmetrical reasons that S1 may have is a complicated matter that will depend on what those actual reasons are. The condition regarding circumstantial factors in RIPDs, condition (iv), is also idealized. In idealized cases of peer disagreement S1 has no reason coming from the circumstances to believe that S1 is more or less likely to be correct on this occasion than S2. S1 would have such a reason were he to have reason to believe that there was something about the disputed proposition or the encounter in general which made it more likely for S1 to be correct. For instance, if S1 had reason to believe that S2 is drunk (and that S1 is not), or that S2 is very tired (and that S1 is not), or that p is about the attractiveness of S2’s spouse (and not S1’s) or the usefulness of S2’s occupation (and not S1’s), then S1 would have reason to think that the balance tips in his favor. In such situations S1 would have reason to think that it is S2 that is mistaken due to some circumstantial factor. In typical cases, however, S1 will have at least some such reasons. Often such reasons will be minor and will not significantly mitigate the amount of conciliation required of S1 from the discovery of the disagreement. If S1 is justified in believing that S2 has had a beer or just finished teaching before adopting D2 toward p (while S1 had not), then this might give S1 some reason to discount S2’s conclusion, but not a reason that significantly mitigates the justificatory effect of the discovery of the disagreement. In such a case S1 may not be required to give S2’s conclusion precisely equal weight, but he will be required to give it something very close to it. The more significant the reason to discount the other’s conclusion, the more mitigated the justificatory effect of the reason to make conciliation will be. S1 will typically also have some reasons to discount his own conclusion. Perhaps it is he who has ingested a moderate amount of alcohol before considering the disputed proposition, while his opponent has not. Here   10  I say “typically” since it can be the case that an individual has better reason to trust the testimony of another about the other’s beliefs, than our subject has reason to trust her introspection about her own beliefs. The details of the particular case will matter.   11  Perhaps this is what Jennifer Lackey (2008) is tracking when she argues that “personal information” can often be used to privilege one’s own opinion. As I am understanding it, if one does have such privileging personal information, then one is not in an idealized case of disagreement.   12  S1 will only typically have such a reason since it is possible that the evidence one gains via testimony about S2’s doxastic attitude regarding p is actually stronger that S1’s evidence about which doxastic attitude S1 has himself adopted toward p.

326  Jonathan Matheson again, the reasons on both sides will need to be weighed against each other. If S1 has on balance stronger reasons to discount S2’s conclusion due to circumstantial factors, then S1 has a reason to make less conciliation. If S1 has on balance stronger reasons to discount his own conclusion due to circumstantial factors, then S1 has reason to make even more conciliation. Finally, the discussion of RIPDs has also focused on two-person cases of disagreement. Many of our everyday disagreements are not like this. Regarding most of our political, religious, ethical, and philosophical beliefs in general, we are aware of a multitude of people who disagree with us, as well as a number of people who agree with us. EWV has mistakenly been thought to imply that discovering a solitary peer disagreement can have dramatic effects on what it is rational for us to believe.13 This is perhaps the fault of defenders of EWV since their accounts typically state meeting a disagreeing peer as being a sufficient condition for making substantial doxastic conciliation.14 Doing so, however, would not amount to giving each peer opinion equal weight—something that EWV is committed to. Rather, such an account would give the opinions of disagreeing peers much more weight than the opinions of the agreeing peers in allowing for one disagreeing peer to call for significant doxastic revision in the face of widespread agreement.15 So, extrapolating from the two-party RIPDs, each opinion of a rational cognizer is to be given some weight. Evidence that S1 has regarding the existence of other agreeing parties will give additional evidential force to D1, whereas evidence that S1 has regarding the existence of other disagreeing parties will give evidential force that pushes away from D1 (possibly in the direction of D2, but possibly in the opposite direction). The weight to be given to each of these opinions will depend on the information that S1 has about each of these individuals in ways that we have covered above. Plausibly, the weight to be given to an individual’s opinion corresponds to how likely that individual is to be right on the matter. Clearly, the more individuals whose opinion on the matter S1 is aware of, the more complicated the evidential story will be. In particular, the independence of the various opinions (or lack thereof) will make a difference as to how much weight each opinion is to be given. Issues of independence are difficult indeed. The independence of opinions comes in degrees. Plausibly the more independent two agreeing opinions are, the more weight each individual opinion is to be given. In contrast, the less independent two agreeing opinions are, the less weight there is to be given to the second agreeing opinion.16

  13  See Lackey (2008).   14  For examples of this, see Feldman (2006) and Elga (2007).   15  For more on this point, see Carey and Matheson (2013). So understood, EWV also mishandles cases when disagreeing peers are discovered at different times. On this understanding of EWV, EWV has certain peer opinions counting for more depending upon when they were discovered. Clearly, the time of discovery should not affect the weight of the peer opinion—to do so would not be to give peer opinions equal weight.   16  For more on the consequences of the difficulty of determining independence, see Carey and Matheson (2013).

Disagreement: Idealized and Everyday  327 We have seen in quite broad strokes what justificatory effect would result from stripping away our idealized conditions in various ways. Stripping away the idealizations resulted in either no difference or giving S1 a reason to give either S1’s or S2’s conclusion more weight. Nothing has been said about how those reasons add up in everyday cases. Unfortunately, not much else can be truly said about such non-idealized cases in universal terms. How these reasons to favor one’s own conclusion, or to make greater conciliation, add up will be a matter of what the particular bits of evidence are, and thus it will be unique to particular cases of disagreement. Sometimes a reason to think that there are circumstantial factors counting against S2 will be much more significant than the fact that S1 is justified in believing that S2 has a superior body of evidence regarding the disputed proposition; other times it will not. For instance, if I am justified in believing that S2 has much more evidence on the matter, but I am also justified in believing that S2 is severely intoxicated, then I don’t have much reason to make much (or even any) conciliation toward S2. Sometimes S1 being justified in believing that he is more intelligent than S2 will be more significant than S1 having a reason to discount his own conclusion; other times it will not. In cases of non-idealized disagreement, the particular details matter. How each of these bits of evidence adds up will depend upon what exactly the reasons are and even what the disputed proposition concerns. Although we have seen how inequalities within the various factors contributing to epistemic peerhood give reasons to change one’s opinion, how those reasons add up is a complicated matter that depends upon what those particular inequalities are. This is not to say that investigating idealized cases of peer disagreement is unimportant. Not only is such an investigation intrinsically interesting, but the verdicts in idealized cases of disagreement set the baseline for the verdicts to be given in everyday disagreement. It is important to see what the epistemically appropriate response is in two-party RIPDs since the epistemically appropriate response to everyday disagreements will depend upon the ways in which those everyday disagreements differ from the two-party idealized case. We have seen that we cannot derive true interesting and universal principles about these everyday disagreements, but how much conciliation should be made in everyday disagreements will depend on what the correct story is about idealized disagreements, coupled with how the asymmetrical reasons add up.17

17.4  An Ethical Analogy This account of the epistemic significance of disagreement bears some similarity to W.D. Ross’s ethical theory that utilizes prima facie duties.18 Ross’s theory is a particularist ethical theory in that what one is obligated to do is determined by the particular circumstances. Ross’s theory gives seven prima facie duties which are features of an

   

  For more on this point, see Feldman (2009).   See Ross (1930).

17

18

328  Jonathan Matheson act which tend to make it one’s actual duty (or duty proper). How these prima facie duties add up in any particular case, however, depends on the details of the particular case. One’s actual duty (duty proper) is to follow the most stringent prima facie duty or duties, but what the most stringent prima facie duty or duties are is a feature of particular cases—different prima facie duties can be the most stringent based upon the particular details of the case at hand. So it is with our account of the epistemic significance of everyday disagreement. Regarding it, true and precise universal claims cannot be made about particular non-idealized cases of disagreement. We have seen in broad strokes various ways in which when one is in a non-idealized case of disagreement one thereby gets a reason to make more or less conciliation than in the idealized case, but how those reasons add up depends upon the details of the particular case and what those reasons consist of. Like Ross’s account of what makes for one’s actual obligation, which doxastic attitude one should adopt after discovering the disagreement will depend upon how the particular reasons add up. The result is a kind of epistemic particularism. We have seen how various pieces of evidence provide prima facie justification for believing that oneself is more or less likely to be correct on the matter, but whether one is on balance justified in so believing will depend upon how those bits of evidence intermingle and add up. The resulting picture for everyday disagreements is messy and perhaps less than we would like to have, but unfortunately it’s all we can get.19

References Ballantyne, Nathan and Coffman, E. J. (2011). “Uniqueness, Evidence, and Rationality,” Phil­osophers’ Imprint 11(18): 1–13. Carey, Brandon and Matheson, Jonathan. (2013). “How Skeptical is the Equal Weight View?” in Diego Machuca (Ed.), Disagreement and Skepticism. New York: Routledge, 131–49. Christensen, David. (2007). “Epistemology of Disagreement: The Good News,” Philosophical Review 116: 187–218. Elga, Adam. (2007). “Reflection and Disagreement,” Noûs 41: 478–502. Feldman, Richard. (2006). “Reasonable Religious Disagreements,” in L. Antony (Ed.), Philo­sophers without Gods: Meditations on Atheism and the Secular Life. New York: Oxford University Press, 194–214. Feldman, Richard. (2009). “Evidentialism, Higher-Order Evidence, and Disagreement,” Episteme 6(3): 294–312. Kelly, Thomas (2005). “The Epistemic Significance of Disagreement,” in T. Gendler and J. Haw­thorne (Eds.), Oxford Studies in Epistemology, Vol. 1. Oxford: Oxford University Press, 167–96. Kelly, Thomas. (2010). “Peer Disagreement and Higher-Order Evidence,” in R. Feldman and T. Warfield (Eds.), Disagreement. New York: Oxford University Press, 111–74.   19  Special thanks to Tomas Bogardus, Brandon Carey, Earl Conee, Richard Feldman, Nate King, and Kevin McCain for helpful comments on earlier drafts.

Disagreement: Idealized and Everyday  329 King, Nathan. (2011). “Disagreement: What’s the Problem? Or A Good Peer is Hard to Find,” Philosophy and Phenomenological Research 85(2): 249–72. Lackey, Jennifer. (2008). “A Justificationalist View of Disagreement’s Epistemic Significance,” in Adrian Haddock, Alan Millar, and Duncan Pritchard (Eds.), Social Epistemology. Oxford: Oxford University Press, 298–325. Matheson, Jonathan. (2009). “Conciliatory Views of Disagreement and Higher-Order Evi­dence,” Episteme: A Journal of Social Philosophy 6(3): 269–79. Matheson, Jonathan. (2011). “The Case for Rational Uniqueness,” Logos and Episteme:  An Inter­national Journal of Epistemology 2(3): 359–73. Ross, W. D. (1930). The Right and the Good. Oxford: Oxford University Press. White, Roger. (2005). “Epistemic Permissiveness,” in J. Hawthorne (Ed.), Philosophical ­Perspec­tives, Vol. 19, Epistemology, Malden, MA: Blackwell, 445–59.

Index Abelson, R. P.  169, 178–9 Abramson, K.  205 n. 12 acceptance  44, 44 n. 24, 210, 229, 232, 269 actions basic  6, 18–20, 34–5, 37, 46–7 non-basic 18–20 Adams, F.  42 n. 22 Adams, R. M.  16 n. 3, 51 n. 2, 71 Addelson, K. P.  230 Additivity of Evidence (AE)  296 Adler, J.  35 n. 8, 41 n. 20, 250 n. 8, 257 n. 18, 259 n. 21 Aguilar, J.  46 n. 25 akrasia  52, 63, 65–9, 74 Alarracín, D. 30 Alcoff, L. 226 Alston, W.  17 n. 7, 34 n. 3, 44 n. 24, 112 n. 1, 137 n. 6 Altmann, R. W.  205 n. 12 Alznauer, M. 261 Ambiguous Question Thesis  3, 4 Anderson, E.  216, 221 n. 5, 224, 234, 237 Anderson, J. 192 Annas, J.  64 n. 22 Anscombe, G. E. M.  38 n. 17, 51, 267 n. 11, 268, 269, 271 Aristotle  2, 6, 7, 51–2, 63–6, 66 n. 23, 67, 67 n. 24, 68–70, 73, 73 n. 33, 74, 140, 158, 160 n. 27, 162 Aronson, E.  28 n. 22, 178, 185 n. 2 Aquinas, T.  2, 162 Asch, S.  206 n. 14 assertion  9, 10, 17, 84–5, 90, 101, 247, 261–82 Assertion-Belief biconditional (AB)  278–80 Assertion-Belief-Link (ABL)  278–80 attunement 9, 244 Audi, R.  35 n. 10, 154 n. 15, 162 n. 32, 164 n. 35, 203 n. 8 authority, see epistemic authority Axtell, G.  140 n. 14, 146 n. 2, 148 n. 6 Bach, K.  262 n. 1 Baehr, J.  71, 72, 146 n. 2, 147 n. 5, 148 n. 6, 160, 160 n. 28 Ballantyne, N.  106 n. 16, 322 n. 8 Balloun, J. 22 Barnett, B.  290 n. 8, 291 n. 9 Bar On, B. A.  225 Bar-Tal, D. 175 Battaly, H.  6, 68 n. 26, 69 n. 29, 74 n. 34, 261 Bedke, M. S.  138 n. 19

Bélanger, J. 170 beliefs first-order  10, 83, 84, 256 higher-order  10, 274, 277 Bennett, J.  20, 28, 35 n. 8, 35 n. 9, 36 n. 12 Bergmann, M.  287 n. 4 Bessenoff, G.  30 n. 26 Bird, A.  96 n. 3, 103 n. 13, 106 n. 18 Bishop, J.  34 n. 4, 44 n. 24, 46 n. 25 Bleier, R. 222 Bogardus, T.  302 n. 3, 328 n. 19 Bonjour, L.  112 n. 1, 252 n. 11 Boyatzi, L. 8 Brand, M.  42 n. 22 Bratman, M.  17 n. 7, 42 n. 22, 44 n. 24 Brehm, J. W.  180 Brock, T. 22 Brogaard, B.  7, 139, 140, 141 n. 15 Broome, J.  135 n. 5, 138 n. 10 Bruner, J. 175 Brunero, J. 138 Buckareff, A.  6, 33 n. 1, 34 n. 5, 43 n. 23, 44 n. 24, 46 n. 25 Burge, T.  139 n. 12 Burr, V. 188 Cantril, H. 21 Carelli, P. 261 Carey, B.  326 n. 15, 326 n. 16, 328 n. 19 Carlsmith, K.  185 n. 2 Categorical Imperative  9 Chen, X. 170 Chignell, A. 132 Chisholm, R.  112 n. 1, 218 Choi, I. 176 Christensen, D.  249 n. 6, 287 n. 4, 293, 293 n. 11, 302 n. 2, 309, 317, 318 n. 3, 320 n. 4 Chun, Y. W.  172, 176 Church, J.  35 n. 8, 41 n. 20 Cialdini, R.  8, 193, 195 Clarke, D.  35 n. 8, 41 n. 20 Clifford, W.K .  1–4, 7, 130–4, 141–3, 147, 147 n. 4, 148, 152–3 Code, L.  223 n. 8, 224–6, 230–3, 237 n. 21 Coffman, E. J.  97 n. 5, 106 n. 16, 261, 322 n. 8 Cohen, L.  44 n. 24 Cohen, A. R.  180 Cohen, S.  232, 302 n. 3 Comfort, N. C.  254 n. 15

332 Index Common Knowledge regarding the Norm of Assertion (CKNA)  262–3, 266–7, 270, 272–3, 275, 282 Conciliationism  10, 302, 302 n. 4, 303–11 conditional fallacy  95, 107 Conee, E.  48 n. 26, 105 n. 15, 107 n. 19, 147 n. 5, 155, 155 n. 20, 156, 159, 228, 249 n. 7, 285 n. 3, 328 n. 19 confirmation bias  21 contextualism  85, 87, 217, 224, 232 competence, see epistemic competence Coppenger, B.  164 n. 35 Crenshaw, M. 190 critical rationalism  8–9, 186 Cullison, A.  48 n. 26 Dancy, J.  135 n. 5 Danto, A.  34 n. 6 Darwall, J.  135 n. 5 Daukas, N.  231, 237–8 David, M.  112 n. 1, 112 n. 2, 114 n. 5, 139 n. 12, 159 n. 25 Davis, R. 261 DeAngelis, T. 190 Debes, R.  205 n. 12 Dechesne, M.  8, 173, 176, 190, 261 defeat  269, 272–4, 277–8, 296–9, 323–4 deliberation  16–18, 27, 35, 40, 42–6, 135, 158–9, 270 demon world  53, 55, 57–8, 60–2, 70–2, 95–6, 245 DeGrada, E. 177 DePaul, M.  112 n. 2, 118 n. 11, 119 n. 13 de Roon, C.  8 DeRose, K.  101 n. 9, 155, 156, 157, 158, 232 Descartes, R.  1, 2, 245 Diaz-Leon, E.  89, 98 n. 6 Dijkersterhuis, A. P.  177 Direct Doxastic Voluntarism (DDV)  6, 34–6, 41, 47 disagreement  10, 284–5, 287–90, 292–3, 295, 299, 301–5, 312, 315–7, 319–28 dogmatism  57, 63, 73, 77–8, 80, 250–1 Dougherty, T.  8, 146 n. 2, 147 n. 5, 148 n. 7, 163 n. 34 doxastic control  4–6, 34–6 doxastic voluntarism  4, 6, 15–21, 27–8, 30, 34, 154 see also Direct Doxastic Voluntarism Driver, J.  53 n. 3 Duggan, K. E.  261 Dutch Books  257 Edgely, R. 39 Ekman, P. 60 Elga, A.  287 n. 4, 301 n. 1, 302 n. 2, 320 n. 4, 326 n. 14 Elgin, C.  9, 69 n. 28, 75 n. 35, 124 n. 19, 227 Eliot, G.  68, 68 n. 27 Ellis, S. 175 Engel, P.  44 n. 24 entitlement  264, 265, 267, 269

epistemology feminist epistemology 9, 216, 217 , 218, 219 knowledge-fi rst epistemology 7, 89–90, 101–10 naturalized epistemology 9, 217, 224, 229–30 social epistemology 5, 216, 218, 228 epistemic authority  8, 175, 176, 180, 187, 195, 207–8, 213, 227, 256, 258, 263–8, 270–1, 273, 282 epistemic competence  53, 78, 81–7, 295, 302 epistemic individualism  9, 244 epistemic injustice,  9, 56, 58–62, 141, 219, 231, 234, 236, 254, 269, 270–1 epistemic peers, see peers epistemic privilege  225, 230, 232–3, 244, 302, 323, 325 epistemic responsibility  8, 33, 147–51, 160–1, 163–4, 237, 264, 266–7, 271–2, 282 epistemic value  4, 7, 21–2, 40, 54–9, 61–4, 69–70, 72, 74–5, 112–27, 151, 155, 162, 219–21, 235, 255 Equal Weight View  10, 304, 309, 318–20, 326 Erb, H. P.  172 Ethics of Assertion biconditional (EA)  278–80 Ethics of Higher-Order Belief (EHOB)  277 eudaimonia 140–1 Evans, J. S. B. T.  173 evidence, first-order  290, 320 evidence, higher-order  163, 285, 324 Evidence of Evidence Principle  10, 290–2, 295–9 Evidential Duplication and Knowledge Producing Theory  109 Evidential Duplication Theory  105 Evidence sufficicient to Warrant Belief (EWB) 280 evidentialism  10, 77, 83, 132–3, 142, 147, 155, 285–8, 294 externalism  77, 83, 217 fake barn cases  103 Fales, E.  164 n. 35 Fehr, C.  226 n. 12 Feldman, R.  10, 20 n. 11, 34 n. 3, 48 n. 26, 105 n. 14, 105 n. 15, 107 n. 19, 132, 137, 147 n. 5, 148 n. 8, 154 n. 15, 154 n. 17, 155, 155 n. 20, 156, 163 n. 34, 261, 284 n. 1, 285 n. 2, 285 n. 3, 287 n. 4, 288 n. 5, 288 n. 6, 302, 302 n. 3, 302 n. 4, 317, 320 n. 4, 322 n. 8, 326 n. 14, 327 n. 17, 328 n. 19 Festinger, L.  178, 179 Feyerabend, P. 169 Finchman, J. R. S.  254 Finn, H.  66 n. 23 Fitelson, B.  298, 298 n. 13, 306 n. 7 folkpsychology  8, 184, 185, 186 Foley, R.  112 n. 1, 159, 159 n. 26 Fong, G. 178 Frankish, K.  6, 16–18, 18 n. 9, 26 n. 19, 27, 35 n. 8, 44 n. 24 Freedman, J. L.  178

Index  333 Freund, T. 177 Fricker, M.  56, 58, 58 n. 13, 59, 59 n. 14, 59 n. 15, 60, 231, 234, 234 n. 19, 236, 254 n. 13, 267 n. 11, 269, 270, 271, 271 n. 16 Fumerton, R.  164 n. 35 Funkhouser, E.  35 n. 8, 36 n. 12 Gadsden, C.  164 n. 35 Garcia, J.  42 n. 22 Garrett, D.  212 n. 25, 213 n. 27 Gawronski, B.  169, 178 Geach, P. T.  114 n. 4 Gelfert, A.  124 n. 18 Gendler, T.  252, 252 n. 10 Gergen, K. 188 Gergen, M. M.  188 Gerken, M.  102 n. 11 Gettier, E.  58 n. 12, 95, 100–1, 103, 105, 124, 158 n. 24, 217, 266, 275 Gigerenzer, G.  170, 172–5 Gilbert, D.  17 n. 7, 23, 24, 29 n. 25, 30 n. 26 Gilbert, J.  96 n. 4 Ginet, C.  17 n. 4, 34 n. 3 Giuliano, T. 170 Glaser, J. 178 God  20, 90, 154, 308, 309, 311 Goldberg, S.  9–10, 69 n. 28, 75 n. 35, 258 n. 19, 262 n. 1, 265 n. 9, 265 n. 10 Goldman, A.  5, 58, 105 n. 14, 105 n. 15, 108 n. 20, 112 n. 1, 113 n. 3, 119 n. 13, 218, 218 n. 2, 258 n. 19, 261 Goldstein, D. G.  173 Gorman, M.  202 n. 6 Govier, T.  35 n. 8, 36 n. 12 Grasswick, H.  9, 75 n. 35, 224 n. 9, 261 Greco, J.  105 n. 15, 118 n. 10, 140 n. 14 Green, A.  213 n. 27 Greenspan, P.S.  135 n. 5 Grimm, S.  124 n. 19, 154 Guimond, S.  213 n. 26 Ha, Y.  21 n. 14 Haack, S.  220, 221 n. 5 Habermas, J. 255 Haddock, A.  113 n. 3, 114 n. 4, 117 n. 9, 119 n. 12, 124 n. 19, 239 n. 23 Haney, M.  75 n. 35, 261 Haraway, D.  222, 224 Hardin, C. D.  180 Harding, S.  224, 226, 226 n. 11, 227, 231, 238 Harman, G.  77, 80 Harnisch, R.  262 n. 1 Hasan, A.  164 n. 35 Haslanger, S. 219 Hastorf, A. 21 Hawthorne, J.  101 n. 9, 143 n. 16 Haybron, D. 140

Hazlett, A.  127 n. 21 Heider, F. 178 Heil, J.  154 n. 15 Hertel, P. 170 Hieronymy, P.  21 n. 13, 34 n. 5 Higgins, E. T.  180 Hill, T.  135 n. 5 Hinchman, T.  265 n. 6, 265 n. 10, 267 n. 11, 269 Hitler, A. 72 Holdon, T.  213 n. 27 Holton, R.  18, 43 Hookway, C. 68 Houminer, D. 175 Howard-Snyder, D. 89 Howard-Snyder, F. 89 Hsiao, T.  149 n. 9 Hubbard, R. 222 Hudson, H.  89, 298 n. 12 Hume, D.  1, 9, 149 n. 10, 154, 198–9, 199 n. 1, 199 n. 2, 200–13 Hursthouse, R.  55 n. 8 Identity Thesis  148 inferential processes  8, 28–9, 170 inferential rules  8, 170, 174–6 internalism  70–1, 77, 83, 217 inquiry  5, 7–8, 67, 69, 78–81, 112, 114, 119–26, 147–8, 155, 162–4, 220–1, 224, 228, 230, 234–5, 238–9, 254–5, 257–8 Inquiry-Stopper Problem  123–5 Irwin, T. 73 Jaggar, A. M.  225 James, W.  1, 3 Jamieson, D. W.  177 Jehle, D.  306 n. 7 Jeter, D. 68 Johnson, O. A.  201 n. 6 Jones, W.  39, 48 n. 26, 117 n. 9 Jost, J. T.  178 Kant, I.  9, 256 Kappel, K.  124 n. 18 Keller, E. F.  254 n. 15 Kelly, T.  10, 105 n. 14, 138, 138 n. 8, 142, 143, 153 n. 26, 261, 285 n. 2, 287 n. 4, 301 n. 2, 302 n. 2, 307 n. 8, 308 n. 10, 318 n. 3, 320, 320 n. 5, 324 n. 9 Kelp, C.  124 n. 18 Kempen, H. 191 Kesebir, P. 187 King, N.  315 n. 1, 321 n. 7, 328 n. 19 Kitayama, S. 176 Kitcher, P.  229, 234 Klayman, J.  21 n. 14 Kleffner, D. A.  174 Klein, P.  163 n. 34

334 Index Knowles, J. 217 Komdat, J.  164 n. 35 Köpetz, C. 170 Korcz, K. A.  157 n. 23 Kornblith, H.  147 n. 5, 154 n. 15, 302 n. 3 Koro-Ljunberg, M. 188 Korsgaard, C.  134 n. 3 Kourany, J. 233 Kraut, R. 140 Kripke, S.  77, 78, 80 Kruglanski, A.  8–9, 169, 170, 173, 177–9, 187, 191–2, 195, 207 n. 16, 213 n. 26, 261 Krull, D. S.  177 Kuhn, T.  169, 180, 223 Kuhnen, U. 176 Kumkale, G. T.  30 Kunda, Z. 178 Kvanvig, J.  112 n. 2, 117 n. 9, 118 n. 11, 120 n. 14, 124 n. 18, 124 n. 19, 151 n. 13, 157 n. 22, 159 n. 25 Lackey, J.  261, 287 n. 4, 325 n. 11, 326 n. 12 Lakatos, I.  169, 180 lay epistemics  8, 170, 171, 175 Lebiere, C. 192 Lehrer, K.  112 n. 1, 157 n. 22 Leiter, B. 147 Leman, P. J.  213 n. 26 Levy, N.  6, 16 n. 3 Littlejohn, C.  89, 97 n. 5, 101 n. 8, 164 n. 35 Lloyd, L. A.  222 Locke, J. 154 Longino, H.  220–1, 221 n. 5, 225, 226–7, 227 n. 13, 229, 229 n. 15, 230, 238 n. 22 Lord, C. 21 lottery propositions  280 n. 23 luck  53, 55, 57–8, 62, 69–70, 72, 87, 124, 155, 193, 253, 258–9, 266, 275 Lugones, M. 233 MacFarlane, J.  262 n. 1 McCain, K.  328 n. 19 McCann, H.  42 n. 22 McClintock, B.  254, 254 n. 14 McCormic, T.  35 n. 11 McGlynn, A.  89, 103 n. 13 McGowan, M. K.  254 n. 13 McGuire, W. J.  178, 195 McHugh, C.  35 n. 11 Macrae, C. 187 Maffee, J. 217 Malmgren, A.  143 n. 16 Mandelbaum, E.  6, 18 n. 9, 25, 27 n. 20, 29 n. 24 Manley, D.  103 n. 12 Mannetti, L. 177 Markus, H. R.  176

Marlow, K.  143 n. 16 Marsh, G. 89 Martin, D.  105, 106 Martin, E. 222 Masuda, T. 176 Matheson, J.  10, 75 n. 35, 89, 106 n. 16, 127 n. 21, 139 n. 13, 143 n. 16, 216 n. 1, 259 n. 21, 261, 288 n. 6, 289 n. 7, 290 n. 8, 302 n. 3, 317 n. 2, 320 n. 5, 322 n. 8, 324 n. 9, 326 n. 15, 326 n. 16 Mele, A.  35 n. 9, 39 n. 18, 41 n. 21, 42 n. 22 Mental Duplication Theory  103 Milgram, S. 187 Mill, J. S.  255, 255 n. 16, 259 Millar, A.  113 n. 3, 114 n. 4, 117 n. 9, 118 n. 10, 119 n. 12, 124 n. 18, 124 n. 19, 217, 239 n. 23 Mischel, W. 173 Mittag, D.  48 n. 26 Molyneux problem  249 n. 5 Monti, M. M.  19 Montmarquet, J.  54 n. 7, 55, 55 n. 9, 57 n. 10, 137 n. 6, 140 n. 14, 160, 237 Moon, A.  164 n. 35 Moran, R.  265 n. 6, 267 n. 11, 269 Morris, M. W.  177 Moser, P.  112 n. 1 Mueller, A.  143 n. 16 Musgrave, A.  169, 180 Narrow-Scope Animal Killing Theory  91–2 Narrow-Scope Counterfactual Theory  95 Narrow-Scope Identity Theory  93, 94 Narrow-Scope requirements  91–3, 97, 135 Narrow-Scope Revised  96 Nelson, L. H.  222, 225, 230, 238 Neta, R.  105 n. 14, 232 Newcomb, T. M.  178 Non-Belief Conditions Theory  98–9 Non-Belief Knowledge-Constitutive Conditions Theory 99 Non-Belief Requiring Conditions Theory 98–9 Non-Belief Requiring Knowledge-Constitutive Conditions Theory  100–1 Norenzayan, A. 176 Nottelmann, N.  34 n. 1, 137 n. 6 Nisbett, R.  39 n. 18, 175, 176, 177 Obama, B.  4, 95–7 O’Brien, L.  33 n. 1 Olsson, E.  113 n. 3 Orehek, E. 173 Orwell, G.  9, 245, 245 n. 1, 246 n. 2, 254, 255 O’Shaugnessy, B.  35 n. 8, 41 n. 20 Owen, A. M.  19 Owens, D.  33 n. 2, 67 n. 25, 265 n. 6 Oyserman, D. 176

Index  335 Pagin, P.  262 n. 1 Palmer, D. L.  213 n. 26 Pascal, B.  20, 90 Past Evidential Duplication Theory  105 Past Mental Duplication Theory  103, 104 Peacocke, C.  139 n. 12 Pears, D.  202 n. 7 peers  10, 272, 284–5, 287–90, 292–3, 299, 301 n. 1, 304, 306–8, 315–22, 326–7 Peng, K.  176–7 Pepitone, A. 180 Pieper, L.  35 n. 8, 36 n. 12 Pierro, A.  173, 177 Pinnick, C.  230 n. 17, 231 n. 17 Plantinga, A.  112 n. 1, 137 n. 6, 149 n. 12 Plato  2, 5, 53 n. 3, 158, 169 Pogin, K.  164 n. 35 Pojman, L.  35 n. 8 Popper, K.  169, 171, 172, 180, 186, 187 post-modernism 73 Poston, T. 261 Potter, E. 226 pragmatic encroachment  77, 79, 85, 87 Pratkanis, A. R.  29 Price, H. H.  201 n. 6 Pritchard, D.  7, 113 n. 3, 114 n. 4, 115 n. 7, 117 n. 9, 119 n. 12, 124 n. 19, 239 n. 23 process availability duplication theory  108 Proctor, R. N.  236 n. 20 Pyszczynski, T. 187 Quadflieg, S. 187 Radcliffe, D.  35 n. 8 Ramachandran, V. S.  174 Raviv, A.  175, 213 n. 26 Reed, B.  143 n. 16, 261 reflective knowledge  84 reliabilism  53, 77, 81, 83–4, 86 n. 3, 107, 108–9, 117 n. 9, 148, 217, 222 n. 7, 245, 287 see also virtue reliabilism responsibility, see epistemic responsibility Reynolds, S.  103 n. 13 Rieber, R.  184 n. 1 Riggs, W.  117 n. 9 Roberts, R. C.  160, 237 Robinson, D.  184 n. 1 Rogers, J.  139 n. 13 Rolin, K.  232 n. 18 Rooney, P.  223 n. 9, 224, 228 n. 14 Rorty, A. 67 Rosenberg, M. J.  178 Ross, A.  264, 264 n. 5, 265 n. 6 Ross, L.  25, 26, 39 n. 18 Ross, W. D.  327, 327 n. 18, 328 Rowley, W.  290 n. 8

Russell, B.  201 n. 6 Ryan, S.  34 n. 3, 35 n. 11, 137 n. 6, 146 n. 1, 154 n. 15 Saint, M. 89 Santioso, R. 178 Schacter, D. L.  29 n. 26 Schaper, C. 177 Schiebinger, L.  236 n. 20 Schmid, A. 188 Schmitt, F. 216 Schmitt, M.  164 n. 35 Schroeder, M.  91 n. 1, 138 n. 10 Schwitzgebel, E. 18 Scott-Kakures, D.  35 n. 8, 41 n. 20 Searle, J.  38 n. 17 self- deception  33, 39 self-indulgence 68–9 self-trust 84 Sellars, W.  252 n. 9 Setiya, K.  21 n. 13, 35 n. 8 Shafer-Landau, R.  133 n. 2 Shah, N.  38–40, 44 n. 24, 115 n. 7, 139 n. 12 Sherman, J.  30 n. 26 Shipley, J.  164 n. 35 Shoda, Y. 173 Shope, R.  95 n. 2 Shteynberg, G. 170 Siegfried, C. H.  228 n. 14 Simple View (of phenomenal action)  42 n. 22 Single Phenomenon View  42 n. 22 skepticism  69, 232, 245–6, 282 Slote, M.  54 n. 7, 55 n. 8 Smith, A. M.  16 n. 3 Smith, B.  134 n. 3, 135, 141 n. 15 Smith, H.  15 n. 2 Smith, Q.  163 n. 34 Smithies, D.  99 n. 6 social constructivism  8, 188–92, 216 Sosa, E.  7, 51 n. 1, 53, 53 n. 4, 54 n. 5, 54 n. 6, 58, 58 n. 11, 58 n. 12, 75 n. 35, 81 n. 1, 86 n. 3, 112 n. 1, 119 n. 12, 140 n. 14, 220, 235, 261 Soteriou, M.  33 n. 1 Stalnaker, R.  44 n. 24, 267 n. 13 Stanley, J.  101 n. 9 Stapel, D.  8, 193, 194 Stanovich, K. E.  27 Steup, M.  19 n. 10, 34 n. 3, 35 n. 11, 137 n. 6, 227, 228 Steward, S. 89 Stich, S. 24 Stocker, M. 71 Stone, T.  39 n. 18 Strack, F.  169, 178 Strong Value Evidentialism (SVE)  155 Strong Voluntarism  16–17, 19 Sullaway, F. J.  178 Sullivan, S.  235, 236 n. 20

336 Index Sustained at Will, Principle  37 Sutton, J.  91, 93, 97 n. 5, 101 n. 9 Swamping Problem  7, 116, 117 Swank, C.  51 n. 2 Swinburne, R.  117 n. 9 Sylvan, K.  164 n. 35 Tannenbaum, P. H.  178 Taylor, M. 190 terrorism 8, 188–9 testimonial justice  59–62, 231 Thibodeau, R.  28 n. 22 Thompson, E. P.  172, 187 Total Evidence View  302 n. 4, 303 Treanor, N.  89, 121–2, 123 n. 15, 127 n. 21 Trianosky, G.  70 n. 31, 75 n. 35 Trivial Truths Problem  119–20 Truth Norm  139, 140 truth thesis  113–27 Tuana, N.  235, 236 n. 20 Tulving, E.  30 n. 26 Tuomela, R.  44 n. 24 Turri, J.  114 n. 4, 115 n. 6, 127 n. 21, 140 n. 14 Ulrich, W. 187 understanding  7, 55, 119, 124–27, 146, 205, 257 Unified Question Thesis  3–4 Uniqueness Thesis  106, 288, 294, 317 n. 2, 322 Up-to-Now Evidential Duplication Theory 105–6 Up-to-Now Mental Duplication Theory  104 Up-to-Now Evidential and KnowledgeProducing Process Availability Duplication Theory 108 value, see epistemic value Van Knippenberg, A. D.  177 Velleman, J. D.  38 n. 17, 44 n. 24, 115 n. 7, 139 n. 12 veritism, see truth thesis vice  6–7, 51–3, 56–75, 147, 159, 210–11 virtue  4, 6, 9, 51–8, 60, 62–8, 71, 73–4, 86, 136, 140–1, 147–8, 158–60, 184, 186, 200, 208–13, 217, 221, 224, 231, 233–4, 236–9, 317, 322, 324 reliabilism  6, 51, 53–4, 56–8, 60–2, 75, 148, 237 responsibilism  6, 51–2, 55–6, 59, 63–5, 68, 71, 74–5, 148, 237–8

Virtuous Belief Principle  209 Vitz, R.  9, 40 n. 19, 48 n. 26, 75 n. 35, 205 n. 12, 208 n. 19, 209 n. 20, 209 n. 21, 209 n. 22, 212 n. 24, 216 n. 1, 261 Vogel, J. 246 Wallace, R.  135 n. 5 Wanderer, J.  267 n. 11, 270, 270 n. 15 Wasserman, R.  89, 99 n. 7 Watson, G.  265 n. 6 Weak Voluntarism  16–18 Webster, D. M.  177 Wedgwood, R.  115 n. 7, 139 n. 12 Wegner, D. M.  26, 170–1 Weiner, M.  262 n. 1 Whitcomb, D.  7, 101 n. 10, 107 n. 19 White, R.  106 n. 16, 301 n. 2, 317 n. 2, 322 n. 8 Wicklund, R. A.  180 Wide-Scope Animal Killing Theory  91–3 Wide-Scope Counterfactual Theory  97 Wide-Scope Identity Theory  94–5 wide-scope requirements  7, 92–3, 130, 133–5, 138, 143 Williams, B.  6, 34–5, 35 n. 8, 36–8, 47–8, 247 n. 3 Williams, M.  232, 232 n. 18 Williamson, T.  90, 93, 101 n. 9, 105 n. 14, 106 n. 17, 107 n. 19, 124 n. 18, 263 n. 3 Wilson, M.  258 n. 20 Wilson, T.  185 n. 2 Winters, B.  35, 35 n. 8, 36–7, 37 n. 13, 38, 38 n. 15, 48 wishful thinking  33, 58, 63, 67, 251 Wittgenstein, L.  252, 252 n. 9 Wood, W. J.  160, 237 Wright, S.  69 n. 30, 75 n. 35, 261 Wundt, W.  8, 184–6, 188 Wylie, A.  225, 232 Young, A.  39 n. 18 Zagzebski, L.  54 n. 7, 55, 55 n. 9, 56, 59, 63, 116 n. 8, 117 n. 9, 124 n. 19, 140 n. 14 Zajonc, R. B.  178 Zanna, M. P.  177 Zylvan, K.  163 n. 33