Political Epistemology 2021931432, 9780192893338, 9780192645371

Citation preview

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Political Epistemology

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

M I N D A S S O C IAT IO N O C C A SIO NA L SE R I E S Tis series consists of carefully selected volumes of signifcant original papers on predefned themes, normally growing out of a conference supported by a Mind Association Major Conference Grant. Te Association nominates an editor or editors for each collection, and may cooperate with other bodies in promoting conferences or other scholarly activities in connection with the preparation of particular volumes. Director, Mind Association: Daniel Whiting Publications Ofcer: Eliot Michaelson Recently published in the series Te Language of Ontology Edited by J. T. M. Miller Quine, Structure and Ontology Edited by Frederique Janssen-­Lauret In the Light of Experience Edited by Johan Gersel, Rasmus Tybo Jensen, Morten S. Taning, and Søren Overgaard Evaluative Perception Edited by Anna Bergqvist and Robert Cowan Perceptual Ephemera Edited by Tomas Crowther and Clare Mac Cumhaill Common Sense in the Scottish Enlightenment Edited by C. B. Bow Art and Belief Edited by Ema Sullivan-­Bissett, Helen Bradley, and Paul Noordhof

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Te Actual and the Possible Edited by Mark Sinclair Tinking about the Emotions Edited by Alix Cohen and Robert Stern Te Social and Political Philosophy of Mary Wollstonecraf Edited by Sandrine Bergès and Alan Cofee Te Epistemic Life of Groups Edited by Michael S. Brady and Miranda Fricker Reality Making Edited by Mark Jago

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Political Epistemology Edited by

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

ELIZABETH EDENBERG and M IC HA E L HA N N O N

1

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

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 2021 Te moral rights of the authors have been asserted First Edition published in 2021 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: 2021931432 ISBN 978–0–19–289333–8 DOI: 10.1093/oso/9780192893338.001.0001

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Printed and bound in the UK by TJ Books Limited 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.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Acknowledgments Tis volume would not have been possible without generous support from several funders and people. Firstly, we’d like to thank the Aristotelian Society, the Society for Applied Philosophy, and the Mind Association for providing the funds to organize the inaugural Political Epistemology Network event in London 2018, where the idea for this volume frst took shape. We would also like to thank the British Academy, the Society for Applied Philosophy, and especially Georgetown University and Ethics Lab for supporting a workshop on Epistemology, Democracy, and Disagreement in October 2018. Tis allowed us to invite many contributors of this volume to Washington D.C. for two days of spirited discussion about the contents of this book. Tanks to everyone who participated in these events. We’d also like to thank the Institute of Philosophy at the University of London, and especially Barry Smith and Richard Somerville, for providing the venue, funding, and support to run a series of events on political epistemology that shaped this volume, as well as for providing Elizabeth Edenberg with a visiting fellowship to make the collaborative work on this project possible. Michael Hannon would also like to thank Jeroen de Ridder and the Dutch Research Council for supporting his research as part of the NWO Vidi grant on “Knowledgeable Democracy: A Social-­ Epistemological Inquiry” (Project 276-­20-­024). A number of other people deserve thanks for helping us to develop this project over the past two years. In particular, we’d like to thank Maggie Little, Robin McKenna, Scott Sturgeon, Étienne Brown, Boudewijn de Bruin, Kate Elgin, Jonathan Healey, Syndey Luken, August Gorman, Lukas Chandler, Michael Tschiderer, Patricia Martin, Roxie France-­ Nuriddin, Daniel Sulmasy, and two anonymous reviewers. Tanks also to Marie Traore for her fantastic help preparing the index for this volume and to Baruch College, for providing support for her research assistance. We’d also like to thank our editor, Peter Momtchilof, as well as Sarah Sawyer from the Mind Association, who initially suggested that we include this edited volume as part of the Mind Association Occasional Series to be published with Oxford University Press. We are incredibly grateful to both of them for their encouragement. Also, we want to apologize to anyone we forgot to mention by name and say thank you for your help. Tis book is the result of many years of thinking about political epistemology, and we are sure that we have learned things from more people than we can now remember. Finally, we’d like to thank every contributor to this volume: Kristofer Ahlstrom-­ Vij, Elizabeth Anderson, Jason Brennan, Quassim Cassam, Tomas Christiano, David Estlund, Alexander Guerrero, Jennifer Lackey, Michael Lynch, Fabienne Peter, Jeroen de Ridder, Regina Rini, Jennifer R. Steele, Robert Talisse, and Briana Toole. When we frst envisioned our dream team of contributors, we never expected all of you to fnd the time to contribute to this project. Tank you so much for being part of this. It was a pleasure to work on this project with you all.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved. Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Contents List of Contributors

ix

Introduction1 Elizabeth Edenberg and Michael Hannon I .   T RU T H A N D K N OW L E D G E I N P O L I T IC S 1. Epistemic Bubbles and Authoritarian Politics Elizabeth Anderson 2. Weaponized Skepticism: An Analysis of Social Media Deception as Applied Political Epistemology Regina Rini

11

31

3. Bullshit, Post-­truth, and Propaganda Quassim Cassam

49

4. Truth and Uncertainty in Political Justifcation Fabienne Peter

64

5. What Lies Beneath: Te Epistemic Roots of White Supremacy Briana Toole

76

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

I I .   E P I S T E M IC P R O B L E M S F O R D E M O C R AC Y 6. Epistocratic Paternalism David Estlund 7. Te Basis of Political Equality Tomas Christiano

97 114

8. Does Public Reason Liberalism Rest on a Mistake? Democracy’s Doxastic and Epistemic Problems Jason Brennan

135

9. Te Epistemic Pathologies of Elections and the Epistemic Promise of Lottocracy Alexander Guerrero

156

10. Policy, Ignorance, and the Will of the People: Te Case of “Good Immigrants” Kristofer Ahlstrom-­Vij and Jennifer R. Steele

180

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

viii Contents

I I I .   D I S AG R E E M E N T A N D P O L A R I Z AT IO N 11. Problems of Polarization Robert B. Talisse

209

12. Deep Disagreements and Political Polarization Jeroen de Ridder

226

13. Political Disagreement, Arrogance, and the Pursuit of Truth Michael P. Lynch

244

14. Te Problem with Disagreement on Social Media: Moral not Epistemic Elizabeth Edenberg 15. When Should We Disagree about Politics? Jennifer Lackey

259 280

297

Index

319

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

16. Disagreement or Badmouthing? Te Role of Expressive Discourse in Politics Michael Hannon

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

List of Contributors Kristofer Ahlstrom-­Vij is Reader in Philosophy at Birkbeck College, University of London, and Research Fellow at the National Institute of Economic and Social Research, London. He works in social epistemology, both theoretical and applied, with a particular interest in the value of truth and the ways in which society can be set up to promote it. At present, he is exploring the role of factual information in political attitude formation, and how to re-­think democratic processes and institutions in light of the problems posed by public ignorance. His publications include Epistemic Consequentialism (Oxford University Press, 2018), Epistemic Paternalism: A Defense (Palgrave Macmillan, 2013), and articles in Noûs, Philosophy and Phenomenological Research, Philosophical Studies, and other journals.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Elizabeth Anderson is Arthur F. Turnau Professor and John Dewey Distinguished University Professor of Philosophy and Women’s Studies at the University of Michigan. Her research covers topics in moral and political philosophy and epistemology, including: democratic theory, equality in political philosophy and American law, racial integration, the ethical limits of markets, the history of egalitarianism, theories of value and rational choice, pragmatism, and feminist epistemology and philosophy of science. Anderson’s book Te Imperative of Integration (Princeton University Press, 2013), was winner of the American Philosophical Association’s 2011 Joseph B. Gittler Award for an outstanding scholarly contribution in the feld of the philosophy of the social sciences. Her most recent book, Private Government: How Employers Rule Our Lives (and Why We Don’t Talk About It) (Princeton University Press, 2017) is based on her Tanner Lectures, delivered at Princeton’s Center for Human Values. Jason Brennan  is Robert  J.  and Elizabeth Flanagan Family Professor of Strategy, Economics, Ethics, and Public Policy at the McDonough School of Business at Georgetown University. He is the author of thirteen books, including Injustice for All (Routledge Press, 2019), with Chris Surprenant and When All Else Fails (Princeton University Press, 2018). Quassim Cassam is Professor of Philosophy at the University of Warwick. He was previously Knightbridge Professor of Philosophy at Cambridge University, Professor of Philosophy at UCL, and Reader in Philosophy at Oxford University. He is the author of seven books, including Vices of the Mind: From the Intellectual to the Political (Oxford University Press, 2019), Conspiracy Teories (Polity, 2019), and Extremism: A Philosophical Analysis (Routledge, 2021). Tomas Christiano is Professor of Philosophy and Law at the University of Arizona. His current research is in moral and political philosophy with emphases on democratic theory, distributive justice, and global justice. He is co-­editor of Politics, Philosophy and Economics (Sage). His books include Te Constitution of Equality:

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

x  List of Contributors Democratic Authority and Its Limits (Oxford University Press, 2008), Te Rule of Te Many: Fundamental Issues in Democratic Teory (Westview Press, 1996), and the edited volume Philosophy and Democracy: An Anthology (Oxford University Press, 2003). He is currently working on books on economic justice, citizen participation, and the foundations of equality. Elizabeth Edenberg is Assistant Professor of Philosophy at Baruch College, Te City University of New York. Prior to joining Baruch College, she was Senior Ethicist and Assistant Research Professor at Georgetown University’s Ethics Lab where she led translational ethics projects, collaborating with teams of computer scientists, lawyers, and social scientists investigating emerging ethical and political challenges posed by technology. She specializes in political philosophy, political epistemology, and the ethics of emerging technologies. Her articles have appeared in journals such as Te Journal of Political Philosophy, Law and Philosophy, New Media & Society, and IEEE: Security and Privacy.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

David Estlund is Lombardo Family Professor of the Humanities and Philosophy at Brown University. His research interests include liberalism, justice and injustice, democracy, dissent, and ideal theory. He previously taught at University of California, Irvine, and has spent fellowship years at Harvard’s Program in Ethics and the Professions and at Australian National University. He is editor most recently of Te Oxford Handbook of Political Philosophy (Oxford University Press, 2013) and author of Democratic Authority: A Philosophical Framework (Princeton University Press, 2008), and Utopophobia: On the Limits (If Any) of Political Philosophy (Princeton University Press, 2020). Alexander Guerrero  is Henry Rutgers Term Chair and Associate Professor of Philosophy at Rutgers University. His work in epistemology, political and moral philosophy, and legal philosophy has appeared in journals such as Ethics, Philosophy and Public Afairs, Public Afairs Quarterly, and Philosophical Studies. One of his main areas of research is on “lottocracies,” which is the idea that lotteries, not elections, should be used to select political ofcials. He received his PhD from the NYU Philosophy Department in 2012, and he has a JD from NYU School of Law. Michael Hannon is Assistant Professor of Philosophy at the University of Nottingham and founder of the Political Epistemology Network. He is author of What’s the Point of Knowledge? (Oxford University Press, 2019) and co-­ editor of Te Routledge Handbook of Political Epistemology. He writes about skepticism, fallibilism, the value of knowledge, expressive political discourse, and the role of empathy in politics. He is currently writing a book for Routledge titled Political Epistemology: An Introduction. Jennifer Lackey is Wayne and Elizabeth Jones Professor of Philosophy at Northwestern University, Director of the Northwestern Prison Education Program, and Editor-­in-­ Chief of Philosophical Studies and Episteme. Most of her research is in the area of social epistemology, with recent work on the rationality of punishment, credibility and false confessions, eyewitness testimony and epistemic agency, epistemic reparations, the duty to object, and the epistemology of groups. She is the author of Te Epistemology of Groups (Oxford University Press, forthcoming) and Learning from

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

List of Contributors  xi Words: Testimony as a Source of Knowledge (Oxford University Press, 2008), editor of Academic Freedom (Oxford University Press, 2018) and Essays in Collective Epistemology (Oxford University Press, 2014), and co-­editor of Te Epistemology of Disagreement (Oxford University Press, 2013) and Te Epistemology of Testimony (Oxford University Press, 2006). She is the winner of the Dr. Martin R. Lebowitz and Eve Lewellis Lebowitz Prize for Philosophical Achievement and Contribution as well as the Young Epistemologist Prize, and she has received grants and fellowships from the Andrew W. Mellon Foundation, the American Council of Learned Societies, the National Endowment for the Humanities, and the Alice Kaplan Institute for the Humanities. Michael P. Lynch is Board of Trustees Distinguished Professor of Philosophy at the University of Connecticut, where he directs the Humanities Institute. His books include Te Internet of Us: Knowing More and Understanding Less in the Age of Big Data (Liveright, 2017), In Praise of Reason: Why Rationality Matters for Democracy (MIT Press, 2012), Truth as One and Many (Oxford University Press, 2009), and the New York Times Sunday Book Review Editor’s pick, True to Life (MIT Press, 2004). His most recent book, Know-­it-­All Society: Truth and Arrogance in Political Culture (Liveright, 2019) was the recipient of the George Orwell Award for Distinguished Contribution to Honesty and Clarity in Public Language.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Fabienne Peter is Professor of Philosophy at the University of Warwick, specializing in political philosophy, moral philosophy, and social epistemology. She has published extensively on political legitimacy and on political epistemology. She is the author of Democratic Legitimacy (Routledge, 2009) and is currently writing a book on Te Grounds of Political Legitimacy, which explores the meta-­normative foundations of conceptions of legitimacy. She has co-­edited Public Health Ethics and Equity (with Sudhir Anand and Amartya Sen) and Rationality and Commitment (with Hans Bernhard Schmid). She is a past editor of Economics and Philosophy. Jeroen de Ridder is Associate Professor of Philosophy at Vrije Universiteit Amsterdam, and Professor (by special appointment) of Christian Philosophy at the University of Groningen. He is currently president of Te Young Academy, the junior section of the Royal Netherlands Academy of Arts and Sciences. His research focuses on issues in social epistemology, philosophy of science, and philosophy of religion. He is principal investigator for a fve-­year research project on “Knowledgeable Democracy: A Social-­ Epistemological Inquiry,” funded by a Vidi grant from the Dutch Research Council (NWO). With Michael Hannon, he has co-­edited Te Routledge Handbook of Political Epistemology (Routledge, 2021). Regina Rini is Canada Research Chair in Philosophy of Moral and Social Cognition at York University in Toronto. She writes about moral agency, moral disagreement, normative change, social epistemology, and the philosophy of digital technology. Her work has appeared in numerous philosophy journals, as well as in public venues such as the Times Literary Supplement, New York Times, and Los Angeles Times. Her most recent book is Te Ethics of Microaggression (Routledge, 2020). She is currently writing a book about the efects of social media on democratic political culture.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

xii  List of Contributors Jennifer R. Steele is an Associate Professor of Psychology in the Faculty of Health at York University in Toronto, Canada, and the Director of the Interpersonal Perception and Social Cognition Laboratory. In her main line of research, Dr. Steele takes a social cognitive approach to understanding racial biases across the lifespan. She has used a variety of tools, including implicit measures, to assess the attitudes and beliefs of children and adults. Dr. Steele publishes regularly in high-­impact peer-­reviewed journals and her research has been funded by federal granting agencies including the Social Sciences and Humanities Research Council of Canada and the Canada Foundation for Innovation. Robert B. Talisse is W. Alton Jones Professor of Philosophy at Vanderbilt University. He specializes in contemporary political philosophy, with particular interest in democratic theory and liberalism. His most recent work engages issues at the intersection of political philosophy and epistemology; in addition, he pursues topics in pragmatism, analytic philosophy, and ancient philosophy. His books include Overdoing Democracy (Oxford University Press, 2019), Engaging Political Philosophy (Routledge, 2016), and Democracy and Moral Confict (Cambridge University Press, 2009).

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Briana Toole is an Assistant Professor in the Department of Philosophy at Claremont McKenna College. Toole is an epistemologist with interests in the philosophy of race and gender and feminism. Her work examines the relationship between social and political systems and the epistemic frameworks that support them. Toole is also the founder of the philosophy outreach program Corrupt the Youth.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Introduction

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Elizabeth Edenberg and Michael Hannon

As current events around the world have illustrated, epistemological issues are at the center of our political lives. It has become increasingly difcult to discern legitimate sources of evidence, misinformation spreads faster than ever, and the role of truth in politics has allegedly decayed in recent years. It is therefore no coincidence that political discourse is currently saturated with epistemic notions like “post-­truth,” “fake news,” “truth decay,” “echo chambers,” and “alternative facts.” Furthermore, disagreement between citizens is not only about moral and political values but also about what information is true and which experts we should trust. Te irreconcilable clash of political outlooks and the deep polarization characteristic of many contemporary societies endanger the common ground upon which the collective pursuit of truth depends. As a result, a worrying form of skepticism threatens our liberal democratic institutions. Tese issues are part of a rapidly growing area of research called “political epistemology.” While scholars have been interested in topics at the intersection of political philosophy and epistemology at least since Plato’s Republic, the term “political epistemology” only recently entered the academic lexicon and it does not yet point to a clear set of research questions or core topics. Still, we believe this term captures an important area of work at the intersection of political philosophy and epistemology that has become especially important in the current political climate. Tis includes work on propaganda and misinformation, political disagreement, polarization, conspiracy theories, the epistemology of democracy, voter ignorance and irrationality, skepticism wielded for political purposes, and the epistemic virtues (and vices) of citizens, politicians, and political institutions. If political epistemology has old roots, why has it only recently been recognized as a distinctive subfeld of philosophy? We hypothesize that two recent developments largely explain the emergence of political epistemology as a feld in its own right. First, new research in social epistemology (itself a relatively young feld) has centered on topics that bear on political life. Tis led to the development of conceptual tools that are readily applicable to political issues; for example, contemporary social epistemologists have studied the social dimensions of knowledge, the role of trust in testimony and reliance on experts, the epistemology of disagreement, judgment aggregation, and the design of social systems to realize our epistemic goals. Second, recent political developments seem to cry out for epistemological analysis. Te spread of fake news, the rise of conspiracy theories, the intensifcation of political polarization, the decline of trust in experts and the media, and the alleged era of “post-­truth” politics all indicate that we are in the midst of a politically charged epistemic crisis. As the April 2017 issue of the New Scientist declared: “Philosophers Elizabeth Edenberg and Michael Hannon, Introduction. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Elizabeth Edenberg and Michael Hannon. DOI: 10.1093/oso/9780192893338.003.0001

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2  Elizabeth Edenberg and Michael Hannon of knowledge, your time has come.” Philosophers have answered the call by seeking to play a crucial role in understanding and shaping our response to these political developments. Tis volume is part of that response. It brings together leading philosophers working at the intersection of epistemology and political philosophy to explore ways in which the analytic and conceptual tools of epistemology bear on political philosophy, and vice versa. By bringing these scholars together, this volume aims to generate new ideas that may help us better understand and make progress in the current crisis over the relevance of truth and knowledge in contemporary politics. Te volume is organized around three broad themes: the role of truth and knowledge in politics; epistemic problems for democracy; and disagreement and polarization. We selected these themes to refect some of the dominant areas of interest in contemporary political discourse. Tey capture broad questions about the role of epistemic considerations in justifying political authority, as well as more specifc applications to questions of skepticism and disagreement between citizens. Part I is broadly concerned with the role of truth and knowledge in politics. As Hannah Arendt wrote in her famous New York Times article, “no one ever doubted that truth and politics are on rather bad terms with each other.” Lies, bullshit, spin, and propaganda are not only tools of demagogues, but are seemingly permanent features of the political world. And yet we have allegedly entered the era of post-­ truth politics, where political life is characterized by intensifed epistemic anarchy. Is there a new war on truth and, if so, how do we fght back? Elizabeth Anderson examines how political discourse is increasingly distorted by “epistemic bubbles,” which threaten democracy because their members are vulnerable to charlatans, propagandists, and demagogues. She proposes two models of how populism creates epistemic bubbles: frst, by promulgating biased group norms of information processing; and second, by replacing empirically oriented policy discourse with identity-­ expressive discourse. Anderson then considers diferent strategies for popping epistemic bubbles. Her analysis suggests that social epistemology must get more social by modeling cognitive biases as operating collectively and outside people’s heads, via group epistemic and discursive norms. Relatedly, Regina Rini focuses on the ways in which Russian propagandists have hijacked social media to wreak epistemic and political damage. An increasingly common worry is that social media enables authoritarians to meddle in democratic politics; for instance, trolls and bots amplify deceptive content. But Rini argues that these tactics have a more insidious anti-­democratic purpose. Authoritarians who implant lies in democratic discourse ofen intend for these lies to be caught. Teir primary goal is not to deceive citizens but rather to undermine the democratic value of testimony. Rini illustrates how this makes democratic societies less resilient to authoritarian pressure by examining recent Russian social media interference operations. Rini and Anderson both draw general lessons about the new epistemic vulnerabilities that society faces, but they have diferent focal points: Anderson worries about the closed-­ mindedness of epistemic bubbles, whereas Rini is concerned about increasing levels of skepticism toward democratic institutions. While Anderson and Rini both look at problematic consequences of manipulations of our fragmented epistemic landscape, Quassim Cassam analyzes whether the

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Introduction  3 concepts of post-­truth and bullshit are useful tools for political-­epistemological analysis. He compares diferent accounts of bullshit and post-­truth before ultimately arguing that neither concept provides an adequate conceptualization of the political tactics that have come to the fore in recent years. Indeed, he argues that an emphasis on bullshit and post-­truth trivializes and misdescribes these tactics, which are more adequately conceptualized in terms of propaganda and hate speech. Te chapters by Anderson, Rini, and Cassam collectively illustrate a concern about citizens’ ability to access political facts. In the next chapter, Fabienne Peter investigates the possibility of political justifcation under conditions of substantial uncertainty. As Peter observes, political deliberation and decision-­making routinely occur in circumstances of substantial uncertainty about what should be done, where this uncertainty concerns both empirical facts and normative facts. But is uncertainty about empirical and normative facts symmetrical? While many argue that uncertainty about empirical facts should be taken into account for political justifcation, some have argued that normative uncertainty does not demand the same caution. Against this asymmetrical view, Peter argues that political justifcation must also take normative uncertainty into account. In the fnal chapter of this section, Briana Toole tackles a persistent threat to the egalitarian promise of democracy: white supremacy. She argues that white supremacy has remained a challenge to democratic institutions because it is epistemologically resilient. According to Toole, white supremacy is more than a tool of social and political oppression: it is an epistemological system that serves as the foundation for how we understand and interact with the world. Tis makes it especially resilient to dismantling because epistemological systems have the capacity to resist change to their underlying structure while also ofering the appearance of large-­scale reform. Her analysis of white supremacy as an epistemological framework allows us to better understand, and eventually overthrow, this oppressive system. Part II centers on epistemic challenges to democratic institutions. Democratic institutions are frequently defended on the grounds that they treat individuals fairly—whether this is because of democracy’s instrumental value in achieving justice or its intrinsic respect for citizens’ equality. Yet critics have asked whether democracy can withstand criticisms about the epistemic quality of the decisions made by democratic rule. Te chapters in this section focus on diferent aspects of the epistemic challenge to democracy—at times to defend democracy from its critics, at times to show just how worrisome the epistemic challenges should be for those who extoll democratic institutions. Tis section begins with David Estlund’s chapter questioning how weighty the requirement for democratic rule can be when challenged by alternatives that would make better political decisions. He draws an analogy between paternalism and epistocracy (political rule by the wise) to consider the limits to one’s right to self-­ governance. As Estlund points out, epistocracy and paternalism are analogous in at least two respects: merely knowing better is not enough to justify taking charge, but the prohibition on each is unlikely to be absolute. If one person’s competence is very low and the other person would do far better by taking over, then paternalism is plausibly justifed. Estlund asks whether our right against asymmetrical subjection is beholden to epistemic constraints based on a sufcient competence gap. While our

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4  Elizabeth Edenberg and Michael Hannon right against paternalism is limited by certain cases of incompetence, Estlund asks whether (and to what extent) such gaps in competence also impact our right to democracy. While Estlund seeks to clarify the epistemic challenge to democracy, Tomas Christiano defends the value of democracy against epistemic objections based on voter ignorance. He grants that voters are typically not very knowledge about politics, but he claims this provides no good grounds to endorse rule by experts. Christiano argues that democracy can retain its value in advancing citizens’ interests in an egalitarian way, even within contexts of low information decision-­making. However, he acknowledges that critics of democracy do identify an underdeveloped part of the foundations of political equality, which is the theory of citizen participation. He proposes a collaborative conception of how citizens participate in a democracy that can help show how citizens can act on good information and how democracy can be made more efective and egalitarian. Jason Brennan’s chapter draws on empirical political science literature about voter psychology and voter ignorance to critique the public reason project. According to the public reason view, moral or political rules must be justifable or acceptable to all those persons over whom the rules have authority. However, Brennan endorses the politically “realist” view that most citizens lack the kinds of beliefs and attitudes which public reason liberals believe are normatively signifcant. As a result, the public reason theorists’ understanding of reasonable disagreement between citizens is unsubstantiated by the empirical literature and, thus, their proposed solutions are of little value. Alexander Guerrero agrees with critics of democracy that elections introduce a host of problems for democracy, including epistemic problems. But rather than give up on the value of egalitarian rule by the people, he argues for a reconceptualization of democracy in favor of a lottocratic alternative to our usual electoral representative models. Afer identifying the epistemic pathologies of electoral representative institutions, Guerrero argues that lottocratic systems of government perform much better on epistemic terms. In particular, he argues that lottocracy is better able to embody the institutional epistemic virtue of “sensibility,” viz,. appreciating and responding to the world as it is. We should therefore prefer lottocracy over electoral representative models for epistemic reasons. In the fnal chapter of this section, Kristofer Ahlstrom-­Vij and Jennifer Steele empirically study public ignorance and what to do in light of the fact that most citizens have only a superfcial understanding of politically relevant matters. In particular, they look at attitudes towards immigration in the UK and whether or not policies can truly refect the “will of the people” when people lack sufcient knowledge about the relevant issues. Tis case study on immigration serves to highlight the ofen-­overlooked problem that policies implemented with reference to popular sentiments might not capture the will of the people. Part III of this book is about two deeply interrelated issues: political polarization and intractable disagreement in politics. Political debates are becoming increasingly polarized in Western-­style democracies. Citizens have highly unfavorable views of each other, ofen regarding their political opponents as immoral, stupid, lazy, and even threatening to each other’s way of life. Te chapters in this section outline

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Introduction  5 various types of polarization, identify the mechanisms that cause citizens to polarize, and suggest ways to mitigate partisan discord. Robert Talisse’s chapter identifes several distinct kinds of polarization and ofers a view of how their interaction gives rise to various political problems, ranging from legislative deadlock and partisan animosity to escalating extremism. Tis conceptual work is important: political commentary is currently saturated with narratives about the harms of polarization, but these commentaries fail to acknowledge the diferent types of polarization; as a result, they obscure from view the diferent causes of—and solutions to—these problems. Jeroen de Ridder also distinguishes multiple senses of “polarization” in order to provide a more precise account of what polarization is. By identifying the diferent ways in which citizens can polarize, both de Ridder and Talisse enable us to better understand the mechanisms of polarization and address the problems they pose for democracy. While de Ridder agrees with Talisse that polarization undermines democracy, de Ridder’s analysis centers on how deep disagreements erode the common ground needed to harness the epistemic power of democratic deliberation. In deep disagreements, parties disagree about relatively fundamental underlying moral or epistemic principles and therefore see each other as less than fully rational or morally subpar. de Ridder argues that deep disagreements lead to both cognitive and practical polarization, especially when they concern matters that are central to people’s social identities. Tis, in turn, entrenches their disagreement even further, resulting in a vicious feedback loop. Michael Lynch’s chapter explores two contributing factors to polarization. Te frst is epistemic disagreement—or disagreement over what is known, who knows it, or how we know. Te second is intellectual arrogance—or arrogance about what we know or think we know. Lynch argues that even the perception of widespread epistemic disagreement is dangerous. Moreover, he argues that epistemic disagreement and intellectual arrogance are mutually reinforcing. Tis makes them doubly dangerous. By increasing cognitive polarization, they undermine the democratic value of the pursuit of truth. In keeping with the chapters above, Elizabeth Edenberg argues that intractable disagreements threaten our ability to build a cohesive political community. Tis is exacerbated by social media, which allows us to sort ourselves into increasingly likeminded groups that consume information from diferent sources and end up in polarized and insular echo chambers. But in contrast with the previous chapters, Edenberg argues for the need to shif away from epistemic evaluations of our fellow citizens in order to build a political community based on mutual respect. Te breakdown of discourse online provides reasons to draw out not an epistemic but a moral basis for political cooperation among diverse citizens—one inspired by Rawlsian political liberalism. Jennifer Lackey explores when we should, from an epistemic point of view, disagree about politics by asking the general question: when do we have the epistemic duty to object to assertions that we take to be false or unwarranted? She characterizes this as an imperfect duty rather than a perfect duty, and she examines in detail one specifc account of imperfect moral duties before sketching her own view of the duty to object. She then applies her view to the political domain and highlights the ways

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

6  Elizabeth Edenberg and Michael Hannon in which pressing issues arise when we disagree about political matters. An interesting feature of Lackey’s view is its implication that we should ofen increase the level of disagreement with our political opponents. Tis starkly contrasts with the chapters by Talisse, de Ridder, Edenberg, and Lynch, all of whom worry about increasing levels of political division. Lackey suggests that objecting to political opponents may actually diminish the extent of political disagreement in the long run. In the fnal chapter, Michael Hannon argues that the extent and depth of political disagreement is largely overstated. While Hannon agrees with Talisse, de Ridder, Lynch, and Edenberg that voters are increasingly polarized in terms of their attitudes towards each other, he maintains there is comparatively little polarization on the issues. Te extent and depth of political disagreement is exaggerated by the fact that many people will deliberately misreport their beliefs as a way to express their attitudes. Hannon then draws out a number of signifcant implications from this idea. For example, he argues that insincere disagreement explains why debates ofen go so poorly, why people seem to hold blatantly contradictory views, and why it is ofen so difcult to correct false beliefs. Although this book is divided into three parts, the chapters in this volume speak to each other across a variety of other issues. For example, Talisse’s chapter focuses on polarization as a dominant explanation for epistemic bubbles, whereas Anderson questions this hypothesis and provides an alternative account that focuses on the epistemic norms of specifc groups. Like Anderson, both Toole and Lynch examine the epistemic norms of groups. Lynch looks at our epistemic attitudes we take towards each other, while Toole analyses the resilient epistemic system of white supremacy, which frames the conditions under which knowledge is possible. Lackey argues that we have an epistemic duty to object to our political opponents, while Brennan suggests that such objections are ofen the expression of tribal cheerleading. Rini and Edenberg both look at how social media have fragmented our epistemic landscape, with Rini diagnosing the core issue as an intentional weaponization of skepticism and Edenberg arguing (against both Rini and Lynch) that the epistemic lens is important but incomplete. Anderson and Hannon agree with Brennan that political disagreements are ofen insincere and merely refect identity-­expressive discourse. By contrast, de Ridder and Lynch worry that political disagreements have grown especially deep in polarized societies. However, these authors all agree that we cannot understand the nature, extent, and depth of political disagreement without better understanding how it is bound up with socio-­political identities. Christiano investigates whether political ignorance undermines democracy’s promise to advance the interests of citizens in an egalitarian way, whereas Ahlstrom-­Vij and Steele argue that political choices based on ignorance fail to refect what individuals really want. Peter picks up on the idea that uncertainty is signifcant for political legitimacy, arguing that political justifcation should take both normative and empirical uncertainty into account. Estlund, Christiano, Peter, and Guerrero all ofer diferent arguments for what political legitimacy requires to best accommodate democracy’s epistemic and egalitarian goals. Tis is just a small sampling of the ways in which the chapters in this volume have rich connections across diverse issues. Te volume also investigates the extent and implications of political ignorance (Ahlstrom-­Vij and Steele, Anderson, Cassam,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Introduction  7

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Christiano, Guerrero, Lynch, and Rini), democratic deliberation (Anderson, Brennan, de Ridder, Hannon, Lynch, Rini, and Talisse), the signifcance of epistemic considerations for political legitimacy (Brennan, Christian, Estlund, Guerrero, and Peter), the epistemology of political disagreement (Anderson, de Ridder, Edenberg, Hannon, Lackey, Lynch, and Talisse), and identity politics (Anderson, de Ridder, Hannon, Lynch, Toole, and Talisse), among other topics at the intersection of political philosophy and epistemology. Tis volume does not aspire to provide a comprehensive guide to the burgeoning feld of political epistemology, nor does it bring together rival answers to a single question or small cluster of questions. Te frst approach is more suitable for an introductory handbook of political epistemology, while the second is more appropriate for subfelds of philosophy that are already well defned and where there is an established set of research questions that focus the discussion. We view this volume as serving a somewhat diferent aim: to highlight some of the key topics, questions, and problems at the heart of political epistemology. A premise underlying the selection of themes and essays for this volume is that, beyond a certain point, progress on certain foundational issues in both political philosophy and epistemology cannot be achieved without attending to both the epistemic and political dimensions of these problems. By bringing political philosophers into conversation with epistemologists, this volume promotes more cross-­pollination of ideas while also highlighting the richness and diversity of this area of philosophy. Our hope is that through careful analysis of the philosophical issues underlying contemporary political challenges, we can identify constructive paths forward.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved. Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

PART I

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

T RUT H A N D KNOW LE D G E IN P OL ITICS

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved. Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

1

Epistemic Bubbles and Authoritarian Politics Elizabeth Anderson

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.  Epistemic Bubbles and their Treats to Democracy Contemporary American political discourse is distorted by “epistemic bubbles.” An epistemic bubble may be defned as a relatively self-­segregated social network of like-­minded people, which lacks internal dispositions to discredit false or unsupported factual claims in particular domains. Due to factors internal to the network, members are liable to converge on and resist correction of false, misleading, or unsupported claims circulated within it. In the U.S., the rise of politically consequential epistemic bubbles appears to be tied to increasing partisan polarization. As Talisse documents in this volume, Americans are divided by partisan identity not only with respect to political issues, but with respect to numerous otherwise politically irrelevant lifestyle choices. Partisan polarization is higher now than at any time since the end of the Civil War (Hare and Poole,  2014). Moreover, members of diferent parties disagree not only about values, but about facts. Partisan disagreement about certain politically salient factual claims—such as whether human activity is causing climate change, and whether carrying concealed weapons makes people safer—exceeds disagreement about values, such as the justice of progressive taxation (Kahan, 2016, p. 1). Factual disagreement need not indicate that partisans occupy distinct epistemic bubbles. It could be that members of rival parties began with diferent priors, and that evidence is insufcient to justify convergence on a consensus position. However, as the scientifc consensus that humans are changing the climate became more certain, and evidence of this was more widely publicized, partisan disagreement over this claim grew from a 17 percent diference between Democrats and Republicans in 2001 to a 41 percent diference in 2016 (Dunlap et al., 2016, p. 9). Democrats converged on the scientifc consensus, while Republicans departed from it. Tus, whether an epistemic bubble exists is indicated not by within-­group consensus and between-­group disagreement, nor by a group’s initial convergence on certain false, misleading, or unsupported beliefs, but by the failure of a group to update its beliefs in an accuracy-­directed response to new information. Politically consequential bubbles have formed around two types of claims. One concerns scientifc claims about risk. Millions of Americans deny that human activity is causing global warming. Many others believe that vaccines cause autism. Te second type concerns delegitimizing claims about government, public policy, politicians, and infuential people. Many hold that “the deep state” plotted against President Trump, that Bush was complicit in the 9/11 attacks, that some U.S. cities Elizabeth Anderson, Epistemic Bubbles and Authoritarian Politics. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Elizabeth Anderson. DOI: 10.1093/oso/9780192893338.003.0002

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

12  Elizabeth Anderson are run under Sharia law. Sometimes the two types merge—for example, in the belief that claims about the dangers of the novel coronavirus are a hoax to discredit President Trump. Epistemic bubbles threaten both sound policymaking and democracy itself. Groups inside bubbles are vulnerable to manipulation by charlatans, propagandists, and demagogues. Tis disables them from contributing productively to the epistemic functions of democracy, which involve collective learning from accurate information about problems and policies that is asymmetrically distributed across citizens (Anderson, 2006). Bubbles threaten the whole society when intransigent false beliefs concern serious problems we all face. Refusal to acknowledge genuine risks can be worse than disagreement about values. If partisans agree about the facts of anthropogenic climate change, but disagree about values, they can still reach compromises— for example, over whether market-­friendly, revenue-­neutral solutions would be preferable to state regulation. But if one side insists that the science of global warming is a communist hoax, it is likely to see obstruction of all action to mitigate climate change to be the only acceptable course—and to view those who disagree as subversives and traitors. Similarly, intransigent false beliefs that the state is involved in conspiracies against the people sow mistrust of democracy and of rival political parties. Such beliefs do not merely impair the intergroup discussion that lies at the core of possibilities for democratic cooperation. Tey underwrite a hostile exclusionary politics of “enemies within” that undermines democracy itself, by writing fellow citizens out of the social contract, and by licensing destructive attitudes toward democratic institutions. If epistemic bubbles are a problem, we need to fgure out how to burst them. Tis requires analysis of how they form and operate. I begin by considering two prominent models of how bubbles work: Cass Sunstein’s group polarization theory, and Dan Kahan’s cultural cognition theory. Both theorists are inclined to deny that partisan groups difer in their tendencies to form epistemic bubbles. Indeed, laboratory evidence indicates that individuals with diferent partisan and ideological identities do not difer on average with respect to relevant cognitive characteristics. However, evidence from the feld suggests that there is a partisan asymmetry. Republicans today are more vulnerable to epistemic bubbles than Democrats. Group asymmetries are possible even if individual cognitive characteristics do not difer between groups, due to features of the group itself. I argue that an important cause of partisan asymmetry in epistemic bubbles is due to the rise of populist politics within the Republican Party. I then ofer two models of how populist politics generates epistemic bubbles. Te collective cognition model claims that populist groups promulgate group norms for assimilating evidence that generate epistemic bubbles. Te discursive model claims that populist groups adopt rhetorical styles that function similarly to ep­i­ste­ mic bubbles, in suspending accuracy-­guided empirical inquiry with them. Both models suggest that social epistemology needs to get more political, by considering the impact of populist political styles on what people assert and believe. Tey suggest that social epistemology needs to get more social, by locating critical features of ep­i­ ste­mic bubbles outside people’s heads, in the norms by which certain groups operate. I conclude with a discussion of how each model points to potential strategies for bursting epistemic bubbles.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistemic Bubbles and Authoritarian Politics  13

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Two Models of Epistemic Bubbles Cass Sunstein’s (Sunstein,  2009; Sunstein and Vermeule,  2009) group polarization theory ofers one prominent model of epistemic bubbles. Begin with a relatively enclosed social network of relatively like-­minded individuals—people who have similar or skewed values and prior credences on particular types of claims in some domain. Tree sets of universal cognitive biases make that group vulnerable to the spread of false and extreme beliefs within the network, and make it resistant to correction, if the falsehoods ft the group’s priors. Te frst set of biases propels information and reputation cascades. Individuals ofen rely on a popularity heuristic: pay attention to those to whom many others in the network are listening. Claims within a network can thus go viral. A conformity heuristic propels the propagation of belief: if some people in the network express belief in a claim, others are liable to defer, until it is so popular that most agree. Even private dissenters may express agreement to preserve their reputation within the group. A second set of biases leads to group polarization—tendencies to accept extreme beliefs within the group. Expressions of agreement with the belief increase members’ confdence in it, and may also exaggerate its content. For example, among Republicans, evidence that Hillary Clinton was negligent in managing some classifed State Department emails turned into certainty that her behavior was criminal. Claims that pander to the group’s fears and wishes intensify those emotions. Tis motivates group members to commit themselves to more extreme actions related to the claims (“Lock her up!”) (Sunstein, 2009, p. 35). A third set of assimilation biases leads members to resist correction of popular beliefs. Confrmation bias leads people to seek and believe new evidence that confrms their belief. Disconfrmation bias leads people to discredit or rebut evidence that disconfrms their belief. Attempts to refute false beliefs may be self-­defeating: repetition of a claim for the purpose of negating it increases people’s familiarity with it, and people give greater credence to familiar thoughts. People also tend to distrust outgroup sources that attempt to correct beliefs held within the group (Berinsky, 2015, pp. 246, 245). Sunstein’s theory ofers a plausible account of several mechanisms that likely underlie epistemic bubbles. Yet his account lacks specifcity. Because it is content-­ neutral, it over-­predicts the tendency of groups to converge on false beliefs, if they begin with a signifcant prior bias in favor of them. Yet epistemic bubbles appear to form primarily around false or unsupported claims that appeal to specifc values or motives peculiar to the group. Dan Kahan’s cultural cognition theory explains why epistemic bubbles form around claims about risk. Cultural cognition refers to a set of cognitive biases tied to the ways the perception of risk is afected by individuals’ worldviews and cultural identities. Building on Mary Douglas and Aaron Wildavsky’s (1982) “cultural theory of risk,” Kahan (2012) postulates that individuals form perceptions of the risk of activities depending on their perceived threat to or congruence with their ideals of social order, which defne group identities. Individuals view those who challenge the cultural dominance of their worldview as threats not only to their preferred social

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

14  Elizabeth Anderson order, but to their cultural identities and to their social status within their cultural group and relative to those holding rival worldviews. Cultural cognition theory classifes individuals’ worldviews along two axes. Te hierarchical–egalitarian axis sorts individuals on a continuum from those who prefer society to be hierarchically organized to those who prefer social equality. Te individualism–communitarianism axis sorts individuals on a continuum from those who prefer that society be based on market competition and self-­reliance to those who prefer collective action to regulate conduct and provide for individual needs. Te widest gaps in risk perception typically lie between hierarchical individualists and egalitarian communitarians. For example, on a four point scale, hierarchical individualists rate the risk of gun ownership to public safety at 2.26, while egalitarian communitarians rate it at 2.95, with the other two groups in-­between (Kahan et al., 2007, p. 481). Hierarchical individualists view gun ownership as enabling them to practice the virtue of self-­reliance in defending themselves against criminals. It upholds their authoritative status as protectors of their families. (Hierarchical individualists are disproportionately white men.) Egalitarian communitarians are more likely to fear that gun owners threaten them, and prefer public safety to be collectively secured by police. Cultural cognition theory ofers empirically well-­supported and compelling explanations for group diferences in risk perception. It is hard for people to think that an activity they view as virtuous and honorable is dangerous. Hence, hierarchical individualists discount the dangers of gun ownership, infate the dangers of gun regulation, and exhibit biased assimilation of information about gun risks—favoring information that highlights successful uses of guns in self-­defense, and discounting information about gun accidents and misuse. Tey give credence to experts who claim that gun ownership reduces crime, and distrust those who claim that the risk of being killed by a gun rises if one lives in a household with guns. Egalitarian communitarians exhibit biases in the other direction. Cultural cognition theory highlights two identity-­protective biases. One inclines people to conform to the beliefs dominant within their group, to express their commitment to and maintain their standing within the group. To concede the danger of what the group regards as virtuous risks that standing by marking oneself as a threat to the group’s status. By contrast, individuals have no motive to assimilate information about the risk in an accuracy-­enhancing way, because an individual’s personal opinion of objective risk has no impact on its incidence or on policies to address it. Hence cultural cognition is “perfectly rational” for the individual, because it enhances the individual’s welfare (Kahan, 2016, p. 13). However, it is collectively irrational in preventing society from arriving at an accurate consensus on risks and from adopting efective policies to address them. Te second identity-­protective bias concerns how members of identity groups relate to advocates of rival worldviews. Each group is disposed to see the others as threatening to their social standing. Tis disposition is activated by culturally antagonistic memes: ways of framing the stakes or the weight of the evidence in discussions of risk that stigmatize rival identity groups as vicious or stupid (Kahan et al., 2017, p. 5). Public discussion of risks degenerates into expressions of mutual contempt, in a competition over the relative standing of the groups. Tis explains why disagreements

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistemic Bubbles and Authoritarian Politics  15 over certain risks, such as guns and climate change, are full of animus and outrage toward the other side. Cultural cognition theory does a better job than group polarization theory in explaining why particular factual disagreements are so politically toxic. Sunstein’s theory predicts group polarization entirely through ingroup processes: each group is separately driven to extreme beliefs on opposite sides of a particular claim. But unless the content of those beliefs is about the members of the other group, it is not evident why disagreement between groups would generate intergroup animus. All disagreements would be like that between believers and skeptics over the efcacy of homeopathy. Cultural cognition theory explains the toxicity of such disagreements in terms of hostile intergroup processes.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Partisan Asymmetry in Vulnerability to Epistemic Bubbles Group polarization theory and cultural cognition theory explain how epistemic bubbles arise and persist on the basis of universal cognitive biases. If individual variations in degrees of bias are randomly distributed across groups, then both theories would tend to predict symmetry across partisan and ideological groups in susceptibility to epistemic bubbles. Group polarization theory predicts that all networks would tend toward extreme beliefs if they begin with a signifcant skew in one direction. Over time, the Democratic and Republican parties have each become more ideologically homogeneous (Jacobson, 2013, p. 691). Te only thing stopping a runaway train to extreme partisan polarization would be the openness of the group to outside information. However, increased geographical sorting of party members (Bishop and Cushing, 2008) as well as the rise of separate news and social media tailored to distinct ideological audiences (Sunstein, 2007) has likely increased the communicative insularity of both sides. Sunstein is even-­handed in his treatment of conservatives and liberals, Republicans and Democrats. His theory thus appears to support ep­i­ste­ mic symmetry of mainstream groups. In the absence of intergroup rivalry, cultural cognition theory does not expect that relatively insular groups will tend to converge on a common ingroup understanding that the meaning of some activity or policy is threatening to its cultural commitments. For central to cultural cognition theory is the thought that the cultural meaning of any given risk or policy is malleable. Tere is no necessary connection between taking action on climate change and threatening the social identities and status of hierarchical individualists, who are committed to big business and free markets. Tey can promote climate mitigation policies that afrm the identities of hierarchical individualists, such as nuclear power and geoengineering. Once solutions are framed to remove threats to their identities, hierarchical individualists will more likely acknowledge climate change (Kahan, 2010, p. 297). In their laboratory experiments, cultural cognition theorists have not found signifcant group asymmetries in biased assimilation of new evidence (Kahan,  2016, p. 14). If group polarization is driven by culturally antagonistic memes, this is as the theory predicts. Both sides of a rivalry will dig in their heels over claims in which their

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

16  Elizabeth Anderson groups’ relative esteem is at stake. Because both sides form their beliefs for status-­ based reasons, it is only an accident which side’s beliefs correspond to what is warranted by the publicly available evidence. Other researchers confrm the view that ideological groups do not difer in their tendency to believe claims based on ideological afnity. A recent meta-­analysis of 51 experimental studies found no signifcant diference between liberals and conservatives in partisan bias (Ditto et al., 2019). Other investigators argue that there are substantial ideological group asymmetries in cognitive bias. Chris Mooney (2012) claims that Republicans and conservatives are more inclined to deny scientifc claims than Democrats and liberals, because the latter are more curious and open to experience than the former. John Jost (2017), reviewing hundreds of empirical studies, argues that Republicans, conservatives, and right-­wing authoritarians are more dogmatic and prone to various cognitive biases than Democrats and liberals. Nevertheless, he grants that assimilation biases are widespread across the ideological spectrum and party divisions (Jost,  2017, pp. 170–1). Since assimilation biases lie at the core of epistemic bubbles, should we believe, as Kahan (2015b) does, that whatever group diferences in cognition exist have negligible practical signifcance? Field evidence of striking partisan diferences bears on this question. Partisan polarization has largely been driven by Republican extremism. About 80 per cent of the increase in partisan polarization in Congress from 1973–2012 was due to Republicans moving to the right (Jacobson, 2013, pp. 690–1). Rank and fle Republicans are also more ideologically extreme and homogeneous, with 73 percent identifying as con­serv­ a­tive, 22 percent moderate, and 4 percent liberal, compared to Democrats, who identify as 51 percent liberal, 34 percent moderate, and 13 percent conservative (Saad). Two distinguished political scientists argue that “[t]oday’s Republican Party . . . is an insurgent outlier. It has become ideologically extreme . . . unpersuaded by conventional understanding of facts, evidence, and science; and dismissive of the legitimacy of its political opposition” (Mann and Ornstein, 2016, pp. Kindle loc. 1727–9). Teir observation about delegitimization of the opposition is critical, for epistemic bubbles about claims of illegitimacy serve this function. To delegitimize Democrats, Republican leaders and spokespersons have promoted and given credence to numerous false and extreme rumors about Democratic leaders and policies: that Obama was not born in the U.S., that he is a Muslim jihadist and founded ISIS, that millions of fraudulent votes were cast for Hillary Clinton in the 2016 election, that, as Secretary of State, she issued a stand-­down order to stop U.S. troops from protecting Americans in Benghazi . . . the list goes on and on. Democratic leaders do nothing comparable in scale or extremity to delegitimize Republicans. While members of both parties select news sources that tend to agree with their priors (Iyengar and Hahn, 2009), the media ecosystems of lef and right are highly asymmetric. In the 2016 election, Trump supporters drew most of their news from overtly partisan, right-­wing news sources such as Fox and Breitbart, which dominated inlinks and social media. Democratic news sources were far more balanced, from the center and center-­lef, dominated by “long-­standing media organizations steeped in the traditions and practices of objective journalism” such as the New York Times, Washington Post, and CNN (Faris et al., 2017, pp. 5, 8, 10).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistemic Bubbles and Authoritarian Politics  17 Te 2016 campaign featured dramatic asymmetries in fake news. One study found that those who identify as conservative shared fake news much more frequently than self-­identifed liberals, who rarely did so. Tese researchers caution that the ideological asymmetry they found may be due to the vastly greater prevalence of pro-­Trump fake news on the web (Guess et al., 2019, pp. 2–3). However, diferences in the prevalence of pro-­Trump and pro-­Clinton fake news appears to have been demand-­driven. Pro-­Clinton fake news sites folded, while pro-­Trump fake news sites went viral (Higgins et al., 2016). Hunt Allcott and Matthew Gentzkow (2017, p. 230) found in a laboratory study that both Democrats and Republicans were more likely to believe fake news articles catering to their ideologies, and that those with more insular networks were more credulous of ideologically aligned fake news. However, their feld data found 30.3 million Facebook shares of pro-­Trump fake news stories, as opposed to 7.6 million shares of pro-­Clinton fake news stories (Allcott and Gentzkow, 2017, p. 223). Furthermore, Trump supporters appear to be less likely than Clinton supporters to distinguish fake from real news (Pennycook and Rand, 2018, p. 9). More importantly for our purposes, conservatives tend to seek their news in a “distinct and insulated media system, using social media . . . to transmit a hyper-­ partisan perspective to the world.” Te most widely shared stories are

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

disinformation: the purposeful construction of true or partly true bits of information into a message that is, at its core, misleading. Over the course of the election, this turned the right-­wing media system into an internally coherent, relatively insulated knowledge community, reinforcing the shared worldview of readers and shielding them from journalism that challenged it. (Benkler et al., 2017)

Extremism of opinions and sources, network insularity, and credulously accepted fake and misleading news are hallmarks of a group in an epistemic bubble. Tus, feld evidence suggests a substantial partisan asymmetry in entrapment. How can this be reconciled with laboratory data that suggests cognitive symmetry of individuals across ideological lines? Te key to reconciling these two sources of evidence is to locate vulnerability to epistemic bubbles in distinctive features of the groups, rather than in diferences in cognitive disposition among the individuals belonging to diferent groups. Start with the fact that the Republican party is more homogenous than the Democratic party, not only by ideology, as noted above, but by race/ethnicity and religion. In 2016, Republicans were 86 percent non-­Hispanic white, 6 percent Hispanic, 2 percent black, 1 percent Asian, and 4 percent other. Democrats were 57 percent non-­Hispanic white, 12 percent Hispanic, 21 percent black, 3 percent Asian, and 5 percent other. Republicans were 83 percent Christian, 12 percent unafliated, and 4 percent some other religion. Democrats were 59 percent Christian, 29 percent unafliated, and 11 percent some other religion (Pew Research Center,  2016). Race, ethnicity, and religion powerfully shape Americans’ personal experiences, opportunities, and perspectives on social and political issues. Because the information relevant to understanding social and political problems is asymmetrically distributed by social identity (Anderson, 2006), groups that contain

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

18  Elizabeth Anderson

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

less of this diversity are more likely to generate distorted, inaccurate views of social and political issues. Hence, the Republican party is already in an epistemically more vulnerable position. Te critical feature of an epistemic bubble, however, is not whether a group tends to initially converge on mistaken beliefs, but whether it resists correction of its beliefs in light of new evidence. To explain resistance to correction, our social epistemology of epistemic bubbles needs to get even more social. Cultural cognition and group polarization theories focus on cognitive biases, understood as processes instantiated inside individual heads. I shall argue that certain cognitive biases can be instantiated at the group level, by way of group-­specifc epistemic practices and norms that promote epistemic bubbles, and discursive practices that function similarly, to block group members from engaging in accuracy-­guided epistemic practices. Tis can take place even if group members do not difer, on average, in their cognitive dispositions. Some groups’ leaders and media sources may drive opinion within their group by deliberately activating cognitive biases that lie latent in group members; more frequently injecting false, unsupported, and misleading ideas into their group’s networks; encouraging network closure to outside sources; reducing the epistemic diversity of the group’s membership; promulgating biased social norms of information processing; and promoting discursive norms that disrupt accuracy-­oriented discussion. Tese practices promote the formation of epistemic bubbles. Furthermore, to explain politically consequential epistemic bubbles today, our social epistemology needs to get more political. I shall argue that central to understanding the recent asymmetric proliferation of politically consequential epistemic bubbles is the rise of populism within the Republican Party. Te bubble-­promoting epistemic and discursive practices mentioned in the previous paragraph are constitutive of populism. Populist parties are inherently vulnerable to epistemic bubbles because populist leaders mobilize support by deliberately generating epistemic bubbles around claims of illegitimacy.

4.  What is Populism? Populism is a style of politics in which populist leaders or parties claim to exclusively represent “the people” against “elites.” Te people are depicted as virtuous; elites are corrupt. Populism amounts to an “exclusionary form of identity politics,” in which the people are always a subset of the citizens, the “real” people, the only ones who should count. It is illiberal and authoritarian, in that it rejects the legitimacy of opposition, criticism, and institutions designed to protect minority rights, all of which it construes as “enemies of the people” (Müller, 2016, p. 3). Populism is a way of viewing the signifcance of political contestation which can be utilized by demagogues on the right or the lef. Lef-­wing populist leaders and parties, such as Nicolás Maduro of Venezuela, or Greece’s Syriza Party, typically advocate socialism. Tey draw the people/elite line between the working classes and the rich. In Europe and the U.S., recent populist politics is overwhelmingly right-­wing. European populism includes such politicians as Viktor Orbán and Geert Wilders, and parties such as AFD (Germany), and Law and Justice (Poland).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Epistemic Bubbles and Authoritarian Politics  19 In the U.S., the Republican Party has included such right-­wing populists as Newt Gingrich, Sarah Palin, Donald Trump, and members of the Tea Party. Right-­wing populism frames issues in terms of three social divisions: the people, elites, and minority groups that elites have purportedly unjustly favored over the people (Müller,  2014, p. 23; Mudde and Kaltwasser,  2017, pp. 23–5). It defnes the true people in ethno-­racial and/or religious terms, who are middle and working class with traditional values, common manners, and common sense. Te purportedly unjustly favored groups typically include immigrants, foreigners, racial or religious minorities, the poor, and people with nontraditional lifestyles and values, such as feminists and LGBTQ people. Right-­wing populists win elections by mobilizing voters through a narrative of the following form (details for Trump’s version in parentheses). “Te people”—always a subset of the citizens (white, Christian)—are being taken advantage of, humiliated, and threatened by enemies both foreign (China, Mexico, “radical Muslim terrorists”) and domestic (immigrants, Muslims, blacks, the poor, transgender people), with the help of a corrupt elite (liberals, bankers, judges, civil servants, and knowledge workers, especially academics, scientists, and journalists) that has betrayed the people in the name of cosmopolitan, universal values. Te people can be saved only by putting their faith in a leader, who uniquely embodies their aspirations, sympathizes with their fate, has plans and powers to defeat their enemies, and who will act to raise (or restore) the people to their (former) glory. Tese plans involve protecting the people from foreign and domestic enemies by treating as presumptively threatening entire minority groups, and by putting the nation frst in transactions with other nations. In domestic policy this narrative is used to justify racism, immigration restrictions, mass incarceration, escalation of the national security state, torture, arbitrary seizure, police brutality, and other violations of human rights and the rule of law. In foreign policy it is used to justify trade protection, and withdrawal from international treaties and organizations that promote cosmopolitan values (human rights, global economic growth, international cooperation). Populist leaders reject the expertise of civil servants, normal bureaucratic processes, legal constraints, and consultation to check on the consequences of their policies. Populists imagine that the (“real”) people are homogeneous, with a unifed will that the leader implements (Müller, 2016, p. 3). Since citizens in fact disagree with each other, this unitary notion requires an exclusionary defnition of the people. Populism is essentially opposed to pluralism—the idea that society is composed of multiple cross-­cutting interest, ideological, and identity groups, and that the job of democratic institutions is to construct, by means of intergroup discussion and compromise, policies that ofer something to as many as possible (Mudde and Kaltwasser, 2017, pp. 7). Te Democratic Party, which prominently includes and celebrates its racial and religious diversity, and its embrace of LBGTQ people and immigrants, is a pluralist party. Populism is also essentially opposed to (liberal) democracy, understood as a system of government by discussion among all citizens, in which freedom of speech and minority rights are guaranteed by constitutional limits on government, and institutions such as an independent judiciary check legislation that violates individual rights (Müller,  2016, pp. 54–6). Instead, populists invoke Rousseauian notions of

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

20  Elizabeth Anderson popular sovereignty, according to which the will of the people is always right. Hence, once the results of a vote are announced, the only proper response of dissenters is to  admit error (Rousseau,  1983, pp. II.3, IV.2). Tis view rejects the legitimacy of ­opposition. Hence, populist politicians attack the independent press, intellectuals, scientists—anyone who claims expertise and advances considerations against populist policies. Anyone who opposes the will of the people is an enemy of the people, corrupt, and not entitled to speak out or claim rights. In ofce, populists reject institutional constraints on their power, such as constitutional rights, the rule of law, and an independent judiciary. Trump’s frequent attacks on the mainstream media as purveying “fake news,” and on the judiciary for being “unfair” in adjudicating civil suits against him and his policies, refect this view. Populists presuppose the Schmittian idea (Schmitt, 2007, [1932], Sec. 2) that the people can never be wrong in defning who they are—and that, in defning themselves, they also defne who their enemies are. Populist politics needs enemies to mobilize its supporters. For populists, as for Schmitt, politics is essentially a matter of “us” versus “them.” Populist politics has important epistemological implications. I propose two ways to think about the social epistemology of populism. Te frst argues that populist politics invokes epistemic norms that tend to trap followers in epistemic bubbles. Te second argues that populist rhetoric enacts the functional equivalent of ep­i­ste­ mic bubbles in public discourse, by replacing empirical discussion with trolling, insults, and conspiracy theories involving claims of illegitimacy.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

5.  Te Social Epistemology of Authoritarian Populist Politics: A Group Cognition Model Te group cognition approach to epistemic bubbles models cognition as taking place among people who share social norms of joint information processing. Almost any mental state that an individual can have can also be realized by a group. Tis follows from the fact that most activities that someone can do alone can also be done together, in coordination with others. Since activities are intentional, groups can have intentions. (Te Red Sox aim to win the World Series.) Since having an intention involves beliefs and desires, groups can have these mental states as well. A sufcient condition for a group to have a mental state is for the group members to be jointly committed to expressing that state together, in coordinated fashion. Each member (or enough to enable the group to advance its intentions) accepts the normative force of others’ expectations that they do their part in expressing that mental state, on condition that enough others do the same. Tis is common knowledge among members. Margaret Gilbert (1989) pioneered this “classical account” of group mental states, which shall be used in this chapter. (Similar outcomes could be generated by models in which group mental states can be realized under weaker conditions (Michael, 2011; Pacherie, 2011).) Consider collective belief (Gilbert, 1987). A group believes P if it accepts norms like the following. P may be introduced as an uncontested frst premise in the group’s reasoning about what (more) to believe or do together. Others will dispute someone’s

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Epistemic Bubbles and Authoritarian Politics  21 assertion of any claim Q that clearly presupposes ~P.  Te group accepts norms against others’ doubting or disputing P without having frst made the case for (re-) opening consideration of its truth. Someone who insists on continuing to dispute P may be marginalized, censured, or even expelled from the group. A group may believe P, although some or all members have private reservations about it, as long as members follow norms of joint belief. A frm may have adopted a risky business strategy based on earnings projections that some executives doubt. But if they held their tongues about this at the strategy meeting, they have lost their chance to dispute the projections, and are expected to promote the strategy or risk being fred. Hence, when A asserts P in an ofcial capacity, it cannot be inferred that A personally believes P. A may simply be performing A’s expected role in the group. No one supposes that spokespersons for corporations personally believe everything they say. Teir role is to express the ofcial views of the corporation. Tis account of group belief can be extended to less tightly organized groups, such as political parties and social identity groups, so long as members follow norms of group belief. Tis may help explain phenomena such as context-­specifc white ignorance of racism against people of color. Consider a revealing experiment by Samuel Sommers (2007). Sommers recorded the deliberations of 29 mock juries who were shown a video of a trial of a black defendant on trial for sexual assault. Half the juries were all-­white; the other half included both whites and blacks. Not surprisingly, the possibility that racism afected the process was raised more frequently in the racially diverse juries than in the all-­white juries. However, black jurors did not wholly account for this diference. White jurors in racially diverse juries were more likely than black jurors to raise the issue, although whites in all-­white juries were the least likely to raise it (Sommers, 2007, pp. 605–6). Sommers also found that every single time whites mentioned racism in all-­white juries, other jurors called them out for raising the issue, claiming that this was irrelevant to the task at hand (Sommers, 2007, p. 607). In racially diverse juries, both whites and blacks seriously considered racism as a possible factor in the trial. Group belief explains this pattern better than individual belief. Jurors in all-­white juries were enforcing a norm of white ignorance about racism, a group-­level lack of belief—a joint determination to act, as whites, in ignorance of racism. Tat whites in racially diverse juries initiated and participated in discussions of racism displays their implicit awareness of contrary norms of group ignorance and awareness in all-­ white versus racially diverse groups. Tus, racially diverse groups know some things that all-­white groups do not, and may do so even if their members do not difer, on average, in their knowledge or cognitive skills. Let us apply this model of group cognition to the social epistemology of populist groups. Such groups explicitly promote norms for information processing that generate epistemic bubbles. Populist politicians deploy narratives of elite failure and betrayal. Hence, elite leaders and institutions cannot be trusted. Populism casts the opposition as illegitimate. Tis sows mistrust of those who do not belong to the populist party or movement. Te “us” versus “them” frame essential to populism thus encourages tight network closure around just those who support the populist line. Due to their nativist tendencies, right-­wing populist parties are more homogeneous by racial, ethnic, religious, or other salient social identities. Hence, they

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

22  Elizabeth Anderson contain less epistemic diversity—lower diversity of information drawn from personal experiences of occupying diferent social identities—than pluralist parties. Members of right-­wing populist parties are mobilized by a shared grievance about their relative position in society—the complaint that elites have unfairly put other groups ahead of them. Tis is a powerful group motive for culturally afected ignorance about the injustices sufered by those other groups, who typically are less advantaged than the populists in the middle. Observers have frequently noted that populist policies are ofen simplistic, ignoring questions of feasibility, efcacy, and unintended consequences. Tis refects an epistemology that favors the “common sense” of the people over the corrupt sophistication of elite knowledge-­producers, including scientists, academics, and policy experts, who rely on empirical evidence (Mudde and Kaltwasser, 2017, pp. 68). Tus Trump advisor Stephen Miller cited “common sense” in dismissing a reporter’s request for data to support his claim that drastically cutting legal immigration would raise Americans’ wages—a view rejected by most economic studies (Bump, 2017). Because the opposition is illegitimate, criticism of populist parties, leaders, and policies from the independent media must be false. Tose who have dedicated their careers to careful empirical investigation are written out of the populist domain of trusted knowers. No wonder Trump has denounced the mainstream media as “the enemy of the American people,” while 80 percent of Republicans today agree that the mainstream media publishes a lot of fake news (Grynbaum,  2017; Easley,  2017). Most Republicans also hold that colleges and universities have a negative impact on society (Pew Research Center, 2017, p. 3). Te group cognition model explains why populists may rely on experts in some contexts and reject it in others. Most populists still trust their personal physician. It is only when what they say is politically salient—when, for example, doctors publicly question the unscientifc advice of populist leaders in a pandemic—that they follow the group norm to play their partisan role in dismissing doctors’ claims to protect the standing of their leader and his cause. Tis role-­based understanding of group belief makes better sense of the purported rationality of cultural cognition than Kahan’s account. Kahan argues that individual belief in the party line can be “perfectly rational” due to reputational incentives, even if the belief is not accuracy-­directed. However, a person can judge her mental state to be rationally formed only if she can endorse it with awareness of her reasons for adopting it. “If I don’t believe that climate change is a hoax, my social standing among fellow group members will be damaged” is not reasoning that, if consciously entertained by a subject, could rationally persuade him that it really is a hoax. Motivated reasoning requires some self-­deception to yield confdent belief. By contrast, performance of one’s role in a group requires no self-­deception. No group has assumptions that precisely mirror every member’s views. But everyone understands that the group’s chances of success improve if its members speak from common talking points. People can rationally do so without needing external incentives, such as worries about reputation. Simple identifcation with the group—commitment to acting jointly with fellow group members to realize the group’s aims—is enough. What is individually rational in this context is asserting certain views, or at least going along with them, not believing them full stop.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistemic Bubbles and Authoritarian Politics  23 Tis rationale for performing one’s role in group projects does not grant a license to function as spokespersons for epistemically irrational groups. Tese include groups with norms of information processing that tend to trap it in epistemic bubbles. Mobilizing partisans by spreading conspiracy theories and simplistically appealing promises may be a path to power. But this is a formula not for successful policymaking, but for ruin.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

6.  Te Social Epistemology of Authoritarian Populist Politics: A Discursive Model Te group cognition model of epistemic bubbles distinguishes norms of assertion from norms for individual belief. Once assertion is detached from individual belief, the possibility is opened that assertions may play other functions for the group besides expressing group beliefs. Some group-­supported speech acts that take the form of empirical assertions may mean something diferent from what they literally say. Te discursive model of epistemic bubbles develops this possibility by focusing on “identity-­expressive” speech acts. Identity-­expressive discourse expresses the speaker’s group identity, and positions the speaker in relation to people with the same or other identities. It may signal whose side one is on, who is the enemy, or doesn’t belong, who is illegitimate, who is superior to whom. Tere is nothing inherently wrong with such discourse, as long as everyone understands its function. Suppose, however, a group adopts discursive norms to treat certain ostensibly empirical assertions as identity-­expressive. It thereby removes those utterances from empirical inquiry, in which assertibility is governed by evidence. A group following such norms may thus create the functional equivalent of an epistemic bubble, by blocking accuracy-­guided empirical inquiry within the group with respect to these assertions. Te speech acts these norms license, while appearing to make empirical claims, function to replace empirical assertion with identity-­expressive discourse. To illustrate, consider claims commonly made on the right about Obama—that he was not born in the U.S., that he is a Muslim “founder” of ISIS, and a former crack-­ addicted gay prostitute, and a follower of America-­ hating Christian preacher Jeremiah Wright, and a Communist—although he is also hated for bailing out the Wall Street banks. Contradictory attributes pile on, with no attempt to reconcile them with each other or with the evidence. Interpreted as asserting empirical claims, such speech acts are absurd and unhinged. Interpreted as identity-­expressive, however, they simply express in varying ways that Obama is not “one of us,” not a “real” American, but an enemy. Uttering them is a way for right-­wing populists to afrm their own identities. When a group insists on treating empirical counterclaims as merely the identity-­expressive discourse of a rival group, it will dismiss them as “fake news,” insults, or other types of speech act not entitled to consideration as evidence. Populist discourse is able to hijack ostensibly empirical assertions for identity-­ expressive aims because it rejects pluralist, liberal democratic politics, which depends on empirical inquiry. Liberal democratic politics understands the will of the people as contingent and shifing, as politicians assemble diferent, temporary majorities to win elections. Hence, the will of the people is empirically determined by objective

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

24  Elizabeth Anderson methods such as counting the votes. Defeated parties are expected to concede and function as a loyal opposition. Pluralist party policymaking aims at delivering benefts for diverse groups who are diferentially afected by policies. Tere is no way to know the diverse consequences of policies without empirical research. Hence, liberal democracies need accuracy-­oriented experts to help design policies and liberal democratic discourse needs to be empirically grounded. Populism rejects these assumptions. For populists, the (real) people are unifed and speak with one voice—the voice of the populist leader or party. Te opposition is disloyal and illegitimate. Hence, when an election delivers populist defeat, it must have been fraudulent: votes were cast by enemies of the people, who were not morally entitled to vote. But a victorious election is always legitimate, regardless of the means—such as suppression of opposition votes or collusion with foreign powers— used to obtain victory (Müller, 2014, pp. 31–2, 56). Populist policymaking is not fundamentally about delivering empirically determinable consequences. Its deliverables are symbolic and moral. Populist politics is essentially about expressing a grievance: elites have betrayed the people; the people must be vindicated by delegitimizing elites—and, in the right-­wing version, by restoring their rightful position above the outgroups whom elites have unjustly placed ahead of them. Tese goals can be achieved symbolically, by maligning the independent media, attacking the impartiality of the independent judiciary, shutting civil servants and policy experts out of the policymaking process, violating ethical and institutional norms of government, and putting despised minorities in their place. Much populist rhetoric should therefore be understood as identity-­expressive symbolic positioning, rather than demands made on the basis of empirical fndings. “Build the wall” is not a practical way to stop illegal immigration, but rather an af­fi rm­a­tion of who the (real) people are and recognition of who they must be protected from. “Lock her up!” is not based on a legal argument, but a delegitimizing move. “Voter fraud” is not evidence-­based, but a grievance against fellow citizens whom populists think do not deserve to vote. Kahan’s notion of “culturally antagonistic memes” ofers a paradigm of identity-­ expressive discourse. Yet in explaining such discourse in terms of cognitive biases, Kahan confates beliefs with symbolic expression (Kahan et al., 2017, p. 3). Identity-­ expressive discourse should be modeled instead as a game of insults and provocations. When one person utters “yo’ momma . . .” to another, he is not sincerely asserting an empirical claim about the interlocutor’s mother. Rather, getting away with the utterance when the other lacks a witty riposte is a way of putting down the interlocutor. Culturally antagonistic memes function similarly, as vehicles of symbolic positional competition between rival identity groups. Sometimes discourse is obviously identity-­ expressive. Brian Schafner and Samantha Luks (2017) showed Trump voters unlabeled photos of Trump’s and Obama’s inaugural crowds and asked which showed more people. Fifeen percent pointed to the half-­empty Trump inaugural photo. Schafner and Luks acknowledge the absurdity of interpreting this response as expressing these voters’ beliefs. Rather, it was a way of showing those smug liberal academics that Trump voters will stand their ground in repudiating insults toward their group. John Bullock and colleagues

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Epistemic Bubbles and Authoritarian Politics  25 (2015) fnd that, when ofered monetary incentives, partisans give more accurate responses to politically charged questions. Tey conclude that some partisans know the truth, but prefer partisan “cheerleading” if they have nothing to gain from accuracy. Kabir Khanna and Gaurav Sood (2017), fnding similar efects of incentives, argue that what looks like motivated reasoning may really be “motivated responding”—that is, identity-­expressive discourse. Purely identity-­expressive political discourse does not manifest entrapment in epistemic bubbles. Tere is nothing wrong with partisan cheerleading when people express themselves using nonempirical language. Tere is nothing inherently problematic about waving the symbols of one’s party, singing its songs, shouting its slogans. Today’s problems with identity-­expressive discourse are those inherent in populist political discourse. Tese are multiple and grave. First, populist political discourse hijacks the grammar of assertion for expressive purposes, overtaking spaces properly reserved for empirical policy discussion. Tis is its point. It is why trolling has become a normative mode of political discourse on the right. Empirical policy discussion is what pluralist elites and experts do, not what the volk do, whose common sense as interpreted by their leader is sufcient to guide action. Populist parties and leaders will invariably lose the language game of evidence-­responsive, consequence-­sensitive policy discourse. So they switch to a different language game with symbolic stakes. Since such discourse does not express beliefs, it does not create an epistemic bubble. Yet it creates the functional equivalent of a bubble, by blocking the group’s engagement with evidence in an accuracy-­guided way, along with engagement in cooperative empirical inquiry with outsiders. Second, the symbolic stakes of populist discourse are essentially divisive. Its aim is to brand elites and despised minority groups as enemies of the people, not entitled to have their views heard or their interests respected. Culturally antagonistic memes are essential to populism. It aims to replace the pluralist liberal democratic politics of compromise and positive-­sum games with a zero- or negative-­sum game of symbolic positional competition. Tat this amounts to a spiteful and nihilistic politics of destruction is its point. Populists sent Trump to Washington to “shake things up,” “deconstruct the administrative state,” and destroy the establishment. It is an apocalyptic vision (Müller, 2016, p. 42), in which hope for salvation is grounded in blind faith that the populist leader knows how to deliver (or restore) them to greatness. Tird, populist appropriation of empirical language to make expressive claims generates confusion about what people are saying. Some populists credulously interpret populist discourse as empirical rather than expressive, and thereby get trapped in a real epistemic bubble, not just in its functional equivalent. Liberal democratic pluralists interpret denials of anthropogenic climate change as mistaken, and reply with empirical arguments. Te other side interprets such arguments as fghting words—ways of calling them idiots. To Kahan, references to the consensus of “97% of climate scientists” is just the liberals’ culturally antagonistic meme (Kahan, 2015a). Tat may be how it is heard. But Kahan fails to recognize profound asymmetries here. Populist partisans reject empirically responsible policy discourse. Some partisan actors are operating as policy trolls. Tey are like someone arriving at a chess match to overturn the boards and smash the pieces. While provocation can inspire retaliation in kind, this is not inherent in pluralist politics.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

26  Elizabeth Anderson

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

7.  Lessons for Politics, and for Social Epistemology For politics, the point of modeling problematic phenomena in political epistemology is to supply remedies. Remedies for epistemic bubbles depend on their diagnosis. Group polarization theory and cultural cognition theory ofer models of factors that likely play important roles in generating epistemic bubbles. Everyone, to some degree, is disposed to motivated reasoning, especially when reasoning within partisan or identity groups. No group is immune from epistemic bubbles. Nevertheless, these theories do not fully explain the current epistemic crisis in American politics. As partisan polarization has intensifed, epistemic bubbles have proliferated, enclosed more people, and grown more extreme, intransigent, and destructive. As I write, epistemic bubbles in the U.S. concerning the dangers and prevention of COVID-­19 are literally killing people. As in Brazil, they are coming from the right. Partisan asymmetry in the incidence of bubbles requires explanation. I have argued that the rise of right-­wing populist politics explains this asymmetry. Populism is a strategy for mobilizing voters around an authoritarian political agenda of suppressing minority rights and disrupting the critical, pluralistic representation, and epistemic functions of civil society. It essentially depends on promoting distrust of empirically-­oriented institutions and expertise in the academy, the media, and the civil service; on undermining empirically-­guided expert advice in forming public policy; on replacing policy-­oriented empirical discussion with trolling and insults; on delegitimizing the political opposition by promulgating slanders and conspiracy theories against its leaders and policies; on mobilizing its supporters to tighten their networks around sources of information dedicated to whipping up hysterical fears and resentment of despised minority groups, and collective ignorance of their plight; on promoting group norms of dismissing any claims that discredit the populist party line. Tis has important implications for remedies. Sunstein rightly argues that ep­i­ste­ mic bubbles are more likely avoided if we diversify the groups to which we belong, welcome information from outsiders, and consult a variety of media sources likely to disagree with us. Kahan helpfully recommends that we avoid culturally antagonistic memes, and enrich the cultural meanings attached to claims about risks and remedies so that diverse ideological groups can feel afrmed rather than threatened. Te identifcation of populism as a key driver of epistemic bubbles, along with two models of how populist bubbles or their discursive equivalents function, suggests additional measures. According to the group cognition model, populists will be trapped in epistemic bubbles when their partisan identity is salient, leading them to act as spokespersons for the party line. Yet individuals have multiple identities. Addressing them through their other identities can help advance collective problem-­solving agendas even in a politically polarized era. For example, although Midwestern farmers are among the most skeptical of the reality of anthropogenic climate change (Tabuchi, 2017), they have been enthusiastic adopters of wind power (Hensley, 2016). Addressing them as property owners who can proft from siting wind turbines on their farms makes more sense than telling them that their trucks are ruining the planet. Te group cognition model also ofers a useful perspective on the scale of ep­i­ste­ mic bubbles. Te number of individuals who are personally duped by wacky, toxic, and destructive claims about risk and illegitimacy may be relatively small. But they

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Epistemic Bubbles and Authoritarian Politics  27 have outsized infuence in spreading fake news and antagonistic memes (Guess, 2021; Guess et al.,  2018). Moreover, a small group of extreme partisans is chiefy responsible for enforcing the epistemic norms that trap larger groups into going along with falsehoods. Many who go along may know the truth, and may even wish for escape from these norms. If so, the norms underwriting epistemic bubbles will be brittle. If enough ingroup members get the courage to say they are fed up with toxic falsehoods, these norms will collapse. According to the discursive model, populists tend to replace empirically-­guided policy discussions with trolling and insults, triggering responses in kind. Such behavior is difcult to dislodge at the level of partisan mass media, social media, and campaign rallies. However, meetings of concerned citizens from across the political spectrum, visibly led by politically diverse fgures, focused on building local solutions to local problems, have established norms of respectful discourse that shut down trolls and hyper-­partisans. Southeast Florida has adopted climate change adaptation policies through bipartisan meetings in which norms against introducing culturally antagonistic memes are enforced (Kahan, 2015a). Ultimately, to stop populist epistemic bubbles or their discursive equivalents, we must fnd ways to defuse populism. Populism arises when a substantial sector of voters feels that their concerns are not being addressed by established institutions, that they are neglected and despised. Insofar as today’s populist voters are moved by despair over the declining prospects of less educated white males, an economic agenda focused on improving their material prospects, without excluding others, would be apt. Insofar as they are moved by nativist demographic and cultural worries of being “replaced” by ethnic “others,” no democratic society can fulfll their wishes in such terms. Yet the micropolitics of heart-­to-­heart personal testimony across identity lines may defuse such “us against them” frames by reducing prejudice (Kalla and Broockman, 2020). Tis may have helped gay and lesbian people to win acceptance. Finally, I ofer a lesson for social epistemology. Te study of cognitive biases, understood as processes going on inside people’s heads, is indispensable for any social epistemology. Yet this interior focus may obscure the ways cognition takes place collectively, when people think together as a group to advance joint goals, to understand their social identities, to contest where they stand in relation to other groups. Cognitive biases can be instantiated in the external social norms by which people think together. Rationality is also instantiated socially, through norms of joint inquiry, discussion, and criticism (Herzog, 1998, ch. 4). If the vast literature on cognitive biases has taught us anything, it is that individuals cannot correct their biases all on their own. Tey are not even aware of them. Te epistemology of rationality thus also needs to be fully socialized.

References Allcott, H. and M. Gentzkow. 2017. “Social Media and Fake News in the 2016 Election.” Journal of Economic Perspectives 31(2): 211–36. Anderson, E. 2006. “Te Epistemology of Democracy.” Episteme 3(1–2): 8–22. Benkler, Y., R.  Faris, H.  Roberts, and E.  Zuckerman. 2017. “Study: Breitbart-led rightwing media ecosystem altered broader media agenda.” Columbia Journalism Review.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

28  Elizabeth Anderson (March 3, 2020) Available at: https://www.cjr.org/analysis/breitbart-media-trumpharvard-study.php. Berinsky, A. 2015. “Rumors and Health Care Reform: Experiments in Political Misinformation.” British Journal of Political Science 47(2): 241–62. Bishop, B. and R.  G.  Cushing. 2008. Te Big Sort: Why the Clustering of Like-minded America is Tearing Us apart. Boston: Houghton Mifin. Bullock, J., A.  Gerber, S.  Hill, and G.  Humber. 2015. “Partisan Bias in Factual Beliefs about Politics.” Quarterly Journal of Political Science 10: 519–78. Bump, P. 2017. “A reporter pressed the White House for data. Tat’s when things got tense.” Washington Post. (August 2, 2017) Available at: https://www.washingtonpost. com/news/politics/wp/2017/08/02/a-reporter-pressed-the-white-house-for-datathats-when-things-got-tense/?utm_term=.b4f46f84cd2d. Ditto, P., B.  Liu, C.  Clark, S.  Wojcik, E.  Chen, R.  Grady, et al. 2019. “At Least Bias Is Bipartisan: A Meta-Analytic Comparison of Partisan Bias in Liberals and Conservatives.” Perspectives on Psychological Science 14(2): 273–91. Douglas, M. and A.  Wildavsky. 1982. Risk and Culture: An Essay on the Selection of Technical and Environmental Dangers. Berkeley: University of California Press. Dunlap, R., A. McCright, and J. Yarosh. 2016. “Te Political Divide on Climate Change: Partisan Polarization Widens in the U.S.” Environment Science and Policy for Sustainable Development 58(5): 4–23. Easley, J. 2017, 24 May. “Poll: Majority say mainstream media publishes fake news.” Te Hill (May 24, 2017). Faris, R., H.  Roberts, B.  Etling, N.  Bourassa, E.  Zuckerman, and Y.  Benkler, 2017. Partisanship, Propaganda, & Disinformation: Online Media & the 2016 U.S. Presidential Election. Cambridge, MA: Berkman Klein Center for Internet & Society, Harvard University. Gilbert, M. 1987. “Modeling collective belief.” Synthese 73: 185–204. Gilbert, M. 1989. On Social Facts. Princeton: Princeton University Press. Grynbaum, M. 2017. “Trump Calls the News Media the ‘Enemy of the American People.’” New York Times. (February 17, 2017) Available at: https://www.nytimes.com/ 2017/02/17/business/trump-calls-the-news-media-the-enemy-of-the-people.html. Guess, A. 2021. “(Almost) Everything in Moderation: New Evidence on Americans’ Online Media Diets.” American Journal of Political Science. Available at: https://doi. org/10.1111/ajps.12589. Guess, A., B. Nyhan, and J. Reifer. 2018. “Selective Exposure to Misinformation: Evidence from the consumption of fake news during the 2016 U.S.  presidential campaign.” European Research Council 9.3. Guess, A., J. Nagler, and J. Tucker. 2019. “Less than you think: Prevalence and predictors of fake news dissemination on Facebook.” Science Advances 5(1): eaau4586 [DOI: 10.1126/sciadv.aau4586]. Hare, C. and K.  Poole, 2014. “Te Polarization of Contemporary American Politics.” Polity 46(3): 411–29. Hensley, J. 2016. “Wind power blows away records in the Midwest and Texas.” Into the Wind: the AWEA Blog. (November 18, 2016) Available at: http://www.aweablog.org/ wind-energy-blows-away-records-midwest-texas/.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Epistemic Bubbles and Authoritarian Politics  29 Herzog, D. 1998. Poisoning the Minds of the Lower Orders. Princeton: Princeton University Press. Higgins, A., M. McIntire, and G. Dance. 2016. “Inside a Fake News Sausage Factory: ‘Tis Is All About Income.’” New York Times. (November 25, 2016). Iyengar, S. and K.  Hahn. 2009. “Red Media, Blue Media: Evidence of Ideological Selectivity in Media Use.” Journal of Communication 59: 19–39. Jacobson, G. 2013. “Partisan Polarization in American Politics: A Background Paper.” Presidential Studies Quarterly 43(4). Jost, J. 2017. “Ideological Asymmetries and the Essence of Political Psychology.” Political Psychology 38(2): 167–208. Kahan, D. 2010. “Fixing the Communications Failure.” Nature 463: 296–7. Kahan, D. 2012. “Cultural Cognition as a Conception of the Cultural Teory of Risk.” In S.  Roeser (ed.), Handbook of Risk Teory: Epistemology, Decision Teory, Ethics, and Social Implications of Risk (pp. 725–59). Springer. Kahan, D. 2015a. “Against ‘consensus messaging.’” Available at: http://www.culturalcognition.net/blog/2015/6/10/against-consensus-messaging.html. Kahan, D. 2015b. “Te ‘asymmetry thesis’: Another PMRP issue that won’t go away.” Available at: http://www.culturalcognition.net/blog/2015/6/10/against-consensusmessaging.htmlhttp://www.culturalcognition.net/blog/2015/12/21/the-asymmetrythesis-another-pmrp-issue-that-wont-go-away.html. Kahan, D. 2016. “Te Politically Motivated Reasoning Paradigm, Part 1: What Politically Motivated Reasoning Is and How to Measure It.” In Scott and S.  Kosslyn (eds.), Emerging Trends in Social & Behavioral Sciences: An interdisciplinary, searchable, and linkable resource, pp. 1–16R. . Wiley Available at: http://www.culturalcognition.net/ storage/etrds0417.pdf. Kahan, D., D. Braman, J. Gastil, P. Slovic, and C. K. Mertz. 2007. “Culture and IdentityProtective Cognition: Explaining the White Male Efect in Risk Perception.” Journal of Empirical Legal Studies 4(3): 465–505 Available at: https://papers.ssrn.com/sol3/ papers.cfm?abstract_id=995634. Kahan, D., K.  Jamieson, A.  Landrum, and K.  Winneg. 2017. “Culturally antagonistic memes and the Zika virus: An experimental test.” Journal of Risk Research 20(1): 1–40. Kalla, J. and D. Broockman. 2020. “Reducing Exclusionary Attitudes through Interpersonal Conversation: Evidence from Tree Field Experiments.” American Political Science Review 114(2): 410–25. Khanna, K. and G. Sood. 2017. “Motivated Responding in Studies of Factual Learning.” Political Behavior 40: 79–101. doi:10.1007/s11109–017–9395–7. Mann, T.  E. and N.  J.  Ornstein 2016. It’s even worse than it looks: How the American Constitutional System Collided with the New Politics of Extremism (Expanded ed.). New York: Basic Books. Michael, J. 2011. “Shared Emotions and Joint Action.” Review of Philosophy and Psychology 2. [DOI: 10.1007/s13164–011–0055–2]. Mooney, C. 2012. Te Republican Brain: Te Science of Why Tey Deny Science—and Reality. Hoboken, NJ: John Wiley & Sons. Mudde, C. and C. R. Kaltwasser. 2017. Populism: A Very Short Introduction. New York: Oxford University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

30  Elizabeth Anderson Müller, J.-W. 2014. “ ‘Te People Must Be Extracted from Within the People’: Refections on Populism.” Constellations 21(4): 483–93. Müller, J.-W. 2016. What is Populism?. Philadelphia: University of Pennsylvania Press. Pacherie, E. 2011. “Framing Joint Action.” Review of Philosophy and Psychology 2. [DOI: 10.1007/s13164–011–0052–5]. Pennycook, G. and D. Rand. 2018. “Lazy, not biased: Susceptibility to partisan fake news is better explained by lack of reasoning than by motivated reasoning.” Cognition. Pew Research Center. 2016. “Te Changing Composition of the Political Parties” (September 13, 2016) Available at: https://www.people-press.org/2016/09/13/ 1-the-changing-composition-of-the-political-parties/. Pew Research Center. 2017. Sharp Partisan Divisions in Views of National Institutions. Washington, D.C. (July 10, 2017) Available at: http://assets.pewresearch.org/wp-content/ uploads/sites/5/2017/07/11101505/07–10–17-Institutions-release.pdf. Rousseau, J.-J. 1983. On the Social Contract and Discourses. (D. Cress). Indianapolis, IN: Hackett. Saad, L. “U.S. Still Leans Conservative, but Liberals Keep Recent Gains.” Available at: https:// news.gallup.com/poll/245813/leans-conservative-liberals-keep-recent-gains.aspx. Schafner, B. and S. Luks. 2017, 25 January. “Tis is what Trump voters said when asked to compare his inauguration crowd with Obama’s.” Washington Post. (January 25, 2017) Available at: https://www.washingtonpost.com/news/monkey-cage/wp/2017/ 01/25/we-asked-people-which-inauguration-crowd-was-bigger-heres-what-they-said/. Schmitt, C. 2007 [1932]. Te Concept of the Political (G.  Schwab) (Expanded ed.). Chicago: University of Chicago Press. Sommers, S. R. 2007. “Race and the decision making of juries.” Legal and Criminological Psychology 12(2): 171–87. Sunstein, C. R. 2007. Republic.com 2.0. Princeton: Princeton University Press. Sunstein, C. R. 2009. On Rumors: How Falsehoods Spread, Why we believe Tem, What Can Be Done. New York: Farrar, Straus and Giroux. Sunstein, C. and A. Vermeule. 2009. “Conspiracy Teories: Causes and Cures.” Journal of Political Philosophy 17(2): 202–27. Tabuchi, H. 2017, 28 January. “In America’s Heartland, Discussing Climate Change without Saying ‘Climate Change.’ ” New York Times. (January 28, 2017) Available at: https://www. nytimes.com/2017/01/28/business/energy-environment/navigating-climate-change-inamericas-heartland.html?_r=0.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

2

Weaponized Skepticism An Analysis of Social Media Deception as Applied Political Epistemology

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Regina Rini

Tere’s more than one way to skin a democracy. Back during the Cold War, the apparent threat was sneaky: a dagger within a cloak, a hypnotized Manchurian candidacy. Tose who meant to undermine democratic societies would frst have to slip our defenses. We spent millions on intelligence agencies investigating (ofen spurious) allegations of clandestine propaganda. We thought: so long as we catch them red-­handed, they cannot be efective. Since at least 2014, anti-­democratic actors have used social media to distribute fake news and disingenuous provocations with the aim of afecting the course of democratic elections. Teir most famous interventions came in the 2016 UK Brexit vote and US presidential election. It is hard to establish whether this interference played a decisive role in either vote, but few dispute that it had some efect. According to the United States Justice Department, the goal was to “spread distrust toward the candidates and the political system in general.”1 If we follow the old Cold War mentality, the solution to this problem is obvious: stop the perpetrators. Now that we know they are doing it, we must simply be alert to the next attempt. Shut down the fake news vectors. Educate our citizens about the epistemic risks. Knowing that the problem exists is the frst step toward solving it. But this mentality underestimates the destructive power of social media deception. As I will show, those who turned fake news against democracy may have done so expecting and wanting to get caught. Te subversive efects of their interference may be most powerful when they are not clandestine; it is our awareness of the pervasiveness of fake news that does maximal damage to democratic discourse. Te basic idea is simple: democratic practice requires that citizens trust one another, and awareness of pervasive fake news erodes this trust. Tis means that democratic political culture is distinctively vulnerable to the exposure of fake news transmitted through social media, much more so than politics in non-­democratic societies. By participating in defective testimonial chains, citizens become complicit in their own epistemic victimization. And once citizens come to realize this fact, they reasonably begin to distrust one another’s competence as co-­participants in the collective epistemic projects that make democratic culture possible. 1  United States of America v. Internet Research Agency federal indictment (2018, p. 6). Regina Rini, Weaponized Skepticism: An Analysis of Social Media Deception as Applied Political Epistemology. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Regina Rini. DOI: 10.1093/oso/9780192893338.003.0003

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

32  Regina Rini In this chapter I aim to illuminate the politico-­epistemic damage wrought by Russian social media operations, primarily in the years 2014–16. (I wrote this chapter in late 2018 and revised it in early 2020; by the time it is published, the 2020 US presidential election will surely have turned up new complications that it cannot anticipate.) In the frst half I present an analytically focused narrative of these events, drawing out the malefactors’ apparent indiference to disguising their work. In the second half I ofer a philosophical analysis of the social epistemic environment that made this possible—which is now being dangerously warped by its efects. Te point is both to appreciate the troubling specifcity of Russia’s attack on democratic culture, and also to draw general lessons about the new epistemic vulnerabilities created by social media. As we will see, the ultimate goal of anti-­democratic epistemic interference is not simply to trick people. Te point is not to implant false beliefs. Rather, the point is to induce skepticism. By fooding the channels of public discourse with falsehood, then allowing citizens to know that this has happened, anti-­democrats make it reasonable for us to trust no one, least of all our co-­citizens. Tat is the story of how weaponized skepticism slices through democratic culture—not down some dark alley, but out where everyone can watch it happen.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.  Project Lakhta In late May 2016, an American citizen posed for a photo in front of the White House with a large sign reading “Happy 55th Birthday Dear Boss!” Tis person was doing a favor for an online political group, who explained only that the sign was meant for “our boss . . . our funder.” Tree days later, in St. Petersburg, Yegveniy Viktorovich Prigozhin turned 55. Prigozhin was the director of Concord Management and Consulting LLC, a Russian-­ government-­ linked services provider. Te company allegedly shifed millions of dollars to support Project Lakhta, a political interference operation working domestically and in several foreign countries. One of its benefciaries was the St. Petersburg Internet Research Agency (IRA), which created the fake online group responsible for dispatching that White House sign-­bearer.2 Of course that was not all it did. In February 2018, Prigozhin and 12 other Russian nationals were indicted by Special Counsel Robert Mueller for conspiracy to defraud the United States. According to the indictment, since at least 2014 the IRA had been using Social Security and date-­of-­birth information to steal the identities of unwitting American citizens. With these identities, the IRA created hundreds of email accounts, web sites, and social media personae. Te fakes were then were used to disseminate false and provocative news stories, to organize divisive political rallies in the physical 2  US v. Internet Research Agency (2018, pp. 7–8). Prigozhin is also widely believed to direct the Wagner Group, a Russian mercenary force that is allegedly responsible for stirring conficts in Ukraine, Syria, and Libya and allegedly murdered three journalists in the Central African Republic. See Lister and Shukla (2019).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Weaponized Skepticism  33 world, and to coordinate with mostly unwitting American political operators, including members of the Trump campaign. Te fake entities ranged across the political spectrum, including an Instagram account called “Woke Blacks,” a Twitter account purporting to be the Tennessee Republican Party (@TEN_GOP), and an improbable Facebook community called “Stop A.I.” (which, despite its name, was concerned with alleged Clinton voter fraud rather than the coming robot apocalypse).3 Together these accounts had hundreds of thousands of social media followers.4 All of this happened during the then-­little-­appreciated rise of fake news on social media.5 Not all fake news came from Russian intelligence operatives, of course; some was produced by disingenuous domestic political operators. Others were just in it for the money, like some now-­infamous Macedonian teenagers, or the click-­bait fake headlines that 20th Century Fox used to promote a new flm.6 But the IRA was very much in the business of political manipulation. According to the Prigozhin et al. indictment, its agents were given the instruction, “use any opportunity to criticize Hillary and the rest (except Sanders and Trump—we support them).” Six weeks before the election, an operative running the fake “Secured Borders” Facebook page was scolded for their “low number of posts dedicated to criticizing Hillary Clinton.” On November 3, three days before the election, the fake Instagram account “Blacktivist” declared, “Choose peace and vote for [third party candidate] Jill Stein. Trust me, it’s not a wasted vote.” 7 Tough humans created this deceptive content, it was virally amplifed by “bots,” social media accounts operated by sofware and programmed to automatically disseminate messages. A study by Chengcheng Shao and colleagues used mathematical network modeling to conclude that Twitter disinformation typically propagates outward from a core of non-­human accounts that continuously re-­tweet one another.8 Tough Twitter purged tens of millions of bot accounts in 2018, many remain active.9 Te Alliance for Securing Democracy, a NATO-­funded body, operates a web dashboard called Hamilton 68, which tracks the evolving focus of Russian-­ linked Twitter accounts.10 It is difcult to say how much these activities matter to election results. Afer all, individual voters are themselves ofen unsure about exactly when or how they made their decision; they may not remember the source of the information that pushed them over the edge. But it seems hard to deny that, in the 2016 US president contest, there was some efect. Kathleen Hall Jamieson, a media scholar with a non-­partisan reputation who has studied every presidential election since 1976, told the New

3  “Stop A.I.” stood for “Stop All Invaders.” Accounts linked to the group continued to operate into 2017, when they reacted to media reports of Russian interference by posting: “Instead this stupid witch hunt on Trump, media should investigate this traitor [Obama) and his plane [sic) to Islamize our country.” US v. Khusyaynova criminal complaint (2018, p. 28). 4  US v. Internet Research Agency (2018, p. 15). 5  Silverman and Alexander (2016); Vosoughi et al. (2018). 6  Silverman and Alexander (2016); Rainey (2017). 7  US v. Internet Research Agency (2018, pp. 17–18). 8  Shao et al. (2018). 9  Dellinger (2018). 10  You can fnd Hamilton 68 at https://dashboard.securingdemocracy.org/.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

34  Regina Rini Yorker that she thinks Russian digital disinformation “infuenced who voted, or didn’t vote, and that could have changed the outcome.”11 But my point does not turn on whether Russia succeeded in throwing the election to Donald Trump. My point is that Russia’s interference aimed and succeeded at weakening American democracy merely by playing a role in U.S.  debates—and indeed by being caught playing that role. What is remarkable about Project Lakhta is that, for the intelligence operation of a government headed by a former KGB agent, it did not do much to cover its tracks. Hiring a person to stand in front of the White House with a frivolous message to the conspiracy’s leader does not look like good tradecraf. In fact it looks more like trolling, the sort that is meant to be noticed. Several pieces of evidence suggest the IRA made little efort to go undetected. Te pseudo-­American accounts’ activity patterns correlated with St. Petersburg ofce hours. If subterfuge were a priority, presumably they could have tasked espionage professionals to operate overnight. Instead they hired ordinary people who expected to work ordinary schedules. Indeed, much of the public knowledge of the IRA’s operations comes from the testimony of Lyudmila Savchuk, a former employee who sued the agency in 2015 for abusive labor practices such as successive 12 hour shifs.12 Tat Savchuk was hired in the frst place suggests minimal operational security; she has a history as a political activist and says she took the IRA job intending to write an exposé.13 Afer Savchuk’s lawsuit, ofce hours seem to have been standardized. In October 2017 one conspirator allegedly gave this advice to colleagues:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Posting can be problematic due to time diference, but if you make your re-­posts in the morning St. Petersburg time, it works well with liberals—LGBT groups are ofen active at night. Also, the conservative can view your re-­post when they wake up in the morning if you post if before you leave in the evening St. Petersburg time.14

Perhaps most tellingly, the IRA operations continued, mostly unchanged, even afer the operatives learned they were being monitored by American intelligence. Media reports about the investigation began circulating in autumn 2017. On September 13, an IRA operative emailed her family: “We had a slight crisis here at work: the FBI busted our activity (not a joke). So, I got preoccupied with covering tracks together with the colleagues.”15 Yet they continued to use the same tactics. Te advice to time postings within St. Petersburg ofce hours was sent a month afer the IRA learned the FBI was on to them. Tough Facebook and Twitter shut down the fake accounts used during the 2016 election, the IRA created many more. A September 2018 criminal complaint against Elena Khusyaynova, the IRA’s accountant, provides evidence of new fake profles as recently as May 2018, months afer the original indictment was issued.16 Tere is every reason to believe these activities continued into the 2018 US midterm 11  Mayer (2019). See also Jamieson’s (2018) book, Cyberwar: How Russian Hackers and Trolls Helped Elect a President—What We Don’t, Can’t, and Do Know. 12  Chen (2015). 13  Jones (2018). 14  United States v. Khusyaynova (2018, p. 14). 15  United States v. Internet Research Agency (2018, p. 24). 16  United States v. Khusyaynova (2018, p. 38).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Weaponized Skepticism  35 campaign, though the end of the Mueller probe made it more difcult to get public information. Presumably similar plans are in place for the 2020 US presidential campaign. I cannot defnitively establish that those behind Project Lakhta intended to get caught. But the informality of their operations and their persistence afer being detected are consistent with this theory. Further, as I will now show, being seen to easily manipulate American voters would accomplish one of the key ideological goals of the Putin regime.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Ivan Ilyin and the Humiliation of Democracy Ivan Ilyin was a Russian nationalist philosopher, born in 1883 and exiled in 1922 afer opposing the Bolshevik Revolution. Living in Berlin through the 1920s and 1930s, Ilyin came to praise authoritarianism. But his most central views were negative. He saw democracy as a fundamentally corrupting form of government, destined to be brought down by its own hypocrisies. Ilyin blamed inefectual Russian democratization in the late czarist era for allowing the Bolsheviks to come to power and he saw strongman leadership as the best guard against the social corruption of democracy. Ilyin told his readers: “Te principle of democracy is the irresponsible human atom.”17 Ilyin died in 1954 and was nearly forgotten. However, as historian Timothy Snyder has documented, from the 1990s Vladimir Putin reanimated Ilyin’s ideas to guide his ideological reshaping of Russian political society. Putin regularly quotes Ilyin in national addresses. In 2006 he had Ilyin’s archived papers returned to Russia; in 2005 he brought Ilyin’s bodily remains home from Switzerland. By 2014, Ilyin’s collected works were being distributed to Russia’s top civil servants.18 Snyder interprets Putin-­following-­Ilyin as a practitioner of what he calls the “politics of eternity.” Tis “places one nation at the center of a cyclical story of victimhood. Time is no longer a line into the future, but a circle that endlessly returns the same threats from the past.”19 Unlike in democratic narratives of progress, the politics of eternity does not promise that things will get better. Instead, the relentless focus is on holding of external threats of corruption and decay. In its Ilyin-­infused Russian form, according to Snyder, the politics of eternity acquires a mystical element, as a heroic authoritarian leader arises to ward of the “seduction” of corrupt forms of government.20 Most important for my purposes is Ilyin’s analysis of the failure of democracy. Ilyin interpreted democratic traditions of free speech and individual rights as vehicles for cynical or deluded self-­interest: A citizen is given a limitless right to secret self-­seduction, corruption of others, and an imperceptible self-­ commodifcation. He enjoys the freedom of insincere, deceitful, treacherous, insinuating speech and an ambiguous but calculated

17  Ivan Ilyin, Kosciol i wladza, as quoted in Synder (2018, p. 27). 18  Snyder (2018, pp. 17–18). 19  Snyder (2018, p. 8). 20  Snyder (2018, pp. 22–9).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

36  Regina Rini suppression of truth. He is free to “trust” liars and scoundrels or at least pretend to do so, selfshly feigning this or that political persuasion. To freely express all this spiritual temptation, he is provided with an “election ballot.”

Not only are democratic institutions vicious and corrupting, according to Ilyin, they are also vulnerable. Since democratic governments fail to regulate their citizen’s speech, they can do nothing to stop the rot from spreading: Te freedom of opinion is to be absolute, state functionaries dare not subvert and diminish it. Te most idiotic, harmful, disastrous, and disgusting of “opinions” are “untouchable” only due to the fact that a certain mischievous idiot or traitor took the time to express them, hiding behind their “untouchable” nature.

Finally, this vulnerability leads to the self-­destructive failure of democracy:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Tis order will endure until the seduction undermines the very idea of voting and the very eagerness to submit to a majority . . . then the ballot will be supplanted by an uprising, and an organized totalitarian minority will seize power.21

It is important to stress that, for Ilyin, this is not simply a prediction of democracy’s uselessness. It is also an agenda for wise authoritarians who fnd themselves trapped amid democratic corruption. Democracy will inevitably fall—so the advice is to give it a push in the right direction, to make it fall your way, on your terms, at your time. Putin followed this advice throughout his consolidation of power in the early 2000s. Post-­Soviet Russia, riddled with oligarchic factionalism, was never a fully mature democracy. But under Boris Yeltsin it held genuine elections and permitted a relatively free media. Afer Putin came to power in 1999, these democratic trappings were rapidly subverted. Trough selective prosecution for corruption, Putin exposed his political opponents as self-­interested pretenders—the liars elected by democracy. He supported witchhunts against independent media and NGOs, which were accused of spreading a western homosexual agenda. Afer 2010, the state media began to tell Russians that protests against growing authoritarianism were organized by pedophiles and foreign spies.22 Fake news was a key element in Putin’s extirpation of Russian democracy. As Snyder puts it, the purpose was “both to spread confusion about a particular event and to discredit journalism as such. Eternity politicians frst spread fake news themselves, then claim that all news is fake, and fnally that only their spectacle is real.”23

21  Ivan Ilyin, “On Formal Democracy.” As printed in his Our Tasks (Nashi Zadachi) (1948). Tere is no public English translation of this book, and even the Russian original is hard to fnd outside Russia. Te excerpts here are from a bespoke translation prepared for me by a Russian academic who prefers to be identifed as Fluctuarius Argenteus. Te digital source text (in Russian) appears at: http://apocalypse. orthodoxy.ru/problems/094.htm 22  See Gessen (2017), especially chapter 16. 23  Snyder (2018, p. 11).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Weaponized Skepticism  37

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Te Social Media Bomb By 2014, the Putin regime had begun exporting these tactics. Afer a popular revolution overthrew Ukraine’s pro-­ Kremlin president, Ukrainian and Russian Facebook users began reading false stories of anti-­Russian atrocities in Ukraine. Many were later traced to Russian operations.24 Te IRA’s frst major project was a propaganda leadup to the Russian invasion of Crimea and eastern Ukraine in February 2014.25 On June 23, 2014, Russian-­linked forces in occupied eastern Ukraine shot down a civilian jet, Malaysia Airlines fight 17, killing all 298 people aboard. Over the following days, as Timothy Snyder has since reconstructed, Russian television and social media practiced a sophisticated form of disinformation: instead of fatly denying Russian responsibility, they spun multiple conficting alternative stories. Russian media claimed the surface-­to-­air missile came from Ukrainian forces and had in fact been aimed at Vladimir Putin’s own plane (which was nowhere near the attack). Ten they claimed that Ukrainian fghter jets accidentally shot down MH17 due to a deceptive conspiracy among Ukrainian air trafc controllers. Ten another ofcial admitted that Russia had shot down the plane—but said there was no atrocity, since the plane contained only already-­dead corpses planted by the CIA. Obviously these stories could not all be true at once. Yet, as Snyder puts it, “even if all of these lies could not make a coherent story, they could at least break a story—one that happened to be true.”26 Tis would become a regular pattern of Russian interference in other nations’ politics. Te point was not to gain belief in any one falsehood, but to saturate the epistemic environment with conficting accounts so that the truth appears to be only one of many bickering narratives. Te goal was confusion and disarray, generalized skepticism rather than focused false belief. As Michael Lynch and Zeynep Tufekci have noted, this distortion of the epistemic environment is a common authoritarian tactic.27 Te authoritarian leader establishes control over information by spreading so much disinformation that citizens come to distrust all reporting. Te authoritarian may not be able to guarantee positive belief in his lies, but he can hinder belief in the truth. It was also in 2014 that Russia frst employed disinformation tactics beyond its traditional sphere of infuence. According to British cyber security expert Ben Nimmo, early IRA troll farms (including some of the same accounts spreading lies about Ukraine) attempted to interfere in the independence referendum in Scotland. When the referendum failed, these accounts claimed that the result was rigged and posted misleading videos of ballot stufng (the footage was actually recorded during the 2012 Russian elections).28 By 2016, the IRA had built a large system of Twitter “trolls” (humans posting maliciously under false identities), “bots” (programmed accounts set to automatically redistribute content), and “cyborgs” (accounts sometimes operated by human trolls 24  Priest et al. (2018). 26  Snyder (2018, p. 182). 28  Carrell (2017).

25  Snyder (2018, p. 139). 27  Lynch (2016); Tufekci (2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

38  Regina Rini and sometimes running as bots). Researchers at the University of Edinburgh have identifed 419 IRA Twitter accounts that tweeted about the June 2016 Brexit referendum. Teir posting rate spiked on the afernoon of the vote, but ceased at 4pm British time even though voting continued for another 6 hours; it was 7pm in St. Petersburg. Of the 398 Brexit-­related tweets sent by the network on that date, just nine consisted of original content. Te rest were only amplifcation, mostly retweeting other members of the network.29 Te Edinburgh researchers point out that the Brexit bot/troll network did not seem to be pushing any particular outcome; a content analysis showed that its tweets trended only slightly toward the Leave campaign. Tis lack of direction suggests, again, that the goal was incoherence rather than infuence. Further, the researchers suggest that one explanation for the quick repurposing of these accounts (most had cover identities outside the UK and seemed to be designed for election interference in America and Germany) was opportunistic chaos: “Brexit as a controversial issue . . . provided a suitable topic for generalized disruptive tweeting.”30 Te largest known Russian digital interference operation was, of course, the 2016 US federal election. Sean Edgett, Acting General Counsel for Twitter, later told a Senate committee that Twitter had detected 36,746 automated accounts, totalling 1.4 million tweets in the two months before the election—slightly under 1 per cent of all election-­related tweets on the network.31 Of these accounts, Twitter identifed 2,752 as run directly from the St. Petersburg IRA. Tese seem to have been “cyborg” accounts, sometimes operated by a human agent, but around 47 per cent of the time displaying automated posting behavior.32 Twitter banned and deleted all these accounts afer the election. Tis seems to be a common pattern in social media interference operations. Researchers at City University of London examined 13,493 Twitter accounts that had opined on Brexit and then disappeared once the vote was complete. (Te researchers do not speculate on how many of these accounts were operated from Russia, though they note that few seem to be UK-­based.) Analyzing relations among automated accounts, the researchers concluded they were grouped into two distinct bot networks: one responsible for amplifying messages from human-­operated accounts, and the other dedicated to simply retweeting other bots.33 Looking at the content of these tweets, the researchers found that more than 50 per cent linked to “perishable news,” website addresses that would be deadlinks only six months later. Checking the links that were still active, the researchers found content “that blurs the line between traditional tabloid journalism and user-­ generated content, which is ofen anonymous, fact-­free, and places a strong emphasis on simplifcation and spectacularization.”34 Tough the Brexit bots did prefer the Leave side of the referendum, 17 per cent of their tweets included the term “Remain,” 29  Llewellyn et al. (2019, p. 1159). 30  Llewellyn et al. (2019, pp. 1160–1). 31  Testimony of Sean J. Edgett before the US Senate Judiciary Committee, Subcomittee on Crime and Terrorism. October 31, 2017. https://www.judiciary.senate.gov/imo/media/doc/10-­31-­17%20Edgett%20 Testimony.pdf. pp. 9–10. 32  Testimony of Sean J. Edgett, p. 12. 33  Bastos and Mercea (2019, p. 38). 34  Bastos and Mercea (2019, p. 45).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Weaponized Skepticism  39 higher than among human Brexit tweeters. In other words, the botnet gave support to both sides, seeming to be more interested in sharpening partisanship and spreading shocking tabloid headlines, rather than any particular outcome.35 A further sign that divisiveness was the primary goal of the US operation is that interference continued even afer the election. On November 12, 2016, four days afer voting was fnalized, two political rallies were staged in Manhattan. One, organized by social media group BlackMatters, was called the “Donald Trump is NOT my president” rally. Te other, titled “show your support for president-­elect Donald Trump,” was organized by the Being Patriotic social media group. In fact, both BlackMatters and Being Patriotic were fake accounts operated by the IRA.36 Te thousands of Americans who attended had no idea that they were being orchestrated by the same foreign actors. With cinematic irony, the anti-­Trump protesters marched from Union Square up 5th Avenue, until outside Trump Tower they met police barriers at the site of the pro-­Trump rally.37 It is hard to imagine a more symbolic manifestation of Russia’s divisive achievement than this: the election already over, thousands of American citizens unwittingly steered into confrontation with one another on the streets of the largest city. But this was not the end of it. Te next twist of the knife came the following year, as American citizens began to understand what had been done to them. Reports of Russian cyber interference in the U.S. had circulated as early as 2014, following publication of IRA documents stolen by a group of Russian hackers.38 More than a year before the election, journalist Adrian Chen published an extensive investigation of the IRA’s activities in the New York Times Magazine.39 Still, most people did not pay much attention until autumn 2017, when Facebook and Twitter executives testifying before Congress acknowledged the scope of the problem. It was then that Facebook admitted as many as 126 million Americans may have seen content posted by the IRA.40 Facebook created a tool allowing users to check whether they read IRA-­linked material during the campaign.41 Finally, in February 2018 the Project Lakhta indictments led to top headlines around the world. More recent evidence is sketchy, but it is clear that Russian interference operations continued into America’s 2018 midterm elections.42 In July 2018 Facebook banned 32 pages and accounts for “coordinated inauthentic behavior” around hot-­button social issues.43 A month later it deleted 652 fake accounts linked to Russia or Iran.44 In October 2018 Donald Trump announced he would manufacture a “Great Midterm issue for Republicans!” out of a caravan of Honduran refugees feeing north through Mexico.45 Te following week social media flled with fake news about the caravan. According to one estimate, up to 60 per cent of all caravan-­related tweets were emitted by bots.46

35  Bastos and Mercea (2019, p. 47). 36  United States v. IRA, (2018, 23). 37  Haskins (2017). 38  Seddon (2014). 39  Chen (2015). 40  White (2017). 41  Romm and Wagner (2017). 42  D’Souza (2018). 43  Fandos and Roose (2018). 44  Solon (2018). 45 @realDonaldTrump tweet, October 17, 2018. https://twitter.com/realDonaldTrump/status/ 1052556222459727872 (last accessed October 31, 2018). 46  Lapowsky (2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

40  Regina Rini But there is a diference from 2016. American social media users now know that they are being targeted by Russian interference operations. When they read a political post, they might wonder whether it contains fake news originating from the IRA, or whether the provocative stranger trying to engage them in debate is really a paid troll or even just a bot. Tis awareness has profound consequences for how citizens share information and conduct public debate.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.  Testimonial Sabotage To fully understand the political consequences of IRA interference, we need to examine the epistemology of social media. People use social media for many things, such as admiring baby pictures and cyber bullying. But a central use is information-­ gathering; in 2019, 55 per cent of Americans said they ofen or sometimes got news from Facebook, Twitter, etc.47 Even those who do not seek social media news will stumble across newsy stories while scrolling. People who get news from social media are acquiring beliefs through what epistemologists call testimony. If I believe X because someone else said X, then my justifcation for believing X depends upon the reliability of this person. Much of our knowledge of the world comes about through testimony. For instance, I’ve never been to Singapore, but I know it is humid because others have testifed to that. Tough I haven’t done the experiment myself, I know that bronze melts at 950°C because others have testifed to this happening. Further, much of our testimonial knowledge relies upon being able to trust strangers or acquaintances who have not established an epistemic track record. In day-­to-­day life, we do not interrogate the personal history of each testimony-­giver; instead, we rely upon the belief that most people are reliable on most topics most of the time. Of course, this trust is defeasible and context matters. If a stranger in a coastal Japanese village tells me a tsunami is coming, I probably should believe them. If the same thing happens in downtown Omaha, that’s diferent. Obvious implausibility blocks default reliance on testimony. Our reliance on strangers is not groundless, because we have social norms surrounding testimony. People are expected to be truthful when giving testimony, and to only report on things they are competent to speak about (or explicitly qualify their uncertainty). People who frequently or egregiously violate these norms acquire reputations as liars and fools, and fear of such a reputation motivates most people to testify responsibly. Tat fact justifes our default reliance on stranger testimony.48 Unfortunately, several crucial features of testimonial practice are weakened on social media (as I have discussed in greater detail elsewhere).49 Social media users engage in what I call “bent” testimony, where neither the speaker nor the audience fully expect the speaker to know what they are talking about. People share stories 47  Shearer and Grieco (2019). 48 For varying views on the epistemology of testimony, see Coady (1992), Lackey (2008), and Goldberg (2010). 49  Rini (2017).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Weaponized Skepticism  41 thoughtlessly and retweet-­without-­endorsing. Social media testimony is not regulated by strong norms of sincerity and competence, perhaps because it is so hard to remember who shared what and thereby hold unreliable testifers accountable. Most oddly, even though we now know that all of these things happen, we ofen still treat social media posts like ordinary testimonial acts. We act as if social media testimony conveys epistemic justifcation, even though social media practices have much weaker norms around sincerity and competence. Tis situation, combined with partisanship, provides much of the explanation for the rampant spread of fake news on social media. People pass along outrageous stories on the say-­so of their “friends,” without investigating and without much reason to believe these friends investigated either.50 Testimonial practices that are sensible in other environments misfre badly on social media, yet the behavior of most users seems insensitive to this diference. I frst wrote about fake news and social media in January 2017, immediately afer the 2016 U.S. presidential election. At that time, most people were only just beginning to learn of the phenomenon. Donald Trump hadn’t yet co-­opted the phrase “fake news” to refer to any information unfavorable to his interests. In a sense, that was a more innocent time. In the years since, things have changed. Most people are now explicitly aware that social media is awash with deceptive content; a mid-­2018 survey found that 57 per cent of American adults think the news they fnd on social media is “largely inaccurate.”51 Since the IRA indictments in February 2018, many are aware that it is not just social media stories that sometimes turn out fake, but also the people telling them. Tat friendly local Facebook group leader organizing a rally for your party might be just a stolen profle picture covering a St. Petersburg operator—who is simultaneously organizing a rally against your party. Or maybe it is just a bot.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

5.  Weaponized Skepticism Widespread awareness of coordinated social media fakery has deep consequences for public epistemology. Te most important is this: testimonial skepticism is now a much more reasonable position. In philosophical parlance, a skeptic is a person who challenges the reliability of basic sources of knowledge. A complete testimonial skeptic denies that we gain any justifcation for beliefs from the say-­so of others. According to the testimonial skeptic, I cannot justifedly believe that bronze melts at 950 degrees just because others have told me. When the seaside stranger shouts a warning to me, I have no greater justifcation for anticipating a tsunami than I did a moment before. Like most types of philosophical skepticism, pure testimonial skepticism is utterly impractical. You cannot go around living a human life in the modern world without according at least some evidential weight to the words of others. But limited testimonial skepticism makes more sense. In certain places, at certain times, via certain media, we ought to suspend our usual reliance on testimony. You shouldn’t 50  See above, as well as Rini (2018).

51  Matsa and Shearer (2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

42  Regina Rini believe things on the say-­so of people staggering out of bars at 3:00 am, nor should you give much weight to information conveyed via toner-­streaked pamphlets thrust into your hands outside subway exits. Tese represent sensible local exceptions to a general reliance on testimony. It is increasingly plausible to say that social media itself has become such an exception. Given how widespread we now know fakery and misinformation to be, perhaps we should accord little or no evidential weight to the testimony of social media strangers. On this view, I should not accept what I am told by social media users unless I can personally trace a testimonial chain back to a person or organization with a good epistemic track record. In practice, I will ofen not have the time or resources to do this, so I should typically distrust what I learn on social media. Tis is a sad outcome for anyone who once evangelized the democratic potential of social media. A decade ago, activists cheered the emergence of information channels evading the controls of government and powerful interests. Executives at Facebook and Twitter were quick to trumpet their products’ role in the Arab Spring; knowledge of corruption and protests that would once have been suppressed by the state fowed instead through testimonial chains of smartphone-­equipped citizens.52 Te assumption was not only that we could rely on social media testimony, but that doing so would enable a fourishing democratic culture even in societies where self-­ rule had been a thin dream. Tat heroic vision now seems deluded at best. Authoritarian governments rocked by the Arab Spring quickly deployed mass disinformation to addle social media democrats. As we have seen, Putin’s Russia began doing the same in 2011, and by 2014 had packaged the technique for export. It is crucial to understand that these regimes’ legitimization of testimonial skepticism is not accidental. Tis is weaponized skepticism—a calculated deployment of landmines across the epistemic commons. By forcing democratic citizens to confront the fallibility of their testimonial channels, authoritarians have neutralized a challenge to their hold on information. Tis skepticism is not a purely defensive weapon for authoritarians. It also contributes to the proto-­fascist arsenal of Ivan Ilyin’s followers. To see this, notice the crucial way in which social media misinformation difers from the radio and TV broadcast lies of earlier authoritarians. When the state Propaganda Ministry continuously bleats falsehoods over the radio, it may succeed in confusing some citizens. But citizens themselves are not necessarily implicated in the deception. Tey may be passive and relatively innocent epistemic victims. Among friends, they may even achieve a kind of solidarity, sharing a quiet laugh at the latest absurd broadcast. Traditional misinformation does encourage skepticism, but it is a type of skepticism directed at broadcasters, journalists, and, lurking behind them, the regime itself. But social media disinformation is diferent. Because it is spread directly by users, its unreliability implicates them as well. Te gullible citizen who gets caught passing on fake news or unwittingly retweeting a propaganda bot thereby diminishes their own credibility and weakens their future ability to serve as a channel for information 52  See Tufekci (2017).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Weaponized Skepticism  43 contrary to autocrats’ interests. Recall Ilyin’s depiction of democratic discourse: “the freedom of insincere, deceitful, treacherous, insinuating speech . . . free to ‘trust’ liars and scoundrels or at least pretend to do so . . .” Every time a citizen is seen passing testimony from a liar or fake, Ilyin’s insulting image looks a little more accurate. And this is exactly what the authoritarian wants. Tis severely compounds the harm of disinformation. It is not just that citizens, increasingly unsure what to believe, are deprived of the benefts of knowledge transmission. Testimonial skepticism goes beyond this, to undermine citizens’ trust in one another as citizens, not just as epistemic vectors. A person you cannot trust to convey truth is also not a person you can trust to negotiate mutual self-­government in good faith, nor a person you can trust to stand beside you in solidarity against autocrats. A citizenry riven by mutual charges of fake news and gullibility is not a citizenry prepared to resist the pressures of a truth-­indiferent incipient strongman. Tis is the insidious genius of the Putin regime’s social media strategy. It is distinctively harmful to democratic political cultures, inverting the epistemic resilience of open public debate. And the efect is most virulent precisely when democratic citizens know it is happening; local testimonial skepticism becomes reasonable because we now know a malicious actor is planting disinformation, even fake “citizens,” into our testimonial networks. In the age of social media, Ilyin’s anti-­ democratic prescription need not be subtle or covert. It is fne if the targets know that a hostile autocratic power is at work; the fact that their co-­citizens have foolishly allowed themselves to be co-­opted is itself encouragement to see democratic debate as corrupt and venal.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

6.  Centrifugal Partisanship and the Disarm/Detonate Choice Tere is one further complication to this story: the confusing role of partisanship. Partisanship leads people to apply skepticism selectively. We have all witnessed this on social media. A person seems happy to rely on testimonial backing for news that confrms their partisan worldview, but when they encounter opposing evidence they declare “fake news!” or “troll!” or “bot!.” Tis is a dangerous form of local testimonial skepticism, made selective not by medium, but by content. It leads to a winnowing of permissible evidence. Welcome information gets a pass, but unwelcome evidence is dismissed on the grounds that the medium cannot be trusted.53 Donald Trump is a fan of this style of reasoning; as he told a crowd in Erie, Pennsylvania on October 10, 2018: “I believe in polls. Only the ones that have us up, because they’re the only honest ones. Other than that, they’re the fake news poll. Fake news. Fake news.”54

53  See Kahan (2016) on cultural cognition, and Anderson (Chapter  1, this volume). See also Levy (2017) regarding the psychological mechanisms supporting the efcacy of fake news. 54  Factba.se, transcript: “Speech: Donald Trump Holds a Political Rally in Erie, PA—October 10, 2018.” https://factba.se/transcript/donald-­trump-­speech-­maga-­rally-­erie-­pa-­october-­10-­2018 (last accessed December 5, 2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

44  Regina Rini In one sense this is just confrmation bias, a phenomenon as old as human cognition. But it is important to see how these patterns are deliberately aggravated by social media infuence operations. As we saw earlier, IRA bots were particularly active in the days leading up to the Brexit vote and the US presidential election, but they did not campaign only for Leave or Trump. Teir aim seems to have been simply to whip up division. And that goal continued afer, and beyond, electoral politicking. In May 2016, IRA trolls used false identities to arrange two competing rallies—“Stop Islamifcation of Texas” and “Save Islamic Knowledge”—at the same time outside a Houston Islamic center.55 Public health researchers found that IRA troll accounts posted infammatory content both for and against the use of vaccines; according to Mark Dredze, they “seem to be using vaccination as a wedge issue, promoting discord in American society.”56 Even pop culture can be weaponized; in  2018 Russian social media trolls fanned criticism of the Star Wars flm Te Last Jedi, apparently baiting a cultural fght over inclusiveness in Hollywood.57 Te point of all of this is to transform social media from a neutral information-­ vector into a perpetual partisan battleground. Doing so also changes our co-­citizens into perpetual antagonists, who are rendered epistemically suspicious the moment they present any information we do not like. Te cumulative efect mutates confrmation bias from a pathology of individual information acceptance into a disease of testimonial networks, spreading corruption via the same lines that might otherwise have conveyed democratic cooperation. It is exactly as Ilyin imagined: democratic citizens coming to see one another as playthings of liars and cheats. I think this process has come in three steps. First, insertion of demonstrable falsehoods and fake “citizens” into democratic testimonial networks. Tis step began in Russia in 2011 and in western democracies in 2014. Second, exposure of many of these fakes, leading to rapid-­onset testimonial skepticism. Tis began as early as 2015, but in the United States reached fever pitch in recriminations over fake news and IRA interference in 2017. Tird, channeling of testimonial skepticism along lines of cultural and political partisanship. Tis last step has been building throughout, and we may not yet have seen its peak. We may be approaching an infection point, a period where the mechanisms of democratic testimonial culture begin to seize up and shake themselves apart. People now openly talk about the untrustworthiness of social media users, yet at the same time continue to (sometimes) get their information from them. We are witnessing the emergence of a particularly dangerous sort of testimonial skepticism—one that is selective, partisan, and blatantly hypocritical. We now watch our fellow citizens condemn the gullibility of the other side while falling for their own tribe’s misinformation. Our awareness of social media deception has itself become yet another partisan battlefeld, another breeding ground for mutual suspicion and animosity. Is there anything we can do to stop this process? I think that we have two options for handling weaponized skepticism: disarm or detonate. In an ideal world, we would disarm the threat. We would focus on educating social media users to make 55  Allbright (2017).

56  George Washington University (2018).

57  VanDerWerf (2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Weaponized Skepticism  45

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

thoughtful testimonial decisions, or on pressuring the platforms to provide tools for better testimonial hygiene.58 But I am increasingly doubtful that this approach will work. Individual users seem to have little motivation toward epistemic scrupulousness. Perhaps this is not surprising, given that most people use social media for entertainment or boredom alleviation, not to do epistemic labor. Te problem may be motivation rather than education. And there is so far little evidence that the social media platforms will institute serious reforms. Instead, the best option may be detonation: blow it all up. Rather than try to inculcate responsible use of social media testimony, we might do best to embrace testimonial skepticism, but in a stronger form that does not permit partisan selectivity. If we can convince people to become universally skeptical of social media content—to disbelieve everything conveyed through the medium, not just what their partisan afliation makes disagreeable—then we can at least blunt the efects of induced turmoil, and slow the Ilyinic souring of democratic discourse. On this strategy, social media needs to be brought down to the epistemic level of children’s cartoons; a source of diversion, but not a place any adult would think to go for information. How do we do this? How do we change implicit expectations and norms across vast informal networks of millions of information-­consuming citizens? I am afraid that, as a philosopher, this sort of question is beyond my paygrade. If this approach is to work, it will require the coordinated attention of social scientists, media scholars, and policymakers. For the moment, the question is a normative one: is this indeed the best outcome we can realistically seek? Personally, I am not happy to advocate the strategy of detonation. I would like to believe in the bright social media vision of a decade ago: a revolutionary medium for direct engagement between citizens, a decentralized engine of organization in nascent democracies. It is sad to admit that Ilyin’s descendants have corrupted the tool. Yet it does seem, for now at least, that this is our best remaining option. Te casualties of weaponized skepticism can be mitigated at best.59

References Allbright, C. 2017. “A Russian Facebook Page Organized a Protest in Texas. A diferent Russian page launched the counterprotest.” Texas Tribune November 1, 2017 https:// www.texastribune.org/2017/11/01/russian-facebook-page-organized-protest-texasdiferent-russian-page-l/ (last accessed Dec. 5, 2018). Anderson, E. 2021. “Epistemic Bubbles and Authoritarian Politics.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press.

58  See Rini (2017) for the latter. 59  Great thanks to Andrew Buzzell for invaluable research assistance. For feedback on earlier drafs, thanks to Elizabeth Edenberg, Michael Hannon, referees for OUP, and participants at the Political Epistemology Workshop at Georgetown in 2018 and workshop on deception in politics at Pontifcia Universidad Javeriana and Universidad de los Andes in Bogotá in 2019.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

46  Regina Rini Bastos, M. T. and D. Mercea. 2019. “Te Brexit botnet and User-generated Hyperpartisan News.” Social Science Computer Review 37(1): 38–54. Carrell, S. 2017. “Russian Cyber-activists ‘tried to discredit Scottish independence vote.’ ” Te Guardian December 12, 2017. https://www.theguardian.com/politics/2017/ dec/13/russian-cyber-activists-tried-to-discredit-scottish-independence-vote-saysanalyst (last accessed Oct. 29, 2018). Chen, A. 2015. “Te Agency.” New York Times Magazine June 21, 2015. https://www. nytimes.com/2015/06/07/magazine/the-agency.html (last accessed Oct 30, 2018). Coady, C. A. J. 1992. Testimony: A Philosophical Study. Oxford: Oxford University Press. D’Souza, S. 2018. “Bots, Trolls and fake News: Social Media is a Minefeld for U.S.  Midterms.” CBC October 21, 2018. https://www.cbc.ca/news/world/national-usmidterm-elections-bots-trolls-fake-news-1.4863258 (last accessed Oct. 31, 2018). Dellinger, A. 2018. “Twitter is Suspending more than One Million Accounts per Day in Latest Purge.” Gizmodo July 7, 2018. https://gizmodo.com/twitter-is-suspendingmore-than-one-million-accounts-pe-1827409235. Fandos, N. and K.  Roose. 2018. “Facebook Identifes an Active Political Infuence Campaign Using Fake Accounts.” New York Times Aug. 1, 2018. https://www.nytimes. com/2018/07/31/us/politics/facebook-political-campaign-midterms.html (last accessed Oct. 31, 2018). George Washington University. 2018. “Bots and Russian Trolls Infuenced Vaccine Discussion on Twitter, Research Finds: New study discovers tactics similar to those used during 2016 US election.” ScienceDaily August 23, 2018. www.sciencedaily.com/ releases/2018/08/180823171035.htm (last accessed Dec. 5, 2018). Gessen, M. 2017. Te Future is History: How Totalitarianism Reclaimed Russia. London: Granta Books. Goldberg, S. 2010. Relying on Others: An essay in epistemology. Oxford: Oxford University Press. Haskins, C. 2017. “I Unknowingly Went to a Trump Protest Organized by Russian Agents.” Motherboard November 22, 2017. https://www.vice.com/en_us/article/ ywb9kx/nyc-trump-election-protest-hack-russian-agents-trolls-government. (last accessed Oct. 30, 2018). Jamieson, K. H. 2018. Cyberwar: How Russian Hackers and Trolls Helped Elect a President: what we don’t, can’t, and do know. Oxford: Oxford University Press. Jones, D.  S. 2018. “Russian Troll Factory Whistleblower Blasts Media.” Shooting the Messenger February 25, 2018. https://shootingthemessenger.blog/2018/02/25/russiantroll-factory-whistleblower-blasts-media/(last accessed Oct. 31, 2018). Kahan, D. 2016. “Te Politically Motivated Reasoning Paradigm.” Emerging Trends in Social & Behavioral Sciences https://onlinelibrary.wiley.com/doi/abs/10.1002/ 9781118900772.etrds0417. Lackey, Jennifer 2008. Learning From Words: Testimony as a Source of Knowledge. Oxford: Oxford University Press. Lapowsky, I. 2018. “Here’s How Much Bots Drive Conversation During News Events.” Wired October 30, 2018. https://www.wired.com/story/new-tool-shows-how-botsdrive-conversation-for-news-events/(last accessed Oct. 14, 2018). Levy, N. 2017. “Te Bad News about Fake News.” Social Epistemology Review and Reply Collective 6(8), 20–36.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Weaponized Skepticism  47 Lister, T. and S.  Shukla. 2019. “Murdered Journalists were Tracked by Police with Shadowy Russian Links, Evidence Shows.” CNN January 10, 2019, https://www.cnn. com/2019/01/10/africa/russian-journalists-car-ambush-intl/index.html Llewellyn, C., L.  Cram, R.  L.  Hill, and A.  Favero. 2019. “For Whom the Bell Trolls: Shifing Troll Behaviour in the Twitter Brexit Debate.” Journal of Common Market Studies 57(5), 1148–64. Lynch, M. 2016. “Fake News and the Internet Shell Game.” Te New York Times November 28, 2016. Retrieved from https://www.nytimes.com/2016/11/28/opinion/ fake-news-and-the-internet-shell-game.html Matsa, K.  E. and E.  Shearer,. 2018. “News Use across Social Media Platforms 2018.” Chicago: Pew Research Center. Mayer, J. 2019. “How Russia Helped Swing the Election for Trump.” New Yorker July 9, 2019. https://www.newyorker.com/magazine/2018/10/01/how-russia-helped-to-swingthe-election-for-trump Priest, D., J. Jacoby, and A. Bourg. 2018. “Russian Disinformation on Facebook Targeted Ukraine well before the 2016 U.S. Election.” Te Washington Post October 28, 2018. https://www.washingtonpost.com/business/economy/russian-disinformation-onfacebook-targeted-ukraine-well-before-the-2016-us-election/2018/10/28/cc38079ad8aa-11e8-a10f-b51546b10756_story.html (last accessed Oct. 29, 2018.) Rainey, J. 2017. “20th Century Fox Apologizes for ‘A Cure for Wellness’ Fake News Promos.” Variety February 16, 2017. https://variety.com/2017/flm/news/cure-forwellness-fake-news-promos-studio-apologizes-1201990634/ Rini, R. 2017. “Fake News and Partisan Epistemology.” Kennedy Institute of Ethics Journal, 27(2), E–43. Rini, R. 2018. “How to Fix Fake News.” New York Times October 15, 2018. https://www. nytimes.com/2018/10/15/opinion/facebook-fake-news-philosophy.html Romm, T. and K. Wagner. 2017. ‘“Here’s How to Check if you Interacted with Russian Propaganda on Facebook during the 2016 Election.” Recode December 22, 2017. https://www.recode.net/2017/12/22/16811558/facebook-russia-trolls-how-to-findpropaganda-2016-election-trump-clinton (last accessed Oct. 30, 2018). Seddon, M. 2014. “Documents Show How Russia’s Troll Army Hit America.” Buzzfeed News June 2, 2014. https://www.buzzfeednews.com/article/maxseddon/documentsshow-how-russias-troll-army-hit-america (last accessed Oct. 30, 2018). Shao, C., P.-M. Hui, L. Wang, X. Jiang, A. Flammini, F. Menczer, et al. 2018. “Anatomy of an Online Misinformation Network.” PLoS ONE 13(4): e0196087. https://doi. org/10.1371/journal.pone.0196087 Shearer, E. and E. Grieco. 2019. “Americas Are Wary of the Role Social Media Sites Play in Delivering the News.” Pew Research Center, October 2019, https://www.journalism. org/2019/10/02/americans-are-wary-of-the-role-social-media-sites-play-indelivering-the-news/ Silverman, C. and L.  Alexander. 2016. “How Teens in the Balkans Are Duping Trump Supporters with Fake News.” Buzzfeed News November 3, 2016. https://www. buzzfeednews.com/article/craigsilverman/how-macedonia-became-a-global-hub-forpro-trump-misinfo Snyder, T. 2018. Te Road to Unfreedom: Russia, Europe, America. Tim Duggan Books.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

48  Regina Rini

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Solon, O. 2018. “Facebook removes 652 fake accounts and pages meant to infuence world politics.” Te Guardian August 22, 2018. https://www.theguardian.com/ technology/2018/aug/21/facebook-pages-accounts-removed-russia-iran (last accessed Oct. 31, 2018). Tufekci, Z. 2017. Twitter and Tear Gas: Te Power and Fragility of Networked Protest. New Haven, CT: Yale University Press. Tufekci, Z. 2018. “It’s the (Democracy-Poisoning) Golden Age of Free Speech.” Wired February 15, 2018. Retrieved from https://www.wired.com/story/free-speech-issuetech-turmoil-new-censorship/ United States v. Internet Research Agency LLC. 2018. 18 U.S.C.  §§ 2, 371, 1349, 1028A. Retrieved from https://www.justice.gov/fle/1035477/download United States v. Khusyaynova. 2018. 18 U.S.C.  §§ 2, 371. Retrieved from https://www. justice.gov/fle/1035477/download VanDerWerf, E. 2018. “Russian Trolls used Star Wars to Sow Discord Online. Te fact that it worked is telling.” Vox October 4, 2018. https://www.vox.com/culture/ 2018/10/4/17930308/star-wars-russia-trolls-bots-alt-right-gamergate-fandamentalism (last accessed Dec. 5 2018). Vosoughi, S., D.  Roy, and S.  Aral. 2018. “Te Spread of True and False News Online.” Science 359(6380): 1146–51. White, J. B. 2017. “Facebook Says 126 Million Americans May Have Been Exposed to Russia-Linked US Election Posts.” Te Independent October 31, 2017. https://www. independent.co.uk/news/world/americas/us-politics/facebook-russia-adverts-americansexposed-trump-us-election-2016-millions-a8028526.html (last accessed Oct. 30, 2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

3

Bullshit, Post-­truth, and Propaganda Quassim Cassam

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1. Introduction It has become something of a cliché since 2016 to see the major political events of that year as evidence of the “power of bullshit” or the “rise of post-­truth.”1 Te concepts of post-­truth and bullshit have been used, with varying degrees of precision, by academics, journalists, and political commentators to analyze the 2016 Brexit vote in the UK and the election of Donald Trump as U.S. President. Some uses of these concepts have been frivolous but they have also been used as serious, or semi-­serious, tools of politico-­epistemological analysis. In his Post-­Truth: How Bullshit Conquered the World, James Ball sees “the eco-­system of bullshit” (2017, p. 67) in the Brexit and Trump campaigns. For Matthew D’Ancona in Post-­Truth: Te New War on Truth and How to Fight Back, Trump’s election was a “symptom” of “the rise of post-­Truth” (2017, p. 16). In Post-­Truth: Peak Bullshit and What We Can Do About It, Evan Davis describes “a new kind of politics, and a new kind of bullshit to accompany it” (2018, p. xvii). Te question to be addressed here is whether the concepts of bullshit and post-­ truth are useful analytical tools. What do these fashionable concepts add to the sum of political knowledge and understanding? Te answer is: less than is commonly supposed. It is interesting to compare their use as tools of analysis with older explanations of troubling political developments. When a concept is described as a useful tool of politico-­epistemological analysis, at least two things are implied. One is that the concept in question can be deployed to give an accurate or illuminating description of certain political developments. Te other is that it can be used to explain such developments. A question for analysts who write about the power of bullshit or the rise of post-­truth is whether these ideas are either descriptively or explanatorily adequate in relation to the recent events to which they have been applied. For example, is it plausible or helpful to attribute the success of the campaign for Brexit to the “routine use” (Ball,  2017, p. 4) of bullshit? Or is such an analysis itself an example of bullshit? Even if bullshit is an efective political weapon, is it efective because of the “rise of post-­truth” or is there a better explanation? 1  From a British perspective, the most signifcant political event of 2016 was the surprise Brexit vote in favor of the U.K. exiting the European Union. In the U.S. the landmark political event of the year was the election of Donald Trump as President. On Brexit, see Shipman, 2017. Ball, 2017, D’Ancona, 2017, and Davis, 2018 are three books, all by journalists, that use the ideas of bullshit and post-­truth to analyze these events. Ball talks about the “power of bullshit,” and the “rise of post-­truth” is from D’Ancona. Te concept of bullshit also fgures in my account of Brexit in chapter 4 of Cassam, 2019. Quassim Cassam, Bullshit, Post-­truth, and Propaganda. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Quassim Cassam. DOI: 10.1093/oso/9780192893338.003.0004

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

50  Quassim Cassam Tese questions are difcult to answer because there are so many diferent defnitions and uses of “post-­truth” and “bullshit.” One challenge is to clarify these notions. Most writers on bullshit cite Harry Frankfurt’s essay “On Bullshit.”2 Frankfurt notes that the bullshitter is “trying to get away with something” (2005, p. 23) and that “bullshitting involves a kind of bluf ” (2005, p. 46). However, the essence of bullshit is “indiference to how things really are” (2005, p. 34). On this “mental state” view of bullshit, “the mental state of the person who creates some piece of discourse is a crucial factor in determining whether or not what is created is bullshit” (2002, p. 340). Other accounts insist that bullshit is nonsense, and that what counts as nonsense is not determined by the state of mind of the person emitting it.3 For Frankfurt, bullshitting is diferent from lying and “bullshit is a greater enemy of the truth than lies are” (2005, p. 61). On other accounts, some or all lying is bullshitting, and lying is the greater enemy of truth.4 Te concept of post-­truth was introduced in a 1992 essay by Steve Tesich.5 Tesich’s conception of post-­truth is epistemological: to live in a “post-­truth world” is to live in a world in which citizens connive in their own ignorance. Other accounts of post-­ truth see it more as a normative stance: a post-­truth stance is one that downplays the importance or value of truth. Tere is also the view that post-­truth poses a challenge “not just to the idea of knowing reality but to the existence of reality itself ” (McIntyre, 2018, p. 10). On this metaphysical conception of post-­truth, what is true is what is felt to be true. For example, if there is a popular feeling that violent crime is increasing then, to all intents and purposes, violent crime is increasing, even if the statistics suggest otherwise. Te discussion below will proceed as follows: Section 2 will clarify the notion of bullshit. Section 3 will discuss the merits of various diferent ways of understanding the concept of post-­truth. Finally, Section 4 will return to the issue of whether the notions of bullshit and post-­truth are descriptively or explanatorily adequate in relation to recent political events. Bullshit can be calculated and strategic or spontaneous and unpolished. A bullshitter can be a “mindless slob” (Frankfurt, 2005, p. 21) or a crafsman.6 Te power of bullshit in politics is presumably the power of strategic bullshit but questions remain about whether even talk of strategic bullshit does justice to the techniques employed by the political fgures who are most ofen accused of bullshitting.

2.  On Bullshit To fx ideas, consider this example based on a report in the Washington Post in March 2018:

2  Tis essay was originally published in 1986 and republished as a small book in 2005. All quotations here are from the book version. 3  See Cohen 2002 for a version of this approach. 4  See Webber 2013 for a defence of the view that lying is worse than bullshitting. 5 Tesich 1992. 6  For a defence of the possibility of carefully crafed bullshit see Frankfurt 2005, pp. 22–4.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Bullshit, Post-truth, and Propaganda  51

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

TRADE DEFICIT: At a fundraising dinner President Trump described a meeting with Canadian Prime Minister Justin Trudeau. An issue that came up in the meeting was whether the U.S. had a trade defcit with Canada. Trudeau insisted that this was not the case. Trump insisted that it was. Trump bragged at the dinner that he had no idea at the time whether what he told Trudeau was true, but that he had repeated his claim several times.7

A natural way to describe Trump’s conduct in his meeting with Trudeau is to say that he was blufng. If bullshit involves a kind of bluf then this is one reason to describe Trump as bullshitting in this case. Suppose that P is the proposition that America has a trade defcit with Canada. Trump confdently asserted that P even though he knew full well at the time of his assertion that he had no idea whether it was true. In other words, Trump was knowingly ignorant: he knew that he didn’t know. However, instead of confessing his ignorance he tried to give Trudeau the impression that he, Trump, knew that P.  Te description of the bullshitter as trying to get away with something also applies in this case. Trump was trying to get away with laying a claim to a piece of knowledge that he did not possess. Moreover, he was indiferent to the truth of his assertion about the defcit. Tis, from a Frankfurtian perspective, is the crucial sense in which he bullshitted. Te issue is not whether he cared about America’s trading position with Canada but whether he cared whether he was speaking the truth to Trudeau. He did not. Bullshit is unavoidable “whenever circumstances require someone to talk without knowing what he is talking about” (Frankfurt, 2005, p. 63). In these circumstances, a person’s obligations or opportunities to speak exceed his knowledge. Frankfurt notes that this is quite common in public life, “where people are frequently impelled . . . to speak extensively about matters of which they are to some degree ignorant” (2005, p. 63). It is not their ignorance that makes them bullshitters nor even just the fact that they don’t think they know.8 It is possible to be knowingly ignorant, that is, to know that one doesn’t know, without also being a bullshitter. Bullshitters are not just knowingly ignorant. Tey are knowingly ignorant and do not own up to their ignorance. Tey bluf by pretending to be in the know, or to understand what they do not understand. Other key features of TRADE DEFICIT are: (a) Trump’s assertion was not nonsense. What he said made perfect sense but he was still bullshitting; (b) the judgment that he was bullshitting does not depend for its plausibility on supposing that his claim was false. He would still have been bullshitting even if what he said was true; (c) it is not at all clear whether Trump believed what he told Trudeau. Saying that he had no idea whether his claim was true suggests that did he did not believe it. On the other hand, it is not unknown for people to believe their own bullshit. 7  At the time of the meeting the U.S. had a trade surplus with Canada. 8  Frankfurt describes a conversation between Fania Pascal and Wittgenstein in which the unwell Pascal reported that she felt just like a dog that has been run over. Wittgenstein objected, apparently because he thought that Pascal was talking bullshit. In what sense? According to Frankfurt, the problem was that her statement was not germane to the enterprise of describing reality: “She does not even think she knows, except in the vaguest way, how a run-­over dog feels. Her description of her own feeling is, accordingly, something that she is merely making up” 2005, p. 30.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

52  Quassim Cassam However, in these cases they do not assert their bullshit because they believe it; (d) Trump’s primary objective was presumably to induce in Trudeau the belief that America had a trade defcit. Other objectives can also be easily imagined.9 Tere are several accounts of bullshit that are at odds with TRADE DEFICIT. One view is that bullshit is a type of nonsense, or what Cohen calls “unclarifable unclarity” (2002, p. 333). Yet what Trump said to Trudeau was neither unclear nor unclarifable. Another variety of bullshit that Cohen recognizes is rubbish, “in the sense of arguments that are grossly defcient in logic or in sensitivity to empirical evidence” (2002, p. 333). However, Trump was not giving or presenting an argument. It is true that he had no evidence to back up his assertion, and he might in this sense have been talking rubbish, but he was not thereby bullshitting unless he was aware— as he clearly was—that he had no evidence to back up his assertion. Te bullshitter is not lacking in self-­k nowledge: he knows he doesn’t really know what he is talking about. A diferent and more promising approach to bullshit is to relate it to violations of the norms of assertion.10 Many such norms have been proposed but few that are directly relevant to the question whether a person is bullshitting. One supposed norm of assertion states that one should assert P only if one knows that P. It is not true, however, that a person who asserts that P without knowing that P is necessarily bullshitting.11 If the lottery ticket a person has just bought is overwhelmingly likely to lose she is not bullshitting when she asserts that it is a losing ticket, even if she does not, strictly speaking, know that it is a losing ticket. Another possible norm of assertion states that one should assert P only if one believes P.12 An example of Jennifer Lackey’s makes the point that it is possible for a person to assert P and yet not be bullshitting even if she herself does not believe that P.  Imagine a creationist teacher who is required to instruct her biology students about evolution. When she asserts “Modern day Homo sapiens evolved from Homo erectus” she herself “neither believes nor knows this proposition” (Lackey, 2007, p. 599). Yet she is not bullshitting. She is neither blufng nor trying to get away with something, so the fact that a person’s assertion that P breaches the belief norm of assertion is not sufcient for the assertion to be bullshit. Nor is it necessary. If, in TRADE DEFICIT, Trump believes his own bullshit he is not violating the belief norm but is still bullshitting. It is worth adding that there are cases—TRADE DEFICIT might be one—in which it is hard to say what a person really believes but not hard to say whether they are bullshitting. Lackey proposes the following norm of assertion: (RTBNA): one should assert that P only if “(i) it is reasonable for one to believe P, and (ii) if one asserted that P, one would assert that P at least in part because it is reasonable for one to believe that P” (2007, p. 608). Trump is in breach of this norm in TRADE DEFICIT. Whatever he 9  Another objective might have been to unnerve Trudeau or cause him to doubt himself. Judging by the report in the Washington Post, Trump may well have achieved these objectives. 10  See Koetzee 2019 for a version of this approach. 11  See Koetzee 2019 for further discussion of this approach to bullshit and Lackey 2007 for criticism of the knowledge norm for assertion. 12  Koetzee is in favor of the view that bullshitting involves breaking a belief norm for assertion. See Koetzee 2019.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Bullshit, Post-truth, and Propaganda  53 actually believes, it is not reasonable for him to believe that America has a trade defcit with Canada, and he does not assert that America has a trade defcit with Canada because it is reasonable for him to believe this. Now consider the following case:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

QUESTION TIME: A lazy and not very smart government minister is preparing to answer questions in parliament about a complex issue. He is briefed by a capable and reliable ofcial but still has no real grasp of the issues. According to his briefer, the available evidence supports a particular proposition P.  In parliament, the minister confdently asserts that P and tries to sound authoritative in doing so. Yet he has no real understanding of P or the surrounding issues.

In this all too familiar scenario, the minister is bullshitting. He tries to conceal what he recognizes as his own ignorance and bluf his way through a tricky parliamentary occasion. However, it is reasonable for him to believe that P is true because his briefer assured him that it is. Furthermore, he asserts P in part because it is reasonable for him to believe what his reliable briefer has told him. If P were not reasonable, the briefer would have told him so and the minister would not have asserted it. His assertion is, in this sense, responsive to the reasonableness of P. So here we have a case in which a person is not in breach of RTBNA but is still bullshitting. Tis analysis of QUESTION TIME might be questioned on the following grounds: I take myself to know that E=MC2 but I actually have no idea what that really means.13 If asked I would confdently say that E=MC2 because I have it on good authority that this is the case. I have no theoretical understanding of what the claim amounts to but I am not bullshitting when I assert, on the basis of what experts tell me, that E=MC2. How is this any diferent from the Minister in QUESTION TIME relying on an expert briefer? Te diference is that the minister pretends to understand. Whether I am bullshitting in asserting that E=MC2 depends on whether, in making this assertion, I am trying to give the impression that I understand the equation. If that is the spirit in which I confdently say that E=MC2 then I, too, am bullshitting. An important diference between TRADE DEFICIT and QUESTION TIME is this: in the former case, the President has no idea whether his claim about the defcit is true. Tis is not the minister’s situation in QUESTION TIME. Since he has been told by his trustworthy briefer that P is true, he has a good idea that this is the case. What he lacks is not knowledge of the truth of his assertion but a proper understanding of it. Trump in TRADE DEFICIT and the minister in QUESTION TIME are both phonies but there are diferent ways of being a phony and correspondingly diferent ways of being a bullshitter.14 In TRADE DEFICIT the President pretends to know what he knows he does not know. Te minister in QUESTION TIME pretends to understand what he knows he does not understand. Both are bullshitting but in diferent ways.

13  I thank the editors for raising this question. 14  Frankfurt claims at one point that “the essence of bullshit is not that it is false but that it is phony” (2005, p. 47).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

54  Quassim Cassam To sum up: an assertion is bullshit if only if, at the time of the assertion the person making it: (a) Either realizes that she does not understand her own assertion or realizes that she does not know whether her assertion is correct. (b) Tries to conceal her ignorance or lack of understanding by pretending to know what she does not know, or to understand what she does not understand. Tis is still a mental state account of bullshit. In the most egregious examples of bullshit, the bullshitter has no idea whether her assertion is true and also knows that she has no idea.15 Tere are degrees of ignorance and degrees of bullshit. Te extreme bullshitter is not just ignorant but clueless. Accordingly, the gap between what she actually knows and what she pretends to know is an especially large one. A peculiarity of Frankfurt’s discussion is his insistence that when a bullshitter asserts that P, he is not trying to deceive anyone concerning P itself. Rather, what the bullshitter cares about is “what people think of him” (2005, p. 18). Te crucial idea here is that people who perpetrate bullshit “misrepresent themselves in a certain way” (2005, p. 19):

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Te bullshitter may not deceive us, or even intend to do so, either about the facts, or about what he takes the facts to be. . . . His only indispensably necessary characteristic is that in a certain way he misrepresents what he is up to. . . . Te fact about himself that the bullshitter hides is that the truth-­value of his statements are of no central interest to him  (Frankfurt, 2005, pp. 54–5).

Yet in TRADE DEFICIT Trump’s primary objective was to convince Trudeau that America had a trade defcit with Canada and that he, Trump, believed this to be the case. Te only sense in which he did not intend to deceive Trudeau about the facts is that it is unclear what Trump took the facts to be. QUESTION TIME is more plausibly a case of a bullshitter trying to deceive his audience about himself rather than about the facts. However, what the minister is trying to conceal is not a lack of interest in the truth value of P but a lack of understanding. Tis brings out the range of objectives that a bullshitter may have. Bullshitting does not have just one objective. Some bullshit is mainly concerned to misrepresent the facts. Other bullshit is mainly concerned to misrepresent the bullshitter. One of the challenges facing all mental state accounts of bullshit is that it can be, at least to some degree, indeterminate what the bullshitter’s state of mind is. In other words, it can be indeterminate what they truly believe or are trying to achieve by bullshitting. Te notion of psychological indeterminacy comes up in Frankfurt’s essay: “it is preposterous to imagine that we ourselves are determinate, and hence susceptible both to correct and to incorrect descriptions, while supposing that the 15  Having no idea whether P is a deeper form of ignorance than simply not knowing that P.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Bullshit, Post-truth, and Propaganda  55 ascription of determinacy to anything else has been exposed as a mistake” (2005, p. 66). However, this is not in itself an objection to mental state accounts of bullshit. If an assertor’s mental state is a crucial factor in determining whether their assertion is bullshit, and it can be indeterminate what their mental state is, then all that follows is that it can be indeterminate whether they are bullshitting. Tis is intuitively the right result; sometimes this is indeterminate.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  On Post-­truth Tesich begins his 1992 essay by discussing what he calls the “Watergate syndrome.” Te revelation that President Nixon and his cabinet were a bunch of cheap crooks sickened and disgusted the nation. As a result, and because Nixon was so quickly pardoned, “we began to shy away from the truth. We came to equate the truth with bad news and we didn’t want bad news any more” (Tesich, 1992, p. 12). Instead, we looked to our government to “protect us from the truth” (1992, p. 12). Similarly, in the Iran Contra scandal, President Reagan perceived correctly that the public really did not want to know the truth. So, he lied to us, “but he didn’t have to work hard at it” (1992, p. 13). Such examples of the public’s unwillingness to face up to political reality lead Tesich to his conclusion that “we have acquired a spiritual mechanism that can denude truth of any signifcance. In a fundamental way we, as a free people, have freely decided that we want to live in some post-­truth world” (1992, p. 13). As a result, “we are rapidly becoming prototypes of a people that totalitarian monsters could only drool about in their dreams” (1992, p. 13). For Tesich, the essence of post-­truth is wilfull ignorance. To live in a “post-­truth world” is to live in a world in which citizens connive in their own ignorance. Tey allow themselves to be lied to and turn a blind eye to evidence of wrong-­doing by their political masters. What motivates this policy of wilfull ignorance is that they can’t handle the truth. Tesich does not point out, however, that there is something paradoxical or perhaps even self-­defeating about this policy. Afer all, it is only because the public already knows the awful truth at some level that it seeks to avoid it. Not acknowledging or attending to the truth is not the same as not knowing it. Far from being an indication that the truth has been denuded of any signifcance, the policy of wilfull ignorance that Tesich describes shows how much truth matters to us. It matters enough for us to want to avoid it when it is troubling. If we were genuinely indiferent to the truth there would be little need for us to avert our eyes from it. One thing that is reasonably clear is that on Tesich’s conception “post-­truth” is an epistemological notion. It is not a special type of truth, or a special way for a proposition to be true, but an epistemic posture towards perfectly objective truths. It is the posture of not wanting to know them. In contrast, other accounts of post-­truth see it as a valuational stance. On this reading, what is mainly at issue in talk of “post-­ truth” is the value or importance of truth. For example, President Trump is reported to have told his butler, Anthony Senecal, that the tiles in the nursery of his

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

56  Quassim Cassam Mar-­a-­Lago residence had been personally made by Walt Disney. When Senecal questioned the truth of this claim, Trump’s response was: “Who cares?.”16 Tis sums up the post-­truth attitude. Post-­truth in this sense “concerns our attitude to truth, rather than the truth itself ” (D’Ancona, 2017, p. 126). In contrast, a third account of post-­truth is concerned with truth itself. On this account, “what seems new in the post-­truth era is a challenge not just to the idea of knowing reality but to the existence of reality itself ” (McIntyre, 2018, p. 10). In the case mentioned above, if there is a widespread popular feeling that violent crime is increasing then violent crime is increasing, even if the statistics suggest otherwise. What is true is equated with what is taken to be true; perception is reality.17 Tis is not directly a claim about the importance of truth or our desire not to know the truth. It is about what constitutes truth, and is therefore metaphysical in its orientation. One point to emerge from this survey is that “post-­truth” is not a single concept with a single agreed content. Of the three notions of post-­truth, the one that is most closely related to the concept of bullshit is the valuational notion. Te attitude to truth displayed by Trump at Mar-­a-­Lago, his total lack of concern about whether his assertion was true, is the indiference to how things really are that Frankfurt sees as the essence of bullshit. It is possible that Trump believed his statement was false and intended by making it to deceive. Tat would make his statement a lie. However, it is also possible that he didn’t know, and didn’t care, whether his statement was true. Tis would make his statement bullshit rather than a lie, and therefore an expression of a post-­truth attitude. Te remaining question is whether, as many commentators suppose, it is helpful to describe or explain recent political developments by reference to the rise of post-­truth or the power of bullshit.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.  Propaganda and Hate Speech One reason that such analyses are less novel and less interesting than they might seem is that they tend to be proposed by commentators who do not think of bullshit in the way that Frankfurt and other philosophers think of it. For example, Ball sees “bullshit” as a “catch-­all word to cover misrepresentation, half-­truths and outrageous lies alike” (2017, p. 5). Davis has a similarly lax view. He takes bullshit to include “all forms of mendacity and self-­deception as well as pure nonsense” (2018, p. 33). On this conception, merely accusing a person of bullshitting leaves it open whether they have lied, uttered a piece of nonsense, or told a half-­truth. Furthermore, the power of bullshit includes the power of lies, and the idea that lying can be a powerful political tool is hardly new. Davis’ view implies that Hitler and Goebbels were bullshitters but to describe their anti-­Semitic rants as bullshit is surely to trivialize them. Furthermore, 16  As reported by Matthew D’Ancona. See D’Ancona, 2017, p. 15. 17  In case this seems fanciful, see Lee McIntyre’s report of an exchange between Republican politician Newt Gingrich and a CNN reporter. In response to statistics showing a decline in violent crime Gingrich argued that the average American does not feel safer and that he would go with how people feel. See McIntyre, 2018, pp. 3–4. While Gingrich’s approach can be interpreted, as McIntyre interprets it, as evidence of a commitment to some version of “post-­truth,” it can also be seen as an example of how right-­ wing populism addresses what Stuart Hall calls “lived experiences” (1979, p. 20).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Bullshit, Post-truth, and Propaganda  57 there is all the diference in the world between their outrageous lies and the relatively mild examples of political bullshit given above. To use the concept of bullshit to describe such disparate phenomena is to deprive it of at least some of its interest and usefulness. On a narrower view of bullshit there is more substance to the idea that recent political developments are evidence of the power of bullshit. In a notable passage, Frankfurt describes politics, along with advertising and public relations, as “replete with instances of bullshit so unmitigated that they can serve among the most indisputable and classic paradigms of the concept” (2005, p. 22). Te issue with using the concept of bullshit in Frankfurt’s sense as a tool of political analysis is whether the resulting analyses are plausible, not whether they are interesting or substantial. A  key question here concerns the relationship between lying and bullshitting. According to Frankfurt, advertisers may qualify as liars since “they may know that they are purveying falsehoods with an intention to deceive” (2002, p. 341). However, their most fundamental commitment is as bullshitters and “they are liars only, as it were, incidentally or by accident” (2002, p. 341). In contrast, the mendacious politicians that Frankfurt would have us regard as bullshitters are not liars by accident. How is it possible to lie “by accident”? Frankfurt imagines advertisers deciding what they are going to say in their advertisements without caring what the truth is, and this is what makes their advertisements bullshit. If they also happen to know or discover disadvantageous truths about their product then “what they choose to convey is something that they know to be false, and so they end up not merely bullshitting but telling lies as well” (2002, p. 341). In this case the lying is incidental since the advertiser is “not motivated primarily by an intention to deceive” (2002, p.  341). If lying by one or both sides was not incidental to the Brexit referendum campaign then this would be a reason to question the idea that the campaign represented a new kind of politics based on bullshit. Similarly, it is not obvious that Trump’s success in 2016 can be attributed to the power of bullshit rather than the efcacy of more old-­fashioned political methods. In that case, one would have to conclude that the concept of bullshit is overrated as a tool of politico-­epistemological analysis. Tree test cases help to bring the issues here into sharp focus. Te frst is a slogan used by the Leave campaign for Brexit: “We send the EU £350 million a week—let’s fund our NHS instead.” Is it accurate to describe this as “the ultimate bullshit political claim” (Ball, 2017, p. 52)? Te same question can be asked about a Vote Leave poster stating that “Turkey (population 76 million) is joining the EU.” Lastly, there is Trump’s statement in 2015 that he had witnessed thousands of people in Jersey City cheering as the Twin Towers came down on 9/11.18 He later added that New Jersey has a “heavy Arab population” and that what he witnessed was “not good.” If the concept of bullshit is of any politico-­epistemological value then one would expect it to apply in some or all of these cases, each of which is representative of some of the political tactics employed in 2016. One would expect it to apply both in the sense of 18  See the account of Trump’s claim and responses to it in Ball, 2017, pp. 21–4.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

58  Quassim Cassam providing a helpful description of these tactics and a promising explanation of their efectiveness. Unfortunately, the reality is much more complicated. One reason for questioning the relevance of bullshit in these cases is that in at least two of them the putative bullshit was carefully crafed.19 Te bullshitter produces his statements without a concern for the truth and this implies a nonchalance or carelessness that is difcult to reconcile with any idea that a great deal of care and attention has gone into the production of his bullshit. Yet a great deal of care and attention certainly did go into the production of the two Brexit slogans, just as a great deal of care and attention goes into the production of many advertisements. However, for Frankfurt, bullshit can be carefully crafed and need not be unrefned. Crafed bullshit might be called strategic bullshit. Although the notion of crafed bullshit involves “a certain inner tension” (2005, p. 22) Frankfurt insists that it is coherent. Te strategic bullshitter is still trying to get away with something and displays a certain laxity, even if the mode of laxity cannot be equated with “simple carelessness or inattention to detail” (2005, pp. 23–4). Rather, the strategic bullshitter’s laxity takes the form of indiference to reality. If the Brexit slogans were bullshit, then they were strategic bullshit. It is possible to imagine someone coming up with the fgure of £350 million for the UK’s weekly contribution to the EU, not knowing or caring whether the fgure had any basis in reality and using it in a calculated manner to give voters the impression that the UK was paying a large sum for membership of the EU. Tis is not what happened. Te £350 million fgure was not plucked out of the air. Te Treasury estimated that the UK’s notional annual contribution to the EU was £19 billion, and £19 billion divided by 52 is roughly £350 million. However, this takes no account of UK’s rebate from the EU. When this is taken into account, the net fgure is closer to £175 million. Tose responsible for the £350 million claim knew this but continued to use the higher fgure.20 A letter from the Chair of the U.K. Statistics Authority pointed out that the £350 million fgure confused gross and net contributions and represented a “clear misuse of ofcial statistics.”21 Tere is nothing new in the misuse of ofcial statistics. Te £350 million fgure was clearly tendentious and misleading though it could be seen as having some basis in reality. Te campaign in favor of remaining in the EU also made a series of tendentious and misleading claims. What does it add to describe such claims as “bullshit”? If bullshit includes misleading half-­truths then such a description will not 19  It was carefully crafed in at least the frst two cases. It is much less clear to what extent Trump’s bullshit is crafed. 20 In a lengthy blog post published in 2017 (https://dominiccummings.com/2017/01/09/on-­ the-­ referendum-­2 1-­branching-­h istories-­of-­t he-­2 016-­referendum-­and-­t he-­f rogs-­b efore-­t he-­storm-­2 /), Dominic Cummings, one of the architects of the successful Leave campaign in the Brexit referendum, wrote that the aim of the £350 million a week slogan was “to provoke people into argument” and that “there is no single defnitive fgure because there are diferent sets of ofcial fgures but the Treasury gross fgure is slightly more than £350 million of which we get back roughly half.” 21  Te letter from Sir David Norgrove, which was addressed to Boris Johnson, can be viewed here: https://www.statisticsauthority.gov.uk/wp-­content/uploads/2017/09/Letter-­from-­Sir-­David-­Norgrove-­to-­ Foreign-­Secretary.pdf. Te issues are complex, as is clear from this post: https://fullfact.org/europe/ foreign-­secretary-­and-­uk-­statistics-­authority-­350-­million-­explained/?utm_source=content_page&utm_ medium=related_content.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Bullshit, Post-truth, and Propaganda  59 be strictly inaccurate but adds nothing to the analysis given by the Chair of the U.K. Statistics Authority. If, on the other hand, bullshit is understood as Frankfurt understands it, then describing the £350 million claim as bullshit is not just inaccurate but also potentially confusing. Unlike the President in TRADE DEFICIT, there was no question of those responsible for the claim having no idea of its truth value and not caring that they had no idea. Tey were not indiferent to reality, at least to the extent that it mattered to them that the £350 million fgure was, at least loosely, based on an ofcial fgure. Viewed from this angle, the notion that the £350 million claim was bullshit is closer to being bullshit than the £350 million claim itself. In 2016 Turkey was one of a group of nations being considered for EU membership even though it was a long way from satisfying the conditions for membership. Te  claim that “Turkey is joining the EU” was, if not straightforwardly false, then certainly misleading.22 It is debatable how much its impact would have been lessened by a more accurate claim such as “Turkey is negotiating to join the EU.” Te objective was to frighten voters about the prospect of mass immigration from Turkey, and the latter version might have been just as efective as a means of achieving this objective. Te technique used by Brexit campaigners was similar to one described by Jason Stanley in his work on propaganda. Stanley writes:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Imagine, for example, a non-­Muslim politician in the United States saying, “Tere are Muslims among us”. Te assertion is true; there are many Muslims in the United States. But the claim is clearly some kind of warning. Te speaker is raising the presence of Muslims to the attention of his audience to sow fear about Muslims. (Stanley, 2015, p. 42)

“Tere are Muslims among us” is an example of what Stanley calls “demagogic propaganda.” Te example shows that demagogic propaganda claims can be true, even if they communicate something false, in this case the falsehood that Muslims are inherently dangerous to others.23 In the same way, the claim about Turkey was a truth or half-­truth that worked as a warning about Turkish immigration and its adverse consequences. Since Turkish immigration would be Muslim immigration, the slogan indirectly sowed fear about Muslims entering the U.K. Even more than the £350 million claim, describing the Turkey slogan as bullshit is unhelpful. It adds nothing to the characterization of it as propaganda and makes it more difcult to understand what is going on in such cases. Far from displaying a bullshitter’s indiference to truth and falsity, the propagandist is aware of the possibility of employing true assertions to communicate something false or to demonize minorities and foreigners. Tose who came up with the Turkey slogan were using a tried and trusted propaganda technique. To describe them as bullshitting is to give entirely the wrong impression. Te art of demagogic propaganda is to come up with a true statement that nevertheless distorts the nature of reality. Te bullshitter is “neither on the side of the true nor on the side of the false” (Frankfurt, 2005, p. 56). Te demagogic propagandist is on the side of the false but in the guise of the truth or 22  For some reason Ball claims that “strictly speaking, the poster could be described as true” (2017, p. 53). 23  See Stanley, 2015, p. 42.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

60  Quassim Cassam half-­truth. To describe him as a bullshitter is to misrepresent the true nature of his skulduggery. Tere might be concerns about relying too heavily on Stanley’s account of propaganda given its idiosyncrasies and inconsistency with some of his own examples.24 Te key to Stanley’s conception of propaganda is the notion of an ideal. He defnes propaganda as “the employment of a political ideal against itself ” (2015, p. xiii) and represents political propaganda as using the language of cherished ideals to unite people behind objectionable ends. For example, the idea that leaving the EU would enable the U.K. to “take back control” appealed to the cherished ideal of self-­ mastery. What made it self-­undermining was the probability that exiting the EU would drastically diminish rather increase the U.K.’s autonomy and self-­mastery. However, this conception of propaganda—what might be called the ideals conception—makes poor sense of the claim about Turkish immigration and, for that matter, of “Tere are Muslims among us.” In neither case is the language of cherished ideals being employed, let alone being employed against the ideals themselves. Tere is no ideal to which “Tere are Muslims among us” appeals. A diferent conception of propaganda is needed to make sense of these examples. On what might be called an afective conception, propaganda works by manipulating people’s emotions.25 As Brennan notes anti-­Nazi propaganda during World War II “tried to instil fear and racist paranoia” (2017, p. 36), in this case fear of Germans. In much the same way, as Stanley notes, “Tere are Muslims among us” tries to sow fear of Muslims. Te instilling of fear of Muslims was also the point of the Turkey slogan. In the case of Trump’s comment about people cheering the downing of the Twin Towers, the target emotion was hatred of Muslims. Fear and hatred are not the only emotions that afective propaganda tries to instill. It might also attempt to instill emotions like pride and a sense of belonging. Furthermore, it is not built into the notion of afective propaganda that it manipulates people’s emotions for objectionable ends. Anti-­Nazi propaganda was not propaganda in the service of an objectionable end, even if the specifc forms that this propaganda took were sometimes objectionable. Having said that, fear and hatred are certainly the most potent of the emotions that propaganda instills, and it is a fact that propaganda has ofen served objectionable ends. A form of propaganda that is closely related to afective propaganda is what one might call identity propaganda. Identity propaganda appeals to a narrative about who “we” are, who “they” are, and what “they” and doing to “us.” Identity propaganda seeks to sharpen the division between them and us, and promotes the conception of “them” as “the other”—alien, inferior but a threat to “us.” Trump’s comment about 9/11 was a classic piece of identity propaganda. In this case, Muslims are the Other, and what they are doing to “us”—to America—is cheering our destruction. Since this form of “othering” serves to instill fear and hatred of the Other, identity propaganda can also be afective propaganda.26

24  See Brennan, 2017 for some telling criticisms of Stanley’s account. 26  On “othering,” see Brons 2015.

25  See Brennan, 2017.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Bullshit, Post-truth, and Propaganda  61 Another concept that is useful for present purposes is that of hate speech.27 Tis can be defned, very roughly, as any form of expression that denigrates a person or persons on account of their belonging to a specifc social group. Trump’s Jersey City claim denigrated the Muslim population of New Jersey on account of their Muslim identity. It was therefore both an example of identity propaganda and hate speech. Another even more overt example of this combination was Trump’s denigration of Mexican immigrants to the U.S. as criminals and rapists. To describe such remarks as mere bullshit is not only to misdescribe them but also to underestimate their potency and ofensiveness. Hate speech is dangerous in a way that mere bullshit is not. It is speech that promotes, incites, and justifes violence or discrimination against a particular group of people. Te question raised earlier was whether the notions of bullshit and post-­truth are descriptively or explanatorily adequate in relation to recent political events. Enough has been said to raise questions about their descriptive adequacy. Teir explanatory adequacy is no less questionable. Te apparent efectiveness of the various tactics described above has less to do with the power of bullshit than the power of propaganda and hate speech. Te use of “bullshit” as a catch-­all word to describe all of these tactics ignores signifcant diferences between them and creates the false impression of a single, unifed methodology. Te claim about Arab Americans celebrating 9/11 or Mexican rapists entering the U.S.  resonated with many Trump supporters because they appealed to their xenophobia. What needs to be understood is why xenophobia is a such a powerful force in politics, and the theory of bullshit ofers few answers.28 Tese remarks also have a bearing on the descriptive and explanatory value of the concept of post-­truth. Is it plausible that people are susceptible to bullshit because they don’t want to know the truth, don’t care about the truth, or think that perception is reality? Such a blanket assertion is hard to justify, not least because there are many more straightforward explanations of recent political developments: for example, pro-­Brexit voters who believed that the £350 million claim was literally true were presumably not frightened of the truth, however misguided they might have been in other ways. Others may have believed that the £350 million claim expressed a deeper truth, even if not the literal truth. Such voters cannot be accused of not caring about the truth or of supposing that there is no diference between what is taken to be true and what is true. Tis is presumably not their attitude in their daily lives, where the distinction between what seems true and what is true will be both familiar and important to them. Truth also matters for propagandists and strategic bullshitters. It must matter to them which of their techniques is efective and they will want to know the unvarnished truth about that. Te luxury of treating the truth as unimportant or determined by their own beliefs is not one that they can aford. Furthermore, what people say about the nature of truth or facts in the heat of political debate is one thing. Teir actual view is another. When one of Trump’s ofcials asserted that there 27  See Brown 2015 for an analysis. 28  A much more promising approach is to focus on the roots of ethnonationalism in the U.S.  and U.K. See the account of the roots of ethnonationalism more in Appadurai, 2006.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

62  Quassim Cassam were facts and “alternative facts” about the number of people who attended his inauguration this was seized upon as evidence of the rise of post-­truth.29 It is more likely to be evidence of the absurdities that political operators produce when under pressure to explain away what they recognize as awkward facts. For the hapless ofcial who came up with the line about alternative facts, the obstinate and inconvenient truth was that the crowd for President Trump’s inauguration was smaller than the crowd for President Obama’s. Everything else was spin, as she probably realized. Even afer the limitations of recent analyses of the alleged rise of post-­truth or power of bullshit have been exposed, there is the question whether such analyses are not just fawed but harmful. Te case for saying that they are is that they trivialize and misdescribe political techniques that are more pernicious than those of the bullshitter or post-­truther. It is a travesty to describe hate speech as mere bullshit since this does not even come close to capturing what is wrong with it and why it works. In the same way, it is a travesty to describe demagogic or racist propaganda as bullshit. For Frankfurt, bullshit is a greater enemy of the truth than lies. Tere are many grounds to question this claim, and even greater grounds to question the idea that bullshit is more dangerous than hate speech, demagogic propaganda, or various other techniques whose use has become increasingly prevalent since 2016. Each technique merits serious study in its own right rather than as a species of the genus bullshit or post-­truth. Tis is not to question the legitimacy of these concepts or the possibility that some assertions are indeed bullshit or expressions of a post-­truth attitude. It is to question their ultimate value as tools of politico-­epistemological analysis.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Appadurai, A. 2006. Fear of Small Numbers: An Essay on the Geography of Anger London and Durham: Duke University Press. Ball, J. 2017. Post-Truth: How Bullshit Conquered the World London: Biteback Publishing. Brennan, J. 2017. “Propaganda about Propaganda.” Critical Review 19(1): 34–48. Brons, L. 2015. “Othering, an Analysis.” Transcience 6(1): 69–90. Brown, A. 2015. Hate Speech Law: A Philosophical Examination New York: Routledge. Cassam, Q. 2019. Vices of the Mind: From the Intellectual to the Political Oxford: Oxford University Press. Cohen, G. A. 2002. “Deeper into Bullshit,” in S. Buss and L. Overton (eds.), Contours of Agency: Essays on Temes From Harry Frankfurt Cambridge, MA: Te MIT Press, 321–39. D’Ancona, M. 2017. Post Truth: Te New War on Truth and How to Fight Back London: Ebury Press. Davis, E. 2018. Post-Truth: Peak Bullshit and What We Can Do About It London: Abacus). Frankfurt, H. 2002. “Reply to G. A. Cohen,” in S. Buss and L. Overton (eds.), Contours of Agency: Essays on Temes From Harry Frankfurt Cambridge, MA.: Te MIT Press, 340–4. 29  Te ofcial in question was Kellyanne Conway. Te episode is recounted in the Wikipedia entry on alternative facts: https://en.wikipedia.org/wiki/Alternative_facts.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Bullshit, Post-truth, and Propaganda  63

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Frankfurt, H. 2005. On Bullshit Princeton, NJ: Princeton University Press. Hall, S. 1979. “Te Great Moving Right Show.” Marxism Today, January 1979: 14–20. Koetzee, B. 2019. “Bullshit Assertion.” Te Oxford Handbook of Assertion Oxford: Oxford University Press, January 8, 2019 Oxford Handbooks Online https://www. oxfordhandbooks.com/view/10.1093/oxfordhb/9780190675233.001.0001/ oxfordhb-9780190675233-e-27 (last accessed August 15, 2019). Lackey, J. 2007. “Norms of Assertion.” Noûs 41(4): 594–626. McIntyre, L. 2018. Post-Truth Cambridge, MA: Te MIT Press. Shipman, T. 2017. All Out War: Te Full Story of Brexit London: William Collins. Stanley, J. 2015. How Propaganda Works Princeton, NJ: Princeton University Press. Tesich, S. 1992. “A Government of Lies.” Te Nation 254(1): 1–15. Webber, J. 2013. “Liar!” Analysis 73(4): 651–9.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

4

Truth and Uncertainty in Political Justifcation Fabienne Peter

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.  Introduction Political deliberation and decision-­making typically take place in circumstances of substantial uncertainty about what the right political decision is. Here are some examples. Which political candidate is better suited for ofce, and what makes them suited for ofce? What will be the efects of not renewing an existing international trade agreement? How will the climate crisis afect our lives and what should be done to mitigate them? How to respond to the threat of a pandemic and how to balance public health, social, and economic concerns? Te uncertainty that afects deliberation and decision-­ making on these issues might concern empirical or normative facts—and typically both. Participants in political deliberation and in political decision-­making ofen do not know all the relevant empirical and normative facts and, as a result, remain uncertain about them and about what should be done. What are the implications of such uncertainty for the justifcation of political decisions? I shall focus here on pro tanto political justifcation in a context of political deliberation, not overall justifcation. For clarifcation, political decisions that are overall justifed are politically legitimate and political deliberation should conclude that they are. If they are pro tanto justifed, they are only justifed in some respect. In general, a pro tanto justifed claim that the government should do x is a valid contribution to political deliberation. But doing x may not be the politically legitimate choice, that is, doing x may not be overall justifed. I will not take a stand on overall political justifcation in this chapter, and only address pro tanto political justifcation. My key question is, thus, whether uncertainty can make a diference to the pro tanto justifcation of political decisions in political deliberation. If uncertainty can make a diference to the pro tanto justifcation of political decisions, it is normatively signifcant. Tere is widespread agreement among philosophers that empirical uncertainty can be normatively signifcant: actual or proposed decisions that might be justifed under certainty are unjustifably reckless under uncertainty. To illustrate, suppose a local council considers building a much needed new child care centre. If the contamination of the soil in the relevant area could be remedied, the project should go ahead. But, everything else equal, this project might not be such a good idea if the soil contamination is known to be seriously toxic and if there is a chance that the contamination cannot be eliminated. If we are uncertain about relevant empirical facts, this will be normatively signifcant in the sense that it can change a pro tanto justifcation of a political decision. Exactly Fabienne Peter, Truth and Uncertainty in Political Justifcation. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Fabienne Peter. DOI: 10.1093/oso/9780192893338.003.0005

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Truth and Uncertainty in Political Justification  65 how the uncertainty should be factored in is a difcult question and decision theorists argue about how to answer that question. Tat empirical uncertainty is typically normatively signifcant and that it should thus not be ignored when assessing alternative political decisions is not controversial, however. In light of this very common view about the impact of empirical uncertainty on political justifcation, we might expect that normative uncertainty—uncertainty about relevant normative facts—is normatively signifcant as well. Surely, if there is uncertainty about relevant normative facts, this should also impact on how we should assess political decisions, and on which decisions we should support in political deliberation? For example, if we are uncertain about moral facts such as the permissibility of eating meat, shouldn’t this make us cautious and in this way afect the pro tanto justifcation of political decisions concerning food regulation?1 Interestingly, things are not as straightforward. Call the afrmative view the Symmetry View. It holds that normative uncertainty is normatively signifcant in the same way as empirical uncertainty is. Te political version of the Symmetry View focuses on political justifcation. Te moral version of the Symmetry View focuses on the justifcation of actions, more generally. Many philosophers have indeed argued in favor of the Symmetry View, in moral or political contexts (e.g. Lockhart, 2000; Guerrero, 2007; Sepielli, 2009; Moller, 2011). Tis way of thinking about normative uncertainty is also consistent with the broadly Rawlsian view in political philosophy, which holds that epistemic limitations afect political justifcation. John Rawls (1993: p. 56f) captured the epistemic limitations of political justifcation in the form of a list of the burdens of judgment and the burdens of judgments include uncertainty about both empirical and normative facts. Call the opposing view the Asymmetry View. Tis view denies that normative uncertainty is normatively signifcant in the same way than empirical uncertainty is. Tis view, too, has been defended by philosophers, and the focus typically is on showing that it is a mistake to think that empirical and normative uncertainty are symmetric. Some of those arguments focus on theoretical difculties with incorporating normative uncertainty into the framework of decision theory (e.g. Nissan-­ Rozen, 2015). But other arguments aim to show, more directly, that uncertainty about relevant normative facts is not normatively signifcant in the way in which uncertainty about empirical facts is (see Weatherson,  2014,  2019; Harman, 2015). Arguments for this second view have focused on the moral context, not the political context, and there is not much of a literature on normative uncertainty in political contexts. Te issue is of great signifcance for political justifcation, however. Should we be cautious in relation to moral truth claims in a political context? Or is it, above all, important that the moral truth comes out? Consider the meat-­eating example again, to illustrate the force of the idea that moral truth trumps other considerations, even in a political context. It is fair to say that there is considerable normative uncertainty about this. While some argue that existing meat-­eating practices are morally akin to 1  I understand moral facts here as a subcategory of normative facts. Normative facts will also include prudential facts, aesthetic facts, etc.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

66  Fabienne Peter genocide, others maintain that eating meat is morally permissible. Now suppose the former are right. Te question, then, is, given the seriousness of the moral wrong, isn’t banning our meat eating practices normatively warranted even if circumstances are confusing—as they ofen are in politics—and if political decision-­makers are not in a position to judge with sufcient robustness that this particular decision is the right one? While the idea is forceful, my aim in this chapter is to argue that we should not be reckless in the pursuit of moral truth in political contexts. Accordance with the normative facts is not sufcient for political justifcation. Against those who argue in favor of the Asymmetry View, I show that uncertainty, whether it concerns empirical facts or normative facts, calls for caution in a political context. I conclude that a plausible theory of political justifcation must take seriously the epistemic limitations under which political decision-­making typically takes place. Te chapter is structured as follows. I will start with an overview of diferent types of political uncertainty (Section 2). I then present the argument for the Asymmetry View and develop it for the political context (Section 3). In Section 4, I argue that the main problem with the argument for the Asymmetry View in a political context is its disregard of bad case scenarios—scenarios in which our normative beliefs are mistaken. Section 5 concludes.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Political Uncertainty Political uncertainty comes in many guises. In this section, I spend some time distinguishing between diferent types of uncertainty. Tis taxonomy will be helpful to better understand political uncertainty and to isolate the type of political uncertainty that is the main focus of this chapter—normative uncertainty. Before I begin, the following point of clarifcation will be helpful. My concern in this chapter is less with uncertainty as a feeling than with uncertainty as a relation to the world. We are uncertain, on my use of the term here, if we lack knowledge about decision-­relevant properties of alternative political decisions, that is, when we do not have a good grip on what the problems are and on what the right response is. I take uncertainty in this sense to be a common predicament in a political context. Uncertainty as I understand it can be accompanied by a feeling of being uncertain, but it need not. Epistemically responsible agents aim to bring their feelings of certainty and uncertainty in line with what they know, but doing so is not always easy. And a good number of the problems that we encounter in the political context, I contend, arise from debates about uncertain issues that are held hostage by feelings of great certainty. Focusing, then, on uncertainty as a relation to the world, a frst important distinction is between uncertainty about what to believe and uncertainty about what to do. Belief uncertainty, in a political context, is uncertainty about what to believe about the properties of alternative political decisions. For example, does the proposed policy increase the health status of a particular group of the population or not? Tis is an example of belief uncertainty concerning relevant empirical facts. We might also be uncertain about what to believe about normative facts. Call this normative

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Truth and Uncertainty in Political Justification  67 belief uncertainty. For example, we might be uncertain about whether an increase in the health status of a particular social group is a reason to implement the policy, say because of a question about the relative importance of health status versus opportunities to be healthy. Practical uncertainty is uncertainty about what to do. In the political context, it is uncertainty about which political decision should be made. Practical uncertainty is the main guise of political uncertainty, given the emphasis on decision-­making. Most political decision-­making occurs under some form of practical uncertainty and this uncertainty puts a distinctive pressure on political deliberation. Te reason is that while it is ofen possible to suspend judgment on what to believe, suspending judgment tends not to be an option as a response to practical political uncertainty. Political decision-­making typically cannot be suspended for long, not least because even sticking with the status quo is a political decision. Practical political uncertainty itself comes in several guises. A frst possibility, not very common in political life, I think, is pure practical uncertainty. In those cases, we are uncertain about what to do while not being uncertain about what to believe. A good example of pure practical uncertainty is when, everything else equal, possible alternatives are normatively “on a par” (Chang,  2013). In those cases, practical uncertainty arises because what we should do remains normatively underdetermined—there are normative reasons to favor either of two alternatives, and we know what they are, but no normative reasons to favor one over the other. While I think cases of pure practical uncertainty are relatively rare in a political context, they can occur. In a political context, it would be a case where we know all the relevant empirical and normative properties of two possible political decisions, but there is still uncertainty about what to do. Maybe some very carefully researched proposals for a new civic building could create this sort of uncertainty: while all relevant facts are known and taken into account (suppose), citizens remain divided about which building proposal to choose because of diferences in taste. In the cases of practical uncertainty that are more typical in political life, belief uncertainty and practical uncertainty are closely linked, and that is an important point to recognize. In the cases that are common in political life, practical uncertainty stems from belief uncertainty about either empirical or normative properties of actions or decisions. Belief uncertainty about empirical properties of alternative political decisions can be practically relevant because of its efect on how we perceive their valence. In the health policy example used above, uncertainty about whether or not a policy will, in fact, increase the health status of the poor might create uncertainty about whether or not the policy should be implemented. Call this type of practical uncertainty empirical uncertainty. Empirical uncertainty is uncertainty about what to do that stems from belief uncertainty about decision-­relevant empirical facts. Practical uncertainty can also stem from belief uncertainty about normative properties of political decisions—about their moral rightness or wrongness, their moral justifcation, or how they honor or promote certain values, for example. Uncertainty about what to believe about decision-­relevant normative facts will typically translate to uncertainty about what the right political decision is. For example, if you are uncertain about whether saving lives or improving the quality of

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

68  Fabienne Peter

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Table 4.1  

Uncertainty concerns empirical facts

Uncertainty Uncertainty concerns concerns normative neither empirical nor facts normative facts

Uncertainty about what to believe Uncertainty about what to do

Empirical belief uncertainty Empirical (practical) uncertainty

Normative belief uncertainty Normative (practical) uncertainty

  Pure practical uncertainty

people’s lives is more important in given circumstances, this might translate to uncertainty about the right health care reform. Tis type of political uncertainty, which is the main focus in this chapter, I call normative practical uncertainty, or normative uncertainty, for short. Te taxonomy I have outlined can be summarized in Table 4.1. Using this taxonomy, we can characterize political uncertainty as a type of practical uncertainty that concerns uncertainty about what the right political decision is. Tere can be pure cases of political uncertainty—cases in which the uncertainty is purely practical and not linked to belief uncertainty about empirical or normative properties of a decision. But most cases of political uncertainty arise from underlying belief uncertainty about decision-­relevant empirical or normative facts, and typically both at the same time. Before I move on to discuss the signifcance of normative uncertainty for political justifcation, let me note that the prevalence of disagreements in political life is both a symptom of the extensive practical uncertainty that surrounds political decisions and a further cause of political uncertainty. Disagreements are a symptom of political uncertainty because the less we know about what the right decisions are, the easier it is to end up with conficting judgments about what should be done. Political debates about abortion, for example, are, at least to some extent, characterized by normative uncertainty. If abortion was morally unproblematic in all circumstances and we knew that, this would favor a liberal abortion law, or not having a law at all. If abortion was morally problematic in all circumstances, perhaps the equivalent of murder, and we knew that, this would favor the most restrictive abortion law. Te deep disagreements that persist on this issue suggest that political decisions concerning the regulation of abortion have to grapple with considerable uncertainty about the normative properties of abortion in at least some circumstances. Political disagreements may also become causes of political uncertainty, for better or worse. Call this the disagreement efect. Te disagreement efect obtains if we take the fact of disagreement as a reason to query our empirical and normative beliefs that underpin our evaluation of political decisions. In this way, a political disagreement may lead us to reduce our confdence in our political beliefs and the disagreement becomes an engine of both belief and practical uncertainty. Tis efect is desirable in circumstances where people uphold the wrong political decisions with unwarranted certainty. But it can be problematic if it undermines support for the right decisions.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Truth and Uncertainty in Political Justification  69

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Te Argument for Asymmetry Turning now to the main topic of this chapter: is normative political uncertainty normatively signifcant? Te meat-­ eating example I introduced earlier perhaps suggests that it might not be. Suppose meat-­eating is, in fact, morally akin to genocide. Shouldn’t banning genocidal practices take normative priority, even if this decision is surrounded by uncertainty? Political contexts are ofen confusing, but that should not stop us from doing the right thing. Te key point that might be made is this: what justifes banning the practice is that it is genocidal. Whether citizens believe that the ban is the right decision or not, or if they are uncertain about this issue, is irrelevant. What matters, in this case, is that the practice is stopped and animals protected. More generally put, it is a mistake to think that it is the citizens’ attitudes, including their uncertainty, that carry normative signifcance. Instead, we should recognize that it is the content of those attitudes—the moral principles or values that are endorsed, or the normative reality that relevant beliefs represent— that is, above all, normatively signifcant (see Enoch, 2015 for this distinction). Tis is an important point, which deserves careful consideration. But the question is how it can be supported. Specifcally, if we want to maintain that normative uncertainty can be ignored, what explains the diference between normative uncertainty and empirical uncertainty? As we saw in the child care example above, in cases of empirical uncertainty, we tend to think that recklessness is unjustifed. Te explanation has to show that empirical uncertainty warrants caution in a way that normative uncertainty does not. In other words, it has to show that the Symmetry View, however intuitive it may be, is mistaken, and that the Asymmetry View is correct. Brian Weatherson (2014, 2019) has the most carefully worked out answer to this question that I am aware of. While his argument against Symmetry focuses on the moral context, not the political context, it will be helpful to briefy explain his argument, before examining it focusing on the political context. Weatherson claims that arguments for the normative signifcance of normative uncertainty rest on a faulty analogy between empirical uncertainty and normative uncertainty. According to Weatherson, the problem is that while empirical uncertainty puts you at risk of doing something that is known to be morally wrong, there is no corresponding risk in the case of normative uncertainty. Consider the following case of empirical uncertainty. Suppose you are baking a cake for a children’s birthday party. You are unsure about whether a jar marked as baking powder might contain rat poison. In those circumstances, you should not use the contents of the container: you would risk doing something very morally wrong— poisoning the children—when you can just leave out the baking powder. Cases of normative uncertainty are importantly diferent, Weatherson argues, at least as long as a strong form of normative externalism is true. On the version of normative externalism that he endorses, the normative properties of our actions are independent of our judgments. Consider the case of eating meat again. Eating meat is morally wrong, in given circumstances, if the action has certain normative properties; vice versa, it is not morally wrong if it lacks the properties that would make it wrong. Suppose it would not, in fact, be morally wrong to eat meat in those

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

70  Fabienne Peter given circumstances. If you eat a steak now and then, even though you cannot rule out that it might be wrong—even though you have some normative uncertainty in this regard—you are not running the risk of doing something you know would be morally wrong. Tere is thus a clear disanalogy with the baking powder case. In the latter, you would be running the risk of doing something clearly morally wrong— poisoning the children—if you ignored your uncertainty. By contrast, if it is not morally wrong to eat meat in your circumstances, then you cannot run a risk of doing something morally wrong. Your uncertainty about this is not normatively signifcant, or so Weatherson argues. As mentioned, this argument relies on a strong form of normative externalism. If we were to apply the kind of normative externalism that Weatherson’s argument is premised on to political justifcation, we would hold that if a political decision is the right one, then normative uncertainty does not undermine the decision’s justifcation. Tat people have conficting beliefs about what the right thing to do is, or that they would vote in favor of diferent choices does not, on that view, undermine the justifcation of the decision. Call the relevant externalist view adapted to a political context political factualism.2 Political factualism holds that political decisions are justifed in virtue of according with normative facts that are relevant to the decision. More specifcally, I want to focus here on the following externalist claim:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Sufciency: if a political decision accords with the normative facts, it is pro tanto justifed.

Sufciency states that contributions to political deliberation in support of decisions that accord with the normative facts are valid. If we do not know what the relevant normative facts are, or what decision they warrant, this does not undermine a political decision’s pro tanto justifcation.3 Normative uncertainty, on this political factualist view, is thus not normatively signifcant. To illustrate the force of this factualist view, consider the following case. Suppose that a country is witnessing the rise of what some call a fascist party and that party receives considerable popular support. Once in power, it would infict great harm on many. An opportunity has opened to defeat it, but there is considerable uncertainty and confusion surrounding the decision. Is it fascism? Would the net efect of the party being in power be harmful or would they also realize some new values? Etc. Tis kind of uncertainty about what the right decision is typical of many political decisions. Te factualist can say that this uncertainty should not detract us from doing the right thing in this context. Te defeat of the upcoming party is justifed— in virtue of according with the normative facts—even if nobody is in a position to judge with sufcient robustness at the time that the decision is the right one.

2  See Peter (2020) for a more extensive discussion of this view. 3  See Srinivasan (2015) for an argument for why an externalist view of normativity should be taken seriously.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Truth and Uncertainty in Political Justification  71

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.  Taking Bad Case Scenarios Seriously Does the political factualist view pose a serious challenge for the more mainstream view that uncertainty about both empirical and normative facts should make us cautious, especially in a political context? My aim in this section is to argue that it does not.4 My target is the type of normative externalism that Weatherson’s argument for the normative insignifcance of normative uncertainty is premised on. Before showing where this argument, considered in the political context, goes wrong, let me start by pointing out an advantage of political factualism, building on the example of the fascist party. Given what is at stake in many political decisions, it is plausible that an important criterion for assessing theories of political justifcation is what we might call the Right Decisions criterion. Tis criterion says that theories of political justifcation should aim to justify the right decisions—the decisions that are warranted by relevant normative facts. Political factualism fares well in relation to the Right Decisions criterion, for obvious reasons. If we grant that there is a meaningful distinction between right and wrong political decisions that is determined by what the normative facts warrant, political factualism has the advantage that it justifes the right decisions—the decisions warranted by the normative facts. To see this advantage of political factualism more clearly, consider this contrast with an alternative theory of political justifcation, one that is quite popular among political philosophers, the public reason view (Rawls,  1993; Gaus,  2011). On the public reason view, political decisions are justifed to the extent that they can be justifed to the citizens—on grounds of reasons they can all share or reasons they can each endorse. Tis view is vulnerable to justifying the wrong decisions, if what can be justifed to the citizens difers from the decision favored by the normative facts. Similarly, it is possible that there are right decisions that cannot be justifed to the citizens because they do not recognize it as the right decision. But supporting the right decisions is not the only criterion that a plausible theory of political justifcation has to satisfy. Paraphrasing William James, who, in an epistemological context distinguished between the goal of acquiring true beliefs and the goal of avoiding false beliefs, we can say that we also need to test a theory of political justifcation in relation to a second criterion, which is whether it helps us in making the wrong decisions. Call this the Avoiding Wrongness criterion. In political life, making the wrong decisions can be morally very costly, catastrophic even. Aiming for the right decisions should not make us blind to the possibility of making terribly wrong political decisions, and the plausibility of a theory of political justifcation depends on how well it fares in relation to this second criterion.

4  Note that my aim here is not to reject Weatherson’s defense of the irrelevance of normative uncertainty wholesale; that would require further argument. My aim is more limited as I will only question the argument’s soundness in a political context.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

72  Fabienne Peter How does political factualism fare in relation to the Avoiding Wrongness criterion? Well, just as it treats all right decisions as justifed, it also treats all wrong decisions as not justifed. Tat might be seen to count in its favor and, at frst glance, it might thus appear that the factualist view fares equally well in relation to both ­criteria—aiming for the right decisions and avoiding wrong decisions. But this appearance is only superfcial and hides a problem that political factualism has with regard to the second criterion. To see the problem, it will help to take a close look at what political factualism implies for political justifcation in good case and bad case scenarios. Te distinction I have in mind is the following. In the good case, a political decision accords with what the normative facts favor; in the bad case, it does not. In the good case, political factualism is buoyed by epistemic and moral fortuitousness. If accordance with the normative facts is of prime importance, then, in the good case, no further justifcation is required. Support for this view comes from the thought that should care about making the right decision, but not about whether it is the right decision. If we are about to make the right political decision, worrying about whether it is the right decision can be seen as fetishist (e.g. Smith, 1994; Weatherson, 2019, p. 45f). Now contrast this scenario with a bad case scenario. In the bad case, when a decision does not accord with what the normative facts favor, we encounter a puzzle. If we are advocating or are about to make a wrong decision, we cannot remain unconcerned about this prospect. We would be acting on false beliefs and we are at risk of making a moral error—potentially a very serious one. Consider political decision-­making at the time of a new pandemic threat. Advocating the wrong trade-­ of between protecting people’s lives and protecting their livelihoods potentially results in hundreds of thousands of avoidable deaths. Similarly, consider political decision-­making under moral ignorance of the signifcance of race or gender injustice: making the wrong decisions can perpetuate very serious forms social injustice. In the bad case scenario, there are thus both epistemic and moral pressures to avoid advocating and making political wrong decisions, which we must take seriously and to which we must respond. In the bad case scenario, we cannot settle for whatever the normative facts imply for the justifcation of a decision. Instead, both epistemic and moral considerations push us to question whether the political decision is justifed. Te good case and the bad case are importantly diferent in this regard. Te good case and bad case scenarios warrant diferent responses because the epistemic and moral fortuitousness that is distinctive of the good case is absent in the bad case. While I grant that there is some plausibility to the claim that we should not worry about whether it is the right decision in the good case, a parallel argument cannot be made in the bad case. Te attitude that can be, with some justifcation, criticized as fetishist in the good case, looks very diferent in the bad case. Not worrying about whether a decision is, in fact, the right one comes across as unwarranted complacency in the bad case: if we are at risk of making (or advocating) a wrong decision, we should be wary of making a moral mistake and should look for ways to avoid it. In the bad case, we must take an interest in the question of whether or not the decision is the right one and respond if there is a possibility that it is not.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Truth and Uncertainty in Political Justification  73 Focusing on the Avoiding Wrongness criterion, and bad case scenarios, thus reveals an important problem for the factualist view of political justifcation. Te problem is that while the Right Decisions criterion is normatively self-­fulflling on this view, so to speak, the Avoiding Wrongness Criterion is not. In bad case scenarios, or in scenarios that might be bad cases, we cannot be content with what the normative facts imply. To satisfy the Avoiding Wrongness criterion, we must satisfy ourselves, to the extent possible in the circumstances, that we are not about to make a moral mistake. Te Avoiding Wrongness criterion thus brings into view that the factualist view of political justifcation is, at best, incomplete. Admittedly, the type of problem that bad case scenarios pose might not be equally serious in all contexts. In the political context, however, bad case scenarios clearly matter and a plausible theory of political justifcation must take such scenarios seriously. First, there are signifcant costs attached to making the wrong political decisions as political decisions afect the lives of vast numbers of people. Second, bad case scenarios are likely in the political context. Te epistemic circumstances of political decision-­making are ofen such that we lack a good grasp of all the relevant normative (and empirical) facts. Tird, the problem that bad case scenarios create for political deliberation and political decision-­making is further exacerbated by the fact that it is ofen impossible to know whether we are in a good case or a bad case scenario when considering alternative political decisions. While the good case and the bad case are morally and epistemically distinct, they will ofen be doxastically indistinguishable. Tere is great potential for political deliberators or decision-­ makers to mistakenly believe that they are in a good case scenario when they are in a bad case scenario and vice versa. Relatedly, there are plenty of examples of terribly wrong political decisions that have been defended with great certainty and there are probably also many right decisions that remain very controversial. In sum, given the seriousness of bad case scenarios in political contexts, their likelihood, and their frequent indistinguishability from good case scenarios, the default assumption cannot be that we are in a good case scenario. It has to be that we are in a bad case scenario. Given the signifcance of bad case scenarios in a political context, I conclude that the problem I have highlighted is a serious problem for political factualism. In the bad case, we are inclined towards the wrong political decision—a decision that does not accord with what the normative facts imply should be done. Political factualism has no problem with identifying such a decision as unjustifed, as mentioned. But, unlike in the good case, where there is some plausibility for the claim that making the right decision is all that matters, we should be concerned about the prospect of making wrong decisions. Focusing on bad case scenarios has shown that political justifcation depends on how we respond to what we do not know, including to normative uncertainty. I have also argued that because the good case and the bad case are ofen doxastically indistinguishable, the problem that bad case scenarios pose for political justifcation afects most instances of political deliberation and political decision-­making. In other words, the problem that bad case scenarios pose arises whenever we do not have robust knowledge of what the right decision is, which is the vast majority of cases of political deliberation and decision-­making.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

74  Fabienne Peter

5.  Against Recklessness in Political Justifcation Te problem with political factualism that I have just highlighted undermines the argument for the irrelevance of normative uncertainty in a political context. Normative uncertainty matters in a political context and we should not be reckless in the pursuit of moral truth. Te target of my objection was the following sufciency claim:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Sufciency: if a political decision accords with the normative facts, it is pro tanto justifed.

Sufciency invites us to focus on the good case because it invites us to consider the implications of a political decision that accords with the normative facts. If Sufciency were defensible, and if a political decision accords with the normative facts, the decision is justifed and there would be nothing wrong with contributions to political deliberation or decision-­making that ignore normative uncertainty. More generally, in contexts where it is appropriate to focus on good case scenarios only, and to only care about making the right political decisions, ignoring normative uncertainty might be defensible. It is the exclusive focus on the good case, and the epistemic and moral fortuitousness that characterizes the good case, that invites moral recklessness as a response. Once we start focusing on the bad case, the Asymmetry View, which holds that normative uncertainty does not warrant the same caution as empirical uncertainty, loses plausibility. In the political context, bad case scenarios matter, I have argued. Once bad case scenarios come into view as well, we cannot rely on the epistemic and moral fortuitousness that characterizes the good case. In assessing political claims or in making political decisions in circumstances of normative uncertainty, we cannot rule out that we are in a bad case scenario. We must respond to the possibility that our beliefs are incorrect and that we are about to make or have made a serious moral mistake. Taking seriously bad case scenarios puts a break on the reckless pursuit of what we perceive to be moral truths. If there is a possibility that we are in a bad case scenario, which is the standard case in political deliberation, we cannot be content with what the normative facts imply for the justifcation of our claims and decisions. Tat a political decision accords with the normative facts is not, therefore, sufcient for political justifcation, contrary to what Sufciency claims. If Sufciency is rejected, this implies that the validity of contributions to political deliberation depends on factors other than the truth of a claim. Te truth of a claim made does not, as such, confer validity. Or, as we might also put it, political justifcation is not governed by a truth norm. Instead, we must conclude that political justifcation depends on how we respond to what we do not know, including to normative uncertainty. What about the case of the fascist party, you might object? Is that not an example of where the good case scenario matters in a political context? It is. But the case does not undermine my argument against the Asymmetry View and against the Sufciency

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Truth and Uncertainty in Political Justification  75 claim that a political factualist might want to make. How we assess the case of the rise of the fascist party very much depends on what we assume about what is known. Te case will look very diferent from the perspective of an observer who knows what is going on even as political confusion reigns. If we know that there is a threat from a fascist party, then we have a basis to validly claim that the party should be banned. It is important to note, however, that what makes the claim valid, in this case, is not accordance with the normative facts as such, contrary to Sufciency. It is our knowledge of the party’s harmfulness that makes the claim valid. When we do not have a good grip on what is going on, which is the more normal case in politics, we are back in the scenarios that I have highlighted. Most political decisions are not clear cut and they are surrounded by great normative uncertainty, in addition to uncertainty about relevant empirical facts. Te validity of claims made in political deliberation in those scenarios depends on how we respond to what we do not know.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Chang, Ruth. 2013. “Grounding Practical Normativity: Going Hybrid.” Philosophical Studies 164(1): 163–87. Enoch, David. 2015. “Against Public Reason.” Oxford Studies in Political Philosophy Vol 1. Oxford: Oxford University Press, pp. 112–42. Gaus, Gerald. 2011. Te Order of Public Reason. Cambridge: Cambridge University Press. Guerrero, Alex. 2007. “Don’t Know, Don’t Kill: Moral Ignorance, Culpability, and Caution.” Philosophical Studies 136: 59–97. Harman, Elizabeth. 2015. “Te Irrelevance of Moral Uncertainty.” In R.  Shafer-Landau (ed.), Oxford Studies in Metaethics, 10, Oxford: Oxford University Press, pp. 53–79. Lockhart, Ted. 2000. Moral Uncertainty and its Consequences. Oxford: Oxford University Press. Moller, Dan. 2011. “Abortion and Moral Risk.” Philosophy 86: 425–43. Nissan-Rosen, Ittay. 2015. “Against Moral Hedging.” Economics and Philosophy 3: 1–21. Peter, Fabienne. 2020. “Te Grounds of Political Legitimacy.”  Journal of the American Philosophical Association 6: 372–90. Rawls, John. 1993. Political Liberalism. New York: Columbia University Press. Sepielli, Andrew. 2009. “What to Do When You Don’t Know What to Do.” In R. ShaferLandau (ed.), Oxford Studies in Metaethics. Oxford: Oxford University Press, pp. 5–28. Smith, Michael. 1994. Te Moral Problem. Oxford: Blackwell. Srinivasan, Amia. 2015. “Normativity without Cartesian Privilege.” Philosophical Issues 25: 273–99. Weatherson, Brian. 2014. “Running Risks Morally,” Philosophical Studies 167: 141–63. Weatherson, Brian. 2019. Normative Externalism. Oxford: Oxford University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

5

What Lies Beneath Te Epistemic Roots of White Supremacy

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Briana Toole

In response to nationwide protests, riots, and looting following the deaths of George Floyd and Breonna Taylor, businesses rushed to announce their support for Black Lives Matter and to denounce police brutality. Some had more than words to ofer. Te companies that own Aunt Jemima and Uncle Ben vowed to change their branding, abandoning images long-­linked to racist caricatures. Cities painted “Black Lives Matter” murals. Confederate monuments were toppled, streets and schools renamed. Tese actions seem to herald a great change, to indicate a fundamental shif in how we think of systemic racism and a newfound willingness to confront the legacy of white supremacy. And yet, I believe these actions are little more than window dressing. Te display has changed, but what is within remains largely untouched. White supremacy is more than the sum of its parts. While white supremacy may be a social system with political and material consequences, it is also, as I will suggest in this chapter, an epistemological system. As I will argue here, what lies beneath the social system of white supremacy is an epistemological system, one that serves to justify or naturalize the political, material, and social oppression that white supremacy produces. And so, to truly dismantle white supremacy, we must ask, how do we change an epistemological system? To answer this question, I will draw on work from Kristie Dotson in which she introduces a framework for thinking about the relationship between certain systems (in her case, epistemological systems) and the forms of oppression those systems produce. Dotson is notable for her introduction into the philosophical lexicon of the concept of “epistemic oppression”—oppression which harms an agent in his or her capacity as a knower. In her 2012 paper, Kristie Dotson introduces a distinction between epistemic oppression that is reducible to social and political oppression (frst- and second-­order epistemic exclusions), and epistemic oppression that is not so reducible (third-­order epistemic exclusions). Te latter, Dotson argues, follows from a feature of epistemological systems themselves. Tis feature is what she calls epistemological resilience. A resilient epistemological system is one which resists change. Te distinction between frst- and second-­order epistemic exclusions, on the one hand, and third-­order epistemic exclusions, on the other, is signifcant because it indicates how that form of epistemic oppression is to be addressed. Reducible forms, Dotson suggests, can be Briana Toole, What Lies Beneath: The Epistemic Roots of White Supremacy. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Briana Toole. DOI: 10.1093/oso/9780192893338.003.0006

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

What Lies Beneath  77 addressed by utilizing and revising epistemic resources within an epistemological system. But epistemic oppression that follows from features of epistemological systems requires that we recognize the limitations imposed upon us by those systems. Tis chapter is a spiritual descendant of the project begun by Dotson. Where Dotson was concerned with investigating this distinction and how each form of oppression is to be addressed, I want to dive deeper into epistemological systems themselves. My frst goal here is to examine in greater detail epistemological systems, how they resist change, and how they facilitate oppression. But the primary objective in doing so is to arrive at a better understanding of the conditions that allow for a system to persist over time and across generations. My hope is that such an analysis will allow us to better understand, and eventually topple, oppressive epistemological systems, systems like white supremacy. To accomplish this task, I frst attempt to precisify what a resilient epistemological system looks like and the ways in which such a system resists change. To do so I will take white supremacy as a paradigm case of the phenomenon in question and use it to draw out the conditions for counting as a resilient epistemological system. White supremacy presents an interesting, and fairly intuitive, case of a resilient epistemological system. We see that rather than ultimately rejecting the racial norms that allowed for anti-­Black racism, there is instead a subtle shif in how these norms function from slavery, to Jim Crow, to the prison-­industrial complex. Ultimately, we see that white supremacy is not abandoned so much as it is accommodated. I will suggest that it is this feature of white supremacy—its resiliency in perpetuating its ways of thinking—that ultimately produces much of the racialized social oppression we bear witness to today. I begin here with a brief examination of epistemological systems (Section  1). I  next ofer an analysis of white supremacy (Section  2) with the aim of using this analysis to deconstruct what features constitute a resilient epistemological system (Section 3). I then return to Dotson’s analysis of the relationship between epistemological systems and epistemic oppression (Section 4), before concluding with a brief look at how the resiliency of this epistemological system has created an enduring legacy of white supremacy (Section 4).

1.  Diving into Epistemological Systems My primary aim here is develop an account of epistemological systems in order to better understand what it means to say that some such systems are resilient. I thus begin here by situating my project in relation to Dotson’s, examining in detail her deployment of the concept, so that I may develop it in fner strokes.

1.1  Tree Levels of Epistemic Oppression Kristie Dotson (2012) frst introduced the notion of epistemological systems in order to distinguish between diferent levels of epistemic oppression. Epistemic oppression

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

78  Briana Toole refers to the persistent and unwarranted infringement on one’s ability to acquire and share knowledge. More broadly, I propose we think of epistemic oppression as the persistent and systematic exclusion of certain agents from the practices of knowledge production. As Dotson acknowledges, though it is ofen the case that many epistemic exclusions are the by-­product of certain forms of social and political oppression, there are some that are not so reducible. Such exclusions are distinctly epistemic, she argues, the result of some (fawed) feature of an epistemological system. As Dotson argues, recognizing the distinct underlying causes of epistemic oppression better allows us to consider the locus of change required to address each. To that end, she introduces three levels of epistemic oppression: frst-, second-, and third-­order epistemic exclusions. On the one hand we have frst- and second-­order epistemic exclusions. Tese levels of oppression, Dotson argues, are distinct in that they are (1) reducible to social or political oppression and (2) due to the inefciency or insufciency of shared epistemic resources. Tus, frst- and second-­ order oppression can be addressed by fxing issues within an epistemological system. First-­ order epistemic exclusions arise as a result of the inefcient or unjust application of some epistemic value or epistemic resource within an epistemological system. To illustrate, consider the epistemic value that we ought to give all epistemic agents a default level of credibility until we have some reason to modify this default assessment (Jones,  2002). However, we tend to not to assign a default level of credibility to people of color or women (Fricker,  2007). For instance, under our operative epistemological system,1 we may regard women as too emotional, or people of color as too irrational, and as such degrade the level of credibility we assign their testimony.2,3 Addressing this problem does not require a change to the epistemological system, but that we bring our behavior in line with this value. Tus, when engaging with women or people of color, we should attempt to counteract this tendency instead of preemptively downgrading their credibility assessment. Or, consider instead an instance of a second-­order epistemic exclusion. Tis level of epistemic oppression occurs when epistemic resources are insufcient or inadequate such that an epistemic agent cannot communicate her experience. Such levels of exclusion can be easily seen with the example of sexual harassment. Prior to the development of this term in the 1970s, women were unable to communicate their experiences of sex and gender-­based discrimination in the workplace—the language for doing so simply did not exist (Fricker, 2007).4 Te epistemic resources within our operative epistemological system were simply inadequate to attending to the unique experiences shared by women. Once again, this level of epistemic exclusion can be

1  I would suggest that our current epistemological system is what bell hooks (2004) would call a “white supremacist capitalist patriarchy.” It is under such a system that women and people of color are viewed in the way described. 2  When we assign a defated level of credibility to an interlocutor for reasons related to their social identity, this is what Miranda Fricker (2007) calls a testimonial injustice. 3  See Haslanger (2008) for more on schemas that classify women and people of color in these ways. 4  Tis is what Fricker has termed a hermeneutical injustice—when a person’s own experience is made obscure, even to them, because the resources necessary for understanding that experience are unavailable.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

What Lies Beneath  79 addressed from “inside” the system by developing the concepts and names needed to pick out experiences shared by the socially marginalized.5 Both frst- and second-­order levels of epistemic oppression can be addressed by either revising existing epistemic resources, creating new resources, or ensuring that the available resources are applied more justly and efciently. On the other hand, however, we have third-­order epistemic exclusions. According to Dotson, third-­ order epistemic exclusions are (1) not reducible to either social or political oppression because they are (2) a direct efect of the epistemological system itself. Tus, third-­ order epistemic oppression cannot be addressed from within the system, either by adding, revising, or improving the application of existing epistemic resources. Instead, this level of epistemic oppression can only be addressed by revising our operative epistemological system(s). But, what, precisely is an epistemological system?

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.2  Epistemological Systems Deep Dive According to Dotson, an epistemological system is a holistic concept that refers to the epistemic norms, epistemic resources, habits of cognition, and other conditions that make possible the production of knowledge. As Dotson writes, an epistemological system includes “operative, instituted social imaginaries, habits of cognition, attitudes towards knowers and/or any relevant sensibilities that encourage or hinder the production of knowledge” (Dotson, 2012, p. 121). Peter Railton provides a useful way for thinking about epistemological systems, though he speaks in terms of epistemic frames rather than epistemological systems. Railton writes that we might think of an epistemic frame as functioning much like a camera frame. When taking a picture, one does not see the frame of the camera, but what is seen is seen through it. In this way, the frame provides a (artifcially imposed) limit on the otherwise “undelimited and unbounded character of one’s experience” (Railton,  2006, p. 15). Tese epistemic frames consist of “legions of tacit beliefs,” beliefs which frame our epistemic situations. As Railton writes, “Such framing is a matter of the expectations one brings to situations, the features of situations one tends to notice or ignore, the spontaneous interpretations of events one is primed to make, the possibilities for thought and action that come immediately to mind, and so on” (2006, p. 15). I fnd Railton’s talk of epistemic frames to be a useful metaphor for drawing out what is meant by an epistemological system. In the same way that one may not see the frame of a camera when taking a picture, one may not be consciously aware of the epistemic frames one is employing; but what one sees, believes, and knows is seen, believed, and known, through this frame. Epistemological systems, I believe, provide this epistemic frame. Without such frames, we are overburdened with 5  I would argue that patriarchy, like white supremacy is an epistemological system. In that sense, then, we can see that frst- and second-­order revisions of that system—like the inclusion of new concepts or the modifcation of our behavior to be in line with our commitments—mitigates the epistemically oppressive nature of that system, but does not fundamentally alter the system itself. We have seen similar revisions of white supremacy but, as I argue in this chapter, the essential function of this system remains largely unchanged.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

80  Briana Toole information for which we lack the categorical tools necessary for understanding that information. Consequently, this information ends up being meaningless without a flter through which to understand and interpret it. For those working in the continental tradition, this notion may call to mind the perhaps more familiar Foucauldian idea of an episteme. An episteme, Foucault writes, “defnes the conditions of possibility of all knowledge” (Foucault, 1966, p. 68). Derek Anderson (ms) argues that an episteme, though it does not determine which claims are true or false, “determines which knowledge claims and which methods for arriving at knowledge claims count as scientifc, rational, intuitive, or commonsensical.” Te question to consider here is how epistemological systems perform this function. Extrapolating from the accounts above, I suggest that epistemological systems are like governing bodies for knowledge-­ acquisition. Consider, for instance, that a political system constructs rules for the passage and enforcement of policies, stipulates which bodies have authority, and determines the relationship between a government and its people. Epistemological systems, by comparison, construct rules for the formation and revision of beliefs, stipulate which method of forming beliefs we ought to employ or avoid, indicate how we ought to weigh evidence, specify which standards a belief must meet to count as knowledge, and so on. Tus, where political systems govern policies and people, epistemological systems govern beliefs and knowledge. And just as there are multiple political systems—monarchies and oligarchies, democracies and autocracies—so too are there many diferent epistemological systems.6 Such systems range from the prosaic and familiar, like arithmetical knowledge, to the mundane and arcane, like Bayesianism. Epistemological systems serve as the necessary background and starting point from which we engage with the world, gather and interpret evidence, and generate new beliefs. Importantly, these systems shape our experiences, ofer meaning, and direct future inquiry. I submit that epistemological systems shape and constrain what we know in at least three identifable (but not exhaustive) ways: they are normative, predictive, and attendant. As Railton alluded, epistemological systems are normative systems. Epistemological systems are normative in that they license certain beliefs and eliminate from consideration beliefs that are unsupported by or inconsistent with other beliefs in the system. Epistemological systems are also predictive in the sense that they prime us to form certain beliefs, making readily available some hypotheses rather than others to explain or interpret a body of evidence. Finally, these systems are attendant, by which I mean that they infuence which features of the word we attend to and which we ignore. As a simple illustration of the ways in which these features shape and constrain knowledge, consider a toy case. First, let us start with the assumption that one commonplace epistemological system is the basic scientifc commitment to the necessity of concrete, physical evidence in support of a belief. Such a commitment does not license a belief in miracles, for instance, anymore than it licenses a belief in 6  An important distinction between political and epistemological systems to note is that there may be multiple epistemological systems in efect at a given time (that do not confict), whereas competing political systems (e.g. autocracies and democracies) cannot coexist.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

What Lies Beneath  81 magic. Second, imagine that my chapstick—an ever-­present fxture for me—has just rolled of my cofee table on to the foor. Tere are a number of ways to explain this occurrence—perhaps my foor is uneven; there may have been a tiny earthquake; the air conditioner may have just cut on; my apartment may have a ghost; or perhaps my chubby cat’s squishy tail disturbed the air in just the right way. Given my scientifc commitments, I am licensed to believe some of these explanations—for example that the foor is uneven—and not others—for example that a ghost has caused this disturbance. Tis commitment primes me to entertain certain hypotheses—my cat’s tail as the culprit—more readily than others—given that I live in a no-­quake area, I am not primed to entertain this explanation. Finally, it will make me attend to explanations compatible with this commitment—once the chapstick falls, I will not search the air for a ghost, but I may listen for the hum of the air conditioner or the mewing of my cat.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.3  What We (Don’t) Know Importantly for my purposes, these systems shape not just what we know, but what we are in a position to know.7 It has been thoroughly discussed by those working at the intersection of social epistemology, feminism, and philosophy of race, the extent to which excluding certain parties from meaning-­generating practices has also excluded certain bodies of knowledge. An epistemological system that, for instance, excludes women as sources of meaning and knowledge (as was the case in legal theory until the early 1990s), will also be one in which knowledge of certain injuries experienced exclusively by women will not be possible for epistemic agents situated in certain ways. Take, for instance, the now well-­known example from Miranda Fricker (2007) on hermeneutical injustice and sexual harassment. Prior to the development of the conceptual resources needed to understand sexual harassment— and to the subsequent modifcation of the epistemological system of legal theory— we would not have been in a position to know that some act constituted sexual harassment (see also Toole, 2019). Of course, my intention here is not to argue that epistemological systems are intrinsically pernicious. Quite the opposite, these systems are useful, indeed necessary, for making meaning of and for engaging with the world. Rather, the problem lies in the resiliency of faulty or maladaptive epistemological systems.8 In their discussion of resilient social-­ecological systems, Walker et al. defne resilience as “the capacity of a system to absorb disturbance and reorganize while 7  Epistemological systems can be designed (or are such that) they make knowledge of certain states of afairs or facts impossible. 8  It is worth noting that resiliency is not, in and of itself, a bad feature. Rather, the problem is that both good and bad epistemological systems can be resilient. It is beyond the scope of this chapter to draw a distinction between what counts as a good or bad epistemological system. However, broadly speaking, we might think that an epistemological system is faulty to the extent that it is fawed in the ways I note in the preceding paragraphs. Tat is, if the system is designed or functions so as to exclude certain epistemic agents or knowledge contributions, if it rationalizes false beliefs (and renders true beliefs irrational), and so on.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

82  Briana Toole undergoing change so as to still retain essentially the same function, structure, identity, and feedbacks” (Walker et al., 2004). Similarly, I suggest that an epistemological system is resilient to the extent that it can absorb disturbances without changing its underlying structure and maintaining its essential function, or to the extent that it can resist change. It is this feature of epistemological systems that particularly ­troubles me, as it is this feature which, I will argue, leads to a variety of forms of oppression—social, political, and epistemic. But frst, a caveat. My claim is not that any epistemological system that can survive or persist despite change is somehow a fawed system. Rather, epistemological systems must be able to appropriately absorb change. Or, as Walker et al. note, systems must be transformable, capable of creating “untried beginnings from which to evolve a new way of living when existing . . . structures become untenable” (Walker et al., 2004). We might think of this as the fexibility of a system to update and evolve when presented with new information or background conditions. As Gaile Pohlhaus (2011) argues in her examination of the relationship between social identity and epistemic resources, epistemic resources are tools which enable us to understand and communicate our experiences. To the extent that existing resources fail to adequately do this, they must be open to addition and revision. To return to the example ofered by Fricker, the available epistemic resources did not capture the phenomenon of gender-­based workplace harassment, but we were able to add to the body of existing resources the concept of sexual harassment to remedy this inadequacy. Te same is true of epistemological systems. If an epistemological system no longer reliably produces knowledge or systematically excludes or obscures the knowledge and experiences of certain groups, we must revise that system. Problematically, as Dotson notes, the worry ofen lies not merely in the inadequacy or inefciency of parts of the epistemological system, but with the system in its entirety. As I will soon show, resilient epistemological systems are epistemological systems that ofer the appearance of having either been radically revised or altogether abandoned. However, they are resilient precisely because their underlying governing structure remains intact. Tis resilience is problematic because, as Dotson argues, it can contribute to epistemic oppression, as is the case when an epistemic agent is unable to share knowledge that seems impossible given the operative epistemological system (Dotson, 2012, p. 131). An example may help us further draw out what it means both to say that an epistemological system is resilient and that such a system can oppress. Let us turn then to white supremacy.

2.  White Supremacy as an Epistemological System In her pivotal work Te New Jim Crow, Michelle Alexander interrogates the racist ideology that developed under slavery and gave rise to Jim Crow laws. Tis ideology, she notes, remains largely unchanged. My aim here is to argue that the racist ideology under investigation by Alexander, that of white supremacy, is a paradigm case of a resilient epistemological system.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

What Lies Beneath  83 Alexander tracks the evolution of white supremacy from chattel slavery to Jim Crow to mass incarceration. I will retrace her steps here so as to better understand how white supremacy has adapted to survive changing legal, political, and social contexts. Doing so is the key to understanding how white supremacy functions as a resilient epistemological system.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.1  A History of White Supremacy Te “essential function” of white supremacy is to uphold racial hierarchies that position whites as the dominant and superior race. I argue that white supremacy is only able to achieve this function to the extent that it is an epistemological system. White supremacy allows for those who enact, uphold, and maintain its policies to believe in the superiority of the white race by constructing a world in which this appears to be true. White supremacy is a multifaceted system of domination that encompasses a number of dimensions, ranging from the political and economic to the cultural and cognitive. On the cognitive component, Charles Mills writes that white supremacy is a system that will “have a negative efect on the consciousness of both whites and nonwhites, shaping both their descriptive and evaluative conceptualizations of the world,” with one of those negative efects being that “whites will tend to develop theories that justify their position, both morally and in terms of alleged facts about reality” (Mills, 2003, pp. 276–7).9 We can most clearly see this in the evolution of white supremacy following the Civil War. As Alexander notes, the abolition of slavery was followed by the apparent loss of racial order. Suddenly, hundreds of thousands of slaves, previously confned to plantations, were free to roam. It is here we see the emergence of vagrancy acts and “Black codes,” which would serve as a bridge between slavery and the eventual development of Jim Crow. Vagrancy acts and Black codes were enacted to compel Blacks to gain lawful employment, as, according to whites, they “lacked the proper motivation to work” (Alexander, 2010, p. 28). Tese laws not only made it a criminal ofense to be without work, but also permitted the hiring-­out of convicts to plantations. Consequently, with few to no job prospects readily available for freed Blacks, these vagrancy laws functioned more or less as another form of forced free labor. Of course, as Alexander notes, these laws were eventually overturned. But, on their heels was Jim Crow, which essentially functioned as a state-­sanctioned form of terror that allowed for the legal enforcement of the separation of races. In many states, Jim Crow enshrined in law practices that previously had been covert, like the hiring-­out of convicts, as well as the targeting and aggressive enforcement of criminal ofenses of Blacks. 9  Tere are a number of important connections to be drawn here between this project and work in the epistemology of ignorance that focuses on the role of white ignorance in the maintenance of racial hierarchies. See Mills, 2007.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

84  Briana Toole

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

However, following the success of the Civil Rights Movement in the early 1960s and the collapse of Jim Crow, white supremacy shifed instead to a focus on crime. Te seeds of this new system had been planted early under the guise of “law and order” initiatives, which merely served to “generate and mobilize white opposition to the Civil Rights Movement” ( Alexander, 2010, p. 40). It is through the call for “law and order” that we see the birth of mass incarceration. Trough “tough on crime” rhetoric that criminalized problems in Black communities—from the “welfare queen” to the War on Drugs’—a new system of racialized control emerged that efectively maintained segregation. In part, the transition from slavery to Jim Crow to mass incarceration was motivated by a cultivated fear among whites that Blacks were too unruly, aggressive, and dangerous to move among them unconstrained. In each case, what “justifes” the move from one form of racialized control to another is a white supremacist epistemological system that ultimately views Blacks as so morally and intellectually inferior that they require supervision and regulation. It is here that we can see the three elements of an epistemological system emerge. A white supremacist epistemological system licenses the belief that poverty in Black communities is due to natural inferiority rather than systemic structures that have denied Blacks access to education and employment, health care and housing, and other opportunities. Moreover, such a system primes epistemic agents to see, and in turn believe, that Blacks are holding guns even when they possess only innocuous objects, like a wallet or a cell phone (Payne, 2001, 2006). Lastly, this system ensures that we are more likely to attend to Black wrongdoing, even when whites commit crimes in equal or greater degree (Alexander, 2010, pp. 98–100, 106).10 White supremacy, as a racial ideology, is a method of social control that succeeds, I argue, by being so well disguised that it is nearly invisible. As such, white supremacy can be difcult to fully identify or describe. Tis is, in efect, the source of its resilience. As such, let me turn now to the task of drawing out the features of a resilient epistemological system.

3.  White Supremacy as a Resilient Epistemological System I suggested above that white supremacy is an epistemological system. But, my goal here is to argue that not only is white supremacy an epistemological system, but that it is a resilient epistemological system. As Alexander herself notes regarding the resilience of white supremacy and the systems of control it enacts, it has “become perfected, arguably more resilient to challenge, and thus capable of enduring for generations to come” (Alexander, 2010, p. 22). Grounding my analysis in the account of white supremacy ofered above, I identify fve features of a resilient epistemological system. A resilient epistemological system

10  Additionally, consider the research on teacher gaze—the phenomenon in which teachers gaze longer at Black students. See Section 3.2 for more on this.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

What Lies Beneath  85 1. consists of a core set of foundational governing beliefs, values, epistemic norms, and resources that: 2. play a central role in structuring our understanding of and engagement with the world; 3. is self-­masking; 4. is self-­replicating; 5. and silences contrary or dissenting views. I will argue below that white supremacy satisfes these conditions, and as such, I take it be a clear illustration of a resilient epistemological system.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.1  Core Beliefs and Values First, white supremacy consists of a core set of beliefs that remain largely unchanged in content and character despite numerous changes in the social and political context. Tese core beliefs are such that they naturalize racial injustice by appealing to beliefs about the inherent inferiority/superiority of certain races. What are the core beliefs that govern white supremacy? Chief among the beliefs in a white supremacist epistemological system is the belief in the inherent superiority of the white race and the inferiority of the Black race. Additionally, as noted above, is the belief that Blacks are by their nature lazy, dangerous, menacing, and predisposed to criminality. Still further, a white supremacist epistemological system tends to cast Black men as violent and overly aggressive, and Black women as hyper-­sexual (Crenshaw, 1994). Tese latter beliefs (that Blacks are violent and promiscuous) mostly function so as to provide justifcation for and to naturalize the claim that whites are inherently superior. Such systems also infuence the available epistemic resources and norms. Tese resources and norms will in turn infuence what beliefs we go on to adopt or reject and what social distinctions we notice or ignore. Te very concept of “race” is itself an epistemic resource devised to justify the practice of slavery (Alexander,  2010, p. 23; see also Fredrickson, 2002). Te emergence of this resource has produced the consequent belief that race is biological and that there are biological diferences between the races with respect to intellect, culture, and morals. Tese core beliefs make a diference to what epistemic agents embedded in this system are in a position to know about the social world.

3.2  Structuring Understanding Te core beliefs of a resilient epistemological system will, in large part, determine which beliefs are rationally permissible. In the case of white supremacy, the epistemic norms and resources developed under this system function so as to direct our attention such that the knowledge we gather serves to reinforce these core beliefs. For instance, we may develop heightened attention to the wrongdoing of Blacks,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

86  Briana Toole which in turn serves to reinforce the belief in the criminality and aggression of Blacks. Take as an example Reagan’s “war on drugs.” As Alexander writes

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Tere is no truth to the notion that the War on Drugs was launched in response to crack cocaine. President Ronald Reagan ofcially announced the current drug war in 1982, before crack became an issue in the media or a crisis in poor Black neighborhoods. . . . Almost overnight, the media was saturated with images of Black “crack whores,” “crack dealers,” and “crack babies”—images that seemed to confrm the worse negative racial stereotypes about impoverished inner-­city residents (Alexander, 2010, p. 5, italics mine)

Yet, despite the over-­attentiveness of the media to crack in the Black community, it remains the case that whites are more likely than Blacks to engage in drug crime (Alexander, 2010, pp. 7, 98–100). Tus, given the ubiquity of negative attention and portrayals of Blacks in the media, one might (almost) be forgiven for believing that Blacks are inherently criminalistic. Tere is also evidence that primary school teachers are subject to a phenomenon known as “teacher gaze,” in which they gaze longer at Black students in expectation that they will be more disruptive (Gilliam et al., 2016). Teachers are thus more likely to catch Black students engaged in wrong-­doing, even while white students are engaging in the same or similar behaviors. To some extent, one must wonder the extent to which this phenomenon contributes to the school-­to-­prison pipeline. In turn, one could argue that just as vagrancy laws bridged chattel slavery and Jim Crow, the school-­to-­prison pipeline serves a similar link between Jim Crow and the prison-­ industrial complex. Still further, as Lauren Woomer has observed, despite the abundant evidence of police brutality in Black communities, whites are overwhelmingly unlikely to believe that police brutality is a severe issue. Rather than seeing police brutality as the result of systemic racism in the criminal justice system, whites are more likely to claim that instances of police brutality are either “isolated incidents,” were justifed, or could have been avoided if the Black actors involved had complied (Woomer, 2017).11 Moreover, as Gaile Pohlhaus (2011) and Miranda Fricker (1999) have each noted, epistemic resources—the conceptual tools and language used for understanding and communicating our experiences—are largely shaped by the socially dominant. Consequently, the epistemic resources that are readily available are better-­suited to understanding a “white world,” but largely fail at capturing those experiences shared by those at the social margins. As such, our conceptual repertoire is altogether inadequate to attending to the oppression produced by white supremacy because those resources were not designed to be able to serve this function.12 11  Consider as well that these core beliefs structure our understanding such that we are not easily able to admit of counterexamples. Consider, for instance, that talented Black intellectuals tend to be viewed as an exception to the norm of Black inferiority, rather than evidence that requires a revision of that norm. 12  One need only look at the treatment of concepts designed to express the experiences of the socially marginalized to see this. Consider, as one illustration, the ill-­treatment of the concept of “microaggressions.”

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

What Lies Beneath  87 Returning to the example above, the concept of race, along with the belief that race is biological, is so pervasive that even our medical doctors are prey to treating Black and white patients diferently on the basis of race alone. In a study published in the Proceedings in the National Academy of Sciences, researchers found that doctors believed (1) that Blacks have thicker skin than whites, (2) that Blacks are biologically more resistant to pain, and (3) consequently, doctors under-­prescribed medication to Black patients even when they reported the same levels of pain as white patients (Hofman et al., 2016, p. 4296). One might naturally wonder how such a system persists, even and perhaps especially among people who might not endorse (or disavow altogether) racist attitudes. Part of this is owed to the fact, I believe, that resilient epistemological systems are self-­masking.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.3  Self-­masking Self-­masking is a feature of a resilient epistemological system that allows for such a system to ofer the appearance of reform when it has been merely redesigned. Te capacity to self-­ mask is, I suggest, one of the defning features of a resilient epistemological system. It is precisely this features that allows for an epistemological system to be resilient. Self-­masking allows for an epistemological system to survive in a world that is continually calling for its demise. As we have seen, white supremacy adapts to meet accepted standards of the period without actually changing the underlying principal beliefs that govern it. As such, the core set of beliefs remain largely undisturbed. For instance, the transition from Jim Crow to mass incarceration via the school-­to-­prison pipeline mirrors the move from chattel slavery to Jim Crow by way of vagrancy laws. In the examples of both the vagrancy laws and the school-­to-­prison pipeline, we have what looks to be social progress—the abolition of slavery and later, the decision in Brown v. Board of Education to desegregate schools, efectively signaling the end of the Jim Crow era. However, each gave way to yet another form of race-­based social control. One way by which white supremacy masks itself is through the advocation of “color-­blindness.”13 As Jose Medina observes, “the disavowal of racialized . . . perception involves distancing oneself from the social reality of racism . . . and failing to properly acknowledge [its] infuence on social cognition” (Medina, 2013, p. 27). In this way, color-­blindness has an important role to play in the maintenance of white supremacy. As Charles Mills writes, the strategy of colorblindness allows “the white delusion of racial superiority [to insulate] itself against refutation” (Mills, 2007, p. 19). Michelle Alexander echoes this sentiment, writing that “in the era of color-­blindness, it is no longer socially permissible to use race, explicitly as a justifcation for 13  I suggest that resilient epistemological systems employ a number of methods to insulate themselves from criticism. With regard to white supremacy, self-­masking occurs via linguistic hijacking (Anderson ms), epistemic misdirection and appropriation (Davis, 2018), and by actively maintaining structures of ignorance (Alcof, 2007; Mills, 2007; Woomer, 2017).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

88  Briana Toole discrimination” (Alexander, 2010, p. 2). And so, we use the language of criminality instead. In hiding behind the linguistic turn from “race” to “criminality” we remain able to enact white supremacy by couching it in the socially acceptable language of “law and order.” In fact, the core belief of white supremacy—the notion that there is such a thing as “race,” at all—is itself a self-­masking one. As Charles Mills writes in discussing constructivists accounts of race, racial categories “do not pre-­exist white supremacy as natural kinds, but are categories and realities themselves brought into existence by the institutionalization of the system” (Mills,  2003, p. 371). Tat this is a category that was devised for the sole purpose of procuring a ready supply of cheap labor is masked by the more recent belief that white supremacy emerged as a natural result of the moral, cultural and intellectual superiority of the white “race” (Fredrickson, 2002, p. 29; Cornell and Hartmann, 2007, p. 23). Tus, the notion that race is biological is designed to mask the more economic origins of the concept and to justify its continued usage. To elaborate on this point, let me turn now to the fourth feature of a resilient epistemological system: its capacity for self-­replication.

3.4  Self-­replication

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Te self-­masking feature of a resilient epistemological system is aided and abetted by a system’s capacity for self-­replication. Self-­replication refers to the recursive, self-­ sustaining dynamic of an epistemological system to ensure its own reproduction. Self-­ replication, in sum, ensures the replication of an epistemological system’s frameworks and ways of thinking. As it applies to our case study, the capacity to reproduce itself through seemingly innocuous means and under numerous guises has created an enduring legacy of white supremacist mindsets in our schools, policies, and legal structures. As has been argued by noted literary scholar Donnarae MacCann, Cultural and social historians have a useful tool in the record created by children’s books. Te simple, transparent images contrived for the young are ofen an unselfconscious distillation of a national consensus or a national debate. Tey reveal, for example, the degree to which postbellum society retained features of the slavery era; they illustrate how the white supremacy myth infected the mainstream collective consciousness in both [antellbellum and postbellum] epochs. (MacCann, 2002, p. xiii)

Still further, as MacCann writes, “the myth of white superiority was introduced into each successive generation’s social conditioning, and the very act of passing down white supremacist attitudes to children tells us much about the importance of this myth to the child-­raisers” ( MacCann, 2002, p. 233). Te passing down of the white supremacist myth from generation to generation serves two functions. First, it represents white superiority as factive, universal, and natural. Tis representation of white superiority as refecting natural racial hierarchies

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

What Lies Beneath  89 allows for the white experience to be so thoroughly embedded as normative “that its normativity is not even identifed as such” (Mills,  1998, p. 10). Second, the white supremacist myth serves to naturalize white supremacy, ultimately rendering white supremacy both essential and fundamental, and thus not susceptible to challenge. Tis consequently thwarts understanding white supremacy and white normativity as refective of a particular historical context and as a contingent, rather than necessary, state of afairs. White supremacy is able to reproduce itself in this way largely because it is also able to silence dissenting views. Tis leads me to the fnal feature of resilient epistemology systems.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.5  Silences Dissenters A resilient epistemological system cannot survive merely by self-­masking and self-­ replicating—it must also “put down” any potential threats. I suggest that it achieves this function through silencing. Silencing, as it pertains to the maintenance of epistemological systems, is a two-­fold process. It both involves rendering oppositional views illegible and alienating participating members from those oppositional perspectives. Kristie Dotson defnes silencing as a form of epistemic violence in which a “given group’s ability to speak and be heard” is damaged (Dotson, 2011, p. 236). As Dotson argues, this occurs when we privilege existing epistemic practices and “disappear” alternative ways of understanding. Tus, for instance, if as I have argued, our existing epistemic practices are largely shaped by white supremacy, then any epistemic practice that conficts with this operative system may be misunderstood. As an illustration, consider that some hear the claim “Black Lives Matter” as “Only Black Lives Matter” and thus respond with the assertion that “All Lives Matter” (Anderson, 2017). Or consider instead that those who try to draw attention to racism are accused themselves of being racist for talking about race at all. In both cases, the ability to speak and be understood by one’s audience is undermined because the operative epistemological system—that is, white supremacy—makes such talk incomprehensible. Te unintelligibility of oppositional views then provides a basis for adherents to dismiss the credibility or reliability of those espousing contrary views. In this respect, resilient epistemological systems function much like echo chambers. C. Ti Nguyen (2018) defnes an echo chamber as “a social epistemic structure in which other relevant voices have been actively discredited” (Nguyen,  2018, p. 2). As Nguyen writes, echo chambers systematically isolate members from outside epistemic resources, such that the chamber cannot simply be “popped” by exposure to outside information. Echo chambers achieve this by undermining the trustworthiness of those who espouse contrary views, as well as providing counter-­explanations of these contrary views such that the core beliefs of the system are reinforced. To illustrate, consider that a core belief of white supremacy is that Blacks are inherently dishonest or unreliable. As a consequence, any social critique made by

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

90  Briana Toole Blacks is taken merely as further evidence of their unreliability.14 Tus, one need neither take seriously the testimony of Blacks regarding the systemic structures of racialized oppression they experience, nor critically evaluate whether one is living in such a system. By preemptively assigning credibility defcits to those who are most likely to challenge the white supremacist status quo, white supremacy can insulate itself from challenge, criticism, and condemnation. What is especially interesting about the role of silencing in the maintenance of a resilient epistemological system is its “looping” efect. Essentially, we see the use of a frst-­order epistemic exclusion (that we distrust the testimony of Blacks) that enables and is enabled by some feature of the operative epistemological system (the credibility defcit preemptively assigned to Blacks under white supremacy). As I hope is becoming clear, the three orders of epistemic exclusion that Dotson delineates, and the fve features of resilient epistemological systems that I have identifed here, operate cyclically. First- and second-­order epistemic exclusions support third-­order ones, which in turn support those frst- and second-­order exclusions. In much the same way, the core beliefs of a resilient epistemological system seem to support the justifable exclusion of dissenting points of views, which in turn reinforces the core beliefs of that system.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.  Circling Back: Resilient Epistemological Systems and Tird-­Order Epistemic Exclusions I prefaced this chapter with a question: what would it take to change an epistemological system? To some extent, my goal has not been to ofer an answer so much as it has been to show that, in order to provide one, we must frst fully understand what it is that we are endeavoring to change. As I noted, epistemological systems can be modifed through frst-­order revisions—by making one’s behavior refect one’s values—or second-­ order changes—by identifying and addressing gaps in one’s operative resources. However, a resilient epistemological system may preserve and legitimize inadequate resources such that the underlying conditions of the epistemological system—those features that I identify in Section 3 above—remain largely intact. Such a system then yields third-­order exclusions—a form of epistemic oppression in which an individual’s knowledge “may seem impossible given the state of the operative epistemological system” (Dotson, 2014, p. 131). As an illustration of such an exclusion, let us return once more to our discussion of police brutality. Tat police brutality is a direct result of systemic racism—enabled by white supremacy as an epistemic framework—is dismissed in the ways noted in Section 3.2 above—as an isolated incident that is the result of one “bad apple” rather than a practice that has been shaped by anti-­Black racism. Consequently, demands that policing be radically reconceived—either by abolishing or defunding the police—are met with skepticism. Many are inclined to argue, as Alex Vitale writes, 14  Or, consider that white pro-­Black activists are dismissed as self-­hating race traitors. Tus, their actions are understood not as being motivated by the fact that Blacks are discriminated against, but by their internalized white guilt.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

What Lies Beneath  91 “that racist and brutal cops can be purged from the profession and an unbiased system of law enforcement reestablished in the interest of the whole society” (Vitale, 2017). But this ignores the historical origins of policing as rooted in white supremacy. As Vitale goes on to argue, policing emerged as a form of modern social control in response to new economic developments—in the U.S., that economic development was slavery (Bayley, 1998; Vitale, 2017). Te view of police as “dispassionate enforcers of the law” cannot be reconciled with the understanding of police as agents who enforce unjust racial hierarchies.15 Resistance to this conceptualization of the police does not stem from a distrust of testimony (a frst-­order exclusion), as we have seen more and more that people are willing to acknowledge the excessive force with which police ofcers treat Blacks. Nor is it the result of inadequate resources for understanding (a second-­order exclusion), as the epistemic interventions that would allow us to understand police brutality are in place. Such knowledge instead seems impossible because it conficts with our understanding of police ofcers as basically good—an understanding itself provided by white supremacy. Tis resistance is thus the product of the resilience of white supremacy as an epistemological system. Dotson argues that when an individual is “confronted with the epistemological resilience of a maladjusted system . . . [her] epistemic agency is compromised by being rendered incapable of contributing to the domains of inquiry relevant to her insight” (Dotson,  2014, p. 130). Tus, we see that the knowledge contributions of Black activists—pathways for the abolition and defunding of the police—have struggled to be taken up—resulting in a third-­ order epistemic exclusion, one that I argue is owed to the resilient epistemological system that is white supremacy.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

5.  Te “New” New Jim Crow? In the previous sections, I provided an analysis of white supremacy in order to deconstruct the features of a resilient epistemological system. I asserted that a resilient epistemological system consists in the following fve features: (1) it has a core set of beliefs that (2) structure our understanding, and it is (3) self-­masking, (4) self-­ replicating, and (5) it preemptively silences would-­be dissenters. Tough I remain neutral as to whether these features are individually necessary and jointly sufcient for constituting a resilient epistemological system, I strongly suspect that those operative epistemological systems that might strike us as resilient are likely to present all fve features.16 Consider, for instance, that this account will enable us to analyze other operative, and maladaptive, resilient epistemological systems like patriarchy or cults, systems which surely possess all the features described above.

15  Essentially, I am suggesting that police brutality against Blacks is a feature, not a bug, of policing. 16 It is just as likely, however, that resilient epistemological systems function more like family, or “cluster-­concept,” such that a system that is resilient might present any three or more of the features identifed.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

92  Briana Toole I argued that white supremacy satisfes these conditions, and as such, I take it to be a clear illustration of a resilient epistemological system. Understanding white supremacy in this manner is essential if we seek to dismantle it. In treating white supremacy as the mere manifestation of social inequity we have acted as if it can be eradicated by eliminating certain unjust social systems (e.g. segregation). But if, at its core, white supremacy is something more, then addressing unjust social systems is little more than a symbolic change, one which leaves the underlying cause unchanged. In sum, in failing to understand the epistemic dimensions of white supremacy, we have treated the symptoms of a corrupt system without striking at the root cause. What lies beneath the social, cultural, political, and economic forms of white supremacy that we are more familiar with is the epistemic form that makes those possible. Understanding that white supremacy is not merely a political system, but an epistemological one, positions us to see that it still pervades the American social, political, and legal landscape, and will continue to do so, so long as it pervades our epistemic systems. White supremacy, I have argued, is embedded in our thinking—it is the flter through which we see, the starting point from which we believe, the frame through which we understand, interpret, and interact with our world. In unpacking the epistemological roots of white supremacy, in exploring the evolution of white supremacy, my hope is that we can do three things. First, the analysis ofered here provides us with the necessary tools for identifying white supremacist attitudes in our thinking. As bell hooks (2004) and others have noted, naming the problem is the frst step in addressing the problem. Knowing how white supremacy manifests in our thinking, how it reproduces itself in our legal and medical institutions, how it silences dissenting viewpoints, allows us to recognize what we could not before. Second, if the diagnosis that I ofer within these pages is correct, it provides a direction for how we can move forward. Te resistance we have seen to discussions regarding issues ranging from police brutality to the prison-­industrial complex, may not be owed to racist individuals, but to racist epistemological frameworks that distort our ability to understand these issues. What this tells us is that we must both draw attention to the inadequacy of this framework, and strive to provide alternative epistemic frames through which to see and understand the world, ones not shaped by anti-­Black racist sentiment. Tird, and most importantly, this analysis may very well allow us to contemplate how white supremacy might continue to evolve. Alexander, in Te New Jim Crow, turned her gaze to the past—attending to the ways in which Jim Crow gave rise to the mass incarceration of Black Americans that we see today. Our task now is to look forward. With the account I have ofered here, I hope we can work to anticipate what might replace mass incarceration as the “new” new Jim Crow.

References Alcof, L. 2007. “Epistemologies of Ignorance: Tree Types.” In S. Sullivan and N. Tuana (eds.), Race and Epistemologies of Ignorance. New York: State University of New York Press, pp. 39–58.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

What Lies Beneath  93 Alexander, M. 2010. Te New Jim Crow: Mass Incarceration in the Age of Color-Blindness. New York: Te New Press. Anderson, D. (ms). “Linguistic Hijacking.” Anderson, L. 2017. “Hermeneutical Impasses.” Philosophical Topics 45(2): 1–19. Bayley, D. 1998. “Te Development of Modern Police.” In L Gaines (ed.), Policing Perspectives: An Anthology. Oxford: Oxford University Press. Cornell, S. and D. Hartmann. 2007. Ethnicity and Race: Making Identities in a Changing World, 2nd edition. Tousand Oaks, CA: Pine Forge Press. Crenshaw, K. 1994. “Te Marginalization of Sexual Violence against Black Women.” NCASA Journal 2(1): 1–15. Davis, E. 2018. “On Epistemic Appropriation.” Ethics 128: 702–27. Dotson, K. 2011. “Tracking Epistemic Violence, Tracking Practices of Silencing.” Hypatia 26(2): 236–57. Dotson, K. 2012. “How is this paper philosophy?” Comparative Philosophy 3(1): 3–29. Dotson, K. 2014. “Conceptualizing Epistemic Oppression.” Social Epistemology 28(22): 115–38. Dresser, R. 1992. “Wanted: Single, White Male for Medical Research.” Hastings Center Report 22: 24–9. Foucault, M. 1966. Te Order of Ting. New York: Pantheon Books. Fredrickson, G.  M. 2002. Racism: A Short History. Princeton, NJ: Princeton University Press. Fricker, M. 1999. “Epistemic Oppression and Epistemic Privilege.” Canadian Journal of Philosophy 29(suppl.): 191–210. Fricker, M. 2007. Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press. Gilliam, W., A. Maupin, C. Reyes, M. Accavitti, and F. Shic. 2016. “Do Early Educators’ Implicit Biases Regarding Sex and Race Relate to Behavior Expectations and Recommendations of Preschool Expulsions and Suspensions?” Yale Child Study Center. . Grasswick, H. 2004. “Individuals-in-Communities: Te Search for a Feminist Model of Epistemic Subjects.” Hypatia 19(3): 85–120. Haslanger, S. 2008. “Changing the Ideology and Culture of Philosophy: Not by Reason (Alone).” Hypatia 23(2): 210–23. Hofman, K., S. Trawalter, J. Axt, and N. Oliver. 2016. “Racial Bias in Pain Assessment and Treatment Recommendations, and False Beliefs about Biological Diferences between Blacks and Whites.” Proceedings of the National Academy of Sciences USA 113(16): 4296–301. hooks, b. 2004. Te Will to Change: Masculinity and Love. New York: Atria Books. Jones, K. 2002. “Te Politics of Credibility.” In L. Antony and C. Witt (eds.), A Mind of One’s Own: Feminist Essays on Reason and Objectivity. Boulder, CO: Westview Press, pp. 154–76. MacCann, D. 2002. White Supremacy in Children’s Literature: Characterizations of African Americans, 1830–1900. Abingdon: Taylor & Francis. Medina, J. 2013. Te Epistemology of Resistance: Gender and Racial Oppression, Epistemic Injustice, and Resistant Imaginations. Oxford: Oxford University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

94  Briana Toole

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Mills, C. 1998. Blackness Visible: Essays on Philosophy and Race. Ithaca, NY: Cornell University Press. Mills, C. 2003. “White Supremacy.” In T. Lott and J. Pittman (eds.), Blackwell Companions to Philosophy: A Companion to African-American Philosophy. Oxford: Blackwell, pp. 269–81. Mills, C. 2007. “White Ignorance,” in S.  Sullivan and N.  Tuana (eds.), Race and Epistemologies of Ignorance. Albany, NY: State University of New York Press, pp. 11–38. Nguyen, C. 2018. “Echo Chambers and Epistemic Bubbles.” Episteme: 1–21. Payne, K. 2001. “Prejudice and Perception: Te Role of Automatic and Controlled Processes in Misperceiving a Weapon.” Journal of Personality and Social Psychology 81(2): 181–92. Payne, K. 2006. “Weapon Bias: Split-Second Decisions and Unintended Stereotyping.” Current Directions in Psychological Science 15(6): 287–91. Pohlhaus, G. 2011. “Relational Knowing and Epistemic Injustice: Toward a Teory of Willful Hermeneutical Ignorance.” Hypatia 27(4): 715–35. Railton, P. 2006. “Normative Guidance.” In R.  Shafer-Landau (ed.), Oxford Studies in Metaethics, Volume 1, Oxford: Oxford University Press, pp. 3–34. Toole, B. 2019. “From Standpoint Epistemology to Epistemic Oppression.” Hypatia 39(4): 598–618. Vitale, A. 2017. Te End of Policing. London: Verso. Walker, B., C. Holling, S. Carpenter, and A. Kinzig. 2004. “Resilience, Adaptability and Transformability in Social–Ecological Systems.” Ecology and Society 9(2): 5. Woomer, L. 2017. “Agential Insensitivity and Socially Supported Ignorance.” Episteme: 1–19.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

PART II

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

EPIST E MIC PROBLE M S F OR DE MO C R ACY

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved. Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

6

Epistocratic Paternalism David Estlund

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1. Introduction Many people think that political democracy is important. And we all know the traditional puzzle about whether democracy is initially required even if those democratic voters would abolish democracy.1 In this aphoristic form, we see a pattern that I will explore in this chapter, though I will not concentrate on those stakes in particular. More generally, my question is how robust the requirement of democracy might be in the face of alternatives that would make much better decisions. In particular, suppose that there are alternatives in which some people are given more voting power on the ground that they are better able than others, to make good decisions. Some might doubt that any are better than others, though I am not one of them, and I will not consider that question. By supposing it might be true for the sake of argument, we can pointedly ask how weighty we should think a requirement of democracy is, and on what grounds. Te democratic idea that people ought to share equally in political rule is in stark tension with the fact that people will never be even roughly equal in their ability to contribute competently to the morally momentous decisions that governments make. Tis ancient challenge to democracy may not be fatal, but I do think that we need a better understanding of its power, if only to help in developing an answer that is clearer and more powerful than what we have so far. To do this properly, it is helpful to postpone the efort to give an answer, and take seriously the possibility that there will not be one. Tat said, I do not mean to concede that my own attempt (2008) at an answer is mistaken. I won’t pursue it here, though it comes up briefy toward the end. In this chapter, I hope to understand better how the prospect of superior rule by the wise, what I will call epistocracy, challenges a strong commitment to democracy. It might seem that the challenge can be put aside by arguing that the justifcation of democracy is based on wholly non-­instrumental normative considerations. Among non-­instrumental accounts, it is common to speak of a right to equal power over political decisions, or to equal or non-­hierarchical social and/or political arrangements. Such accounts ofen acknowledge that democracy might also have some instrumental value, but they seem to suggest that its justifcation is not contingent on that. If it were, the challenge from epistocracy would have to be more directly engaged. I will explain how an important set of non-­ instrumental accounts—accounts that 1  I do not know its source, and it can be formulated in slightly diferent ways. David Estlund, Epistocratic Paternalism. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © David Estlund. DOI: 10.1093/oso/9780192893338.003.0007

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

98  David Estlund emphasize a non-­instrumental claim against being asymmetrically ruled by others— needs to address the epistemic dimension. I think this can be exhibited by considering an analogy, imperfect as analogies are, with the common view about paternalism, broadly conceived. Even if a right against paternalism is grounded non-­ instrumentally on the individual’s claim of a kind of sovereignty, that right is limited by certain considerations about the person’s competence. In the absence of the relevant competence the right against paternalism does not apply, or is overridden. Putting it very generally for now, if the person is sufciently bad over a certain range of decisions, and some external agent is sufciently capable of avoiding serious errors by interfering, interference is sometimes permitted on that basis. Te case of children is the obvious example, and there may be others, though this is somewhat controversial. Te right against paternalism is not, it is fair to say, grounded in epistemic considerations, but the question of the agent’s epistemic competence is nevertheless a crucial part of the account. Some accounts of the requirement of democracy ground the requirement in some variety of a right to collective self-­governance. Call these sovereignty accounts. Te analogy with paternalism helps us to see how even sovereignty based justifcations of democracy do not obviously avoid the need to defend democracy’s epistemic value, at least as compared with the prospect of what we might call epistocratic paternalism. Tat is the rough idea, which I now go on to lay out in a less condensed way. From there I will go on to extend the point beyond sovereignty accounts to those emphasizing a right to equal power over political decisions, or a right against social hierarchy. Tere is a sort of Holy Grail in the tradition of democratic theory, namely to show how properly democratic arrangements, as such, can avoid people being ruled against their will. Rousseau famously attempted to square this circle by way of what he called the “general will.” Te idea mentioned a moment ago, of “collective self-­ rule” evokes a similar aspiration. However, the Grail seems to me to remain safely hidden. Certainly, political rule need not be any form of slavery. However, we sometimes lazily say that the right against slavery is the right to be subject only to one’s own will. But that would also be a right against democracy. Democracy is not freedom from subjection to others, but a particular pattern of subjection: each is subject to the authority and power of the citizens collectively, and so mostly to the aggregated wills of other people. Democracy is roughly a case of symmetrical legal subjection (Viehof, 2014, p. 18).2 I say, “roughly,” since democracy might entail more than this. Since democracy involves subjection it must still be given a justifcation, which may in some way stem from the subjection’s symmetry. Unless there is a right against democracy, then there must not be a right against subjection itself. However, 2  Viehof (2014) uses the term “nonsubjection” to characterize the absence of hierarchy in a relationship, and might mean nothing more by it. But I am suggesting that subjection can be symmetrical, marking a big diference from the genuine absence of subjection. Arthur Ripstein (2009) also places what he calls “nonsubjection” at the center of his development of a Kantian political philosophy, with subjection consisting in the usurping or destruction of one’s power to set ends. Whether or not that is avoided in a proper legal order, each remains subject to some or all of the others rather than to no one but oneself. Te term “nonsubjection” risks confating those two conditions, and the diference seems to be morally important.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Epistocratic Paternalism  99 if there is a right to democracy it may be, in part, a right against asymmetrical legal subjection in particular. As I have said, paternalism can be justifed in special cases by what I will call a competence gap, as in the case of children, where intervention would tend to avoid sufciently serious mistakes. But then, why not also asymmetrical legal subjection? Why could it not be justifed in a similar way, when privileging the power of those who know better would remedy a sufciently pronounced incompetence by the others? Tat suggests a way of justifying epistocracy. It would be a form of asymmetrical subjection, and the analogy with paternalism brings into engagement considerations that are ofen seen as pointing in diferent directions in democratic theory: a right to equal or symmetrical say over political decisions, on one hand, and the importance of an epistemic dimension—the substantive quality of the procedure’s decisions—on the other.3 I will not consider the general question of how such a case for epistocracy might be answered, but just to avoid misunderstanding, I am by no means suggesting that it can’t. I am just arguing that it must. Te central case of paternalism is interference with someone’s choices with a rationale of promoting or protecting the interests of the interfered agent in cases of their self-­regarding action. Classic anti-­paternalism is, roughly, the position that such interference tends to be wrong. I will assume, with the classic view, that other-­ regarding actions will be less insulated from interference, and that will play a big role in what is to come. Te epistemic justifcation for paternalism in the political case might seem to be especially plausible, since political decisions are profoundly other-­regarding. Viehof (2014, p. 350) essentially makes this point: I might have an interest in contributing to our shaping our collective life through voting, but I also have an interest in shaping my own life and so being immune from the decisions of other voters. In one respect, my ambition here is modest, arguing in the frst instance against an extreme view: an absolute or extremely weighty requirement of democracy. Te challenge is simple: what if some non-­democratic arrangement—focusing on the case of epistocracy—would produce profoundly better and even more just outcomes, or even a more just society? An absolutist requirement of democracy probably does not appeal to many theorists. But the pattern of that simple challenge shows that there is a serious question about how much weight a pro tanto requirement of democracy could plausibly have as against better performing alternatives such as, suppose, epistocracy. Toward the end I will show how the challenge fts a general pattern—perhaps one that can be answered—to a broad class of views according to which democracy is required as a necessary constituent or condition of one or another kind of moral, social, or political equality.

3  Te importance of the quality of the decisions is better called the instrumental dimension, but we will be thinking of the epistemic subset: where better decisions are produced by people intentionally fguring them out. A system of asymmetrical subjection might have special instrumental value for other reasons, such as certain kinds of administrative efciency. Much of what I say in this chapter will have some application to those cases too, though I focus on the epistocratic version.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

100  David Estlund

2.  Paternalism and the Improvement Rationale I am defning “paternalism” broadly for present purposes. In the most commonly discussed cases, interference with an agent’s actions aims to promote her own interests. But I am construing paternalism—in alignment with Shifrin’s (2000) infuential account—to include interference aimed at producing improvement, from a moral point of view, over what the agent would have decided. Tere is also the overarching standard of what we might call rational improvement—improvement from the standpoint of all the reasons that apply to the interfered agent, which I  assume would include prudential and moral reasons among others. In that way, interference for the sake of moral improvement could be subsumed by the rationale of rational improvement. I will put this even broader aspect aside and focus on prudential or moral improvement as follows:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Improvement rationale: An improvement rationale for interfering with an agent’s choices takes, as a motivating reason for such interference, the (supposed) fact that such interference would avoid prudential or moral mistakes by that agent.

People ofen have a right to act without interference, at least so long as their action does not involve others in certain ways, such as risking harm to them or obstructing their legitimate spheres of control. Te relevance of harm to the question of a right against interference is contested. Since there are cases such as harmless trespass that most people believe may be regulated by law, the idea that only harmful or dangerous acts may be regulated would not explain it (Ripstein,  2006). An alternative is to emphasize acts that interfere with the freedom of others to use their means and powers as they see ft. Accordingly, I will use the idea of “self-­regarding” versus “other-­regarding” acts in a fexible way, leaving open whether it is best flled out according to harming others, and/or according to violating their individual sovereignty—two diferent ways in which an act might be, so to speak, with regard to others. Self-­regarding actions, then, should be understood for our purposes as those that are not other-­regarding in either of those ways. It will be helpful to stipulate what I will call, Te Noninterference Principle: A person’s self-­regarding choices, presumptively (or pro tanto) may not be interfered with.

Tis may be too simple in a number of ways, but having it in place will allow us to consider the issues that are my main concern. My question is whether and when an improvement rationale can permit departures from symmetrical legal subjection—much as an improvement rationale can sometimes permit paternalism. I will call that epistocratic paternalism. Te counterpart in the case of ordinary or legal paternalism would be the question when, if ever, the improvement rationale justifes interference by law in individual choices. One way of thinking about the right against paternalism is to hold that the improvement rationale can itself make wrong some interference that would not be

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistocratic Paternalism  101 wrong if not for that particular motive. If that rationale is regarded as wrong-­making, we can speak of strong anti-­paternalism. However, there is also another strand in our anti-­paternalist thought, namely that the improvement motive, which might seem to be a permitting rationale for interference, is not. Tat is obviously diferent from thinking that the presence of the improvement rationale is wrong-­making itself. Instead, what is wrong is the interference, not interference from the motive of improvement. Tis is nothing more than to deny that there is an exception of this kind to the noninterference principle. I will call this,

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Weak anti-­paternalism: Even avoiding mistakes of the interfered agent—thereby promoting the agent’s own prudential interests or moral reasons or imperatives—is (pro tanto) not necessarily enough to justify interference with an agent’s choices.

For my purposes in this chapter, it is weak anti-­paternalism that I am interested in. I emphasize the apparent failure, at least ofen, of the improvement rationale to do its supposed justifying work. On the other hand—and I am making important use of this—interference in an individual’s choices from the motive of an improvement rationale is not always wrong, as we have seen. For one thing, young children may be interfered with, at least by certain agents. More generally, when the interfered agent is below some competence threshold, and interference would to a sufcient degree avoid serious error, paternalism is sometimes justifed. Call this the competence gap proviso to the noninterference principle. Despite the compact name I am giving it, the exception case is not simply the diference between the competences of the interfered and interfering agents. It is not as if one agent is permitted to interfere in the choices of a quite competent agent on the ground that the interfering agent is even better. Tere is plausibly also a threshold of competence of the interfered agent above which the proviso does not kick in even if the potential interfering agent could do much better. So this is the “sufcient competence gap,” or competence gap, for short. Te question these observations are meant to set up is this: If a sufcient competence gap can sometimes justify paternalism, can a competence gap sometimes justify epistocracy? Tis is not the ancient question whether some epistocracy is instrumentally better than democracy. Even if that were so there might be a right of some weight against it. Our question is, rather: even if there is a right of some weight against epistocracy, could it be outweighed by a sufcient competence gap?

3.  Collective Self-­governance Te most direct analogy between paternalism and epistocracy would posit a collectivity of all subjects somehow organized into a single genuine moral agent, with its own right to make choices without interference from others. Epistocratic interference would be interference by other agents, possibly from amongst the members of the collective agent—which would still make them other agents on this view—in the collective agent’s political decisions, on the rationale that these others could do better than the collective agent itself. Tis collective agent might be held to

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

102  David Estlund have a right to be wholly self-­governing. Te objection to epistocracy might naturally arise, then, that it is a case of impermissible paternalism—at least in our extended sense of interference in another agent’s choices on the basis of an improvement rationale. I will not spend much time on this proposal, but limit myself to a few remarks about this vivid but unviable approach, the idea of the collective as an agent with rights of its own, including a right to self-­governance. 1. Tis strong collective agent view does not correspond, I think, to any currently advocated philosophical account of the requirement or justifcation of democracy. 2. It is certainly recognizable, perhaps mainly in the form of what I regard as erroneous interpretations of Rousseau’s doctrine of the General Will. Space prevents me from discussing that matter in any detail, but it should be sufcient here to point out what ought to be uncontroversial, namely that Rousseau’s concern about how political authority can be reconciled with freedom is about the freedom of individuals, not the freedom of the community as such. Afer all, as Rousseau (2019) says:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

“To fnd a form of association that will defend and protect the person and goods of each associate with the full common force, and by means of which each, uniting with all, nevertheless obey only himself and remain as free as before.” Tis is the fundamental problem to which the social contract provides the solution. (Rousseau, 2019, p. 50)

3. While the possibility of a genuine collective agent (as distinct from the utility of such an idea in, for example, law) seems to remain an open question among philosophers, this would not be enough to explain how such a thing might also be a subject of moral concern in its own right, where this is not reducible to concern for entirely diferent agents, namely, the individual members. 4. Even if there were such a collective agent, and it literally governed itself as it had a right to do, this would not yet provide any basis for thinking the individual members are, in any respect, self-­governing. Rule of one agent—an individual, by another agent—a collective, does not posit any agent that governs itself. Te central normative question in the broadly liberal tradition has been how an individual might justifably be subject to political power or authority, and this collectivist gambit is silent on that matter. So, I go on now to extend the paternalism analogy to less directly analogous understandings of democracy and epistocracy.

4.  Symmetrical Subjection as a Baseline Consider an undemocratic political system in which some adults’ legal right to vote is less than that of others, or is denied outright, on the ostensible ground (which

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistocratic Paternalism  103 might be true) that doing so would be an improvement, morally or prudentially, in the political decisions. Tat form of epistocracy involves an improvement rationale, and that suggests a connection to issues about paternalism. But how does this vote-­ weighting count as interference at all, which is a standard element in the idea of paternalism?4 Unless equal voting is already a status quo that epistocracy would upset—indeed if epistocracy is even already the status quo—why think of equal voting, in particular, as a baseline: a condition by contrast with which epistocracy is something like an interference? Tere is a basis for doing so. Notice, frst, that there are two points of possible interference in our purview: Our main topic is the parallel between epistocracy and paternalism, where the epistocrats interfere—or something like it—in democratic choices. Call this epistocratic intervention. But for a moment consider the more standard case of interference by law in people’s choices, as when behaviors such as selling certain drugs are made illegal. Call this legal interference. To simplify, I will suppose that interference by law is sometimes permissible but only in the case of democratically made law. If so, what might it be about democracy that can render some interference in people’s choices permissible? It is notable, as I have said, that the democratic power of legal interference is symmetrical. In democracy, even though any given person is subject to the interference of the others, this is also true of everyone else. And this seems to be a possible key to its justifcation. I do not think it is obvious that there is a right against asymmetrical subjection. Presumably, we would want some moral explanation for such a right, and it is not immediately clear to me what it would be. But it is illuminating to consider that possibility, in order to see that even if there were a right against asymmetrical subjection, the threat of epistocracy is by no means sidelined by such a right. Te subjection relationship in democracy is, as I am calling it,

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Symmetrical Subjection: No one has fundamental legal power over another person that that person does not also have over them.

Te term “fundamental” is important for the following reason. Tere is no question that in any government some people will have more power over the law than others in certain ways, since not all legislating can practically be subject to a vote by all. It will also be unavoidable in any government to consult or even deputize certain people on the grounds of their expertise in certain areas—superior competence. In modern democracies, anyone with formal power over law or policy is either elected, or appointed by somebody elected or by another appointee. When we trace some ofcial’s formal role back through the appointments and elections this eventually comes to a stop. Call that the level of fundamental legal power. If the fundamental legal power consists in elections including equal universal sufrage, then for present purposes that is democracy rather than epistocracy. If it comes to a stop at some level in which some have more formal power than others on the grounds of their superior expertise other than the conventional denial of the vote to children, this is epistocracy. 4 Authors disagree about whether paternalism must consist in interference with another’s liberty. Shifrin (2000) denies it, but Gerald Dworkin (2020) builds it into the defnition.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

104  David Estlund

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

What is symmetrical in democracy, then, is fundamental legal subjection. Whether the extra legal powers of ofcials in a democracy renders legal subjection as a whole asymmetrical or not, I will simply be focusing on symmetrical and asymmetrical fundamental legal subjection. Voting at the fundamental level might not be the only form symmetrical subjection could take. If a subset of citizens were selected on a rotating basis to serve as legislators, this might also count as symmetrical subjection. Te symmetry would lie in the equal chance or opportunity to be a temporary legislator. I agree with Alex Guerrero (2014) that this is best understood as a form of democracy, even though it is not electoral, and I think it might be precisely for this reason—its having the form, at the fundamental level, of symmetrical subjection. Also, for that reason, even if, as in Guerrero’s version of such a “lottocracy,” the temporary legislators are given special epistemic training and resources, and even if the whole model is recommended on grounds of its epistemic advantages, it would not count as a form of epistocracy. Suppose there is a right against asymmetrical subjection. Tat could explain why some democratic law, still a kind of interference, is permissible, while no other law is. Tis is not yet an argument that such symmetry is the justifcation, but I continue with this hypothesis for now. If there is a right against asymmetrical subjection then symmetrical legal subjection can serve as a kind of normative baseline for evaluating deviations from it. In that sense, epistocracy would be a deviation from a rightful baseline condition of symmetrical subjection. Te structural similarity to paternalism is now pretty clear. A natural thought that is parallel to weak anti-­paternalism—the claim that the improvement rationale is not sufcient to justify interference with an individual’s choices—would be that the fact that epistocratic intervention would lead to an improvement in outcomes is not enough to justify the deviation from symmetrical subjection, though—we should keep in mind—with the right size and kind of competence gap it might be. Te epistocracy/paternalism parallel looks like this: Noninterference principle (refned): Each person has a right against legal interference with her choices, even in the case of an improvement rationale, though subject to some limit involving a sufcient competence gap. Democracy principle: Each person has a right against asymmetrical legal subjection, even in the case of an improvement rationale, though subject to some limit involving a sufcient competence gap.

5.  Epistocracy and the Competence Gap Having exhibited what I hope is a fruitful analogy, and one that might be used to support an objection to epistocracy on the ground that it is, or is akin to, impermissible paternalism, we see that to be a weak argument. One reason we have seen is that in the case of ordinary paternalism, it can be permissible in cases of a sufcient competence gap, and the same would plausibly be true in the case of epistocratic paternalism. A further compounding weakness of the critique of

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Epistocratic Paternalism  105 epistocracy as akin to paternalism is this: Te moral objection to legal paternalism is normally thought to stem from a strong claim individuals have to autonomy of a certain kind. But exercising one’s share of political power is to participate in the legal subjection of others, even if it is symmetrical. As I have argued, appeals to such things as “collective self-­government” should not lull us into thinking that a person’s vote is only self-­regarding, a case simply of governing oneself. When we drop the idea that the demos is a moral being with a right and capacity to rule itself, we see instead that democracy involves people ruling each other. Te claim that each might have, against interference, to exercise their share of that ruling power is far less strong than it is in cases that are not so momentously other-­regarding. So, as the right against paternalism has limits based on a competence gap, it would seem at least as plausible that the right against asymmetrical subjection must have a similar limit. Te one does not prove the other, since the disanalogies might make the diference. However, the case of children may seem to prove it in this case as it does in the case of individual paternalism. Why else do we believe that children do not have a right to vote? Te competence gap may explain it: the fact that they are below a pertinent level of competence to vote well (in the interests of themselves or others), combined with the fact that there would be a sufcient improvement under the alternative of adult-­only sufrage. If so, then there would be an important question: what is the threshold and the sufcient superiority such that the right is overridden? Granted, the exclusion of children is a special case. Everyone begins as a child, and most would receive the vote upon maturity. Measured across a person’s lifetime the disenfranchisement of children does not treat any group of people diferently from others. Nevertheless, at any given time it gives only some people a fully weighted vote, excluding children. Giving children the vote would be less exclusive in this obvious way. Maybe the need to justify the exclusion is more salient if we imagine that the right to vote is not granted until the age of ffy. Or rescinded at sixty. Tese would still treat all equally over a lifetime. But their unequal treatment at every actual time requires some justifcation, some explanation of the diference between those of certain ages and others, that makes the diference in the right to vote. Te denial of the franchise to children is not the kind of deviation from equal sufrage we are mainly wondering about, and so it would be unhelpful to defne that as a form of epistocracy despite some similarity. Rather, let us defne epistocracy, with intentional looseness, as follows: Epistocracy: A political system in which law and policy is directly or indirectly authorized by its subjects, but where there is an unconventionally high competence threshold for the fullest right to vote.

Roughly, we want to focus on a competence threshold so high that it would strike the reader as intuitively undemocratic, and it is fne to leave that fuzzy threshold up to the reader. I speak of the fullest right to vote in order to count a system of plural voting, where some have more votes than others, as epistocratic even though everyone gets a vote of some weight. Tis defnition also makes clear that we are not limiting our concern to cases in which full voting power is limited to a small elite,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

106  David Estlund despite what the verbal form “‑ocracy” might suggest.5 A system of competence-­driven plural voting would count as epistocratic even if only a small fraction of (normal adult) voters were denied full voting power on such grounds as education or intelligence. If epistocracy is objectionable, it is not because it proposes a competence threshold on the right to vote. Tat is common ground as proven by the case of children. So the central question about epistocracy is this: On what grounds is the conventional competence threshold (that is, that conventional range) justifed as a limit on the right against asymmetrical subjection, whereas no signifcantly higher threshold is justifed? It is tempting, I think, to suppose that epistocracy is like paternalism in their both being forms of asymmetrical subjection. In that case, epistemic limits to one would suggest epistemic limits to the other. But this would be a mistake: legal paternalism, in a democracy, is not a case of asymmetrical subjection. Such laws, we are assuming, are democratically authorized, and so no one has more fundamental power over such a law than anyone else. Te permissibility of some paternalistic law would not indicate any exception to a right against asymmetrical subjection. It marks some exception to the noninterference principle, but that is entirely diferent. Tis raises the question, though it does no more than that, whether this opening can be exploited against epistocracy, by showing that the right against asymmetrical subjection is more robust against epistemically based exceptions than is the right against paternalism. I do not pursue this suggestion here.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

6.  Decisional Equality vs. Symmetrical Subjection Tere is another model that fgures in thinking about the justifcation of democracy, and that is the idea of each person having an equal share of power over political decisions.6 Distinguishing between symmetrical subjection and decisional equality will avoid certain kinds of confusion, and will also lead to some refnement of the former idea. In this section I want to explain how equality of individual power over political decisions is not, as such, symmetrical subjection. Te latter is not a kind of arithmetic equality of anything, but a form of hierarchy. When the question is about individual power over the whole range of political outcomes, as in the case of voting, then since many of those are laws applying to all, where there is unequal power there is also something structurally diferent and arguably more in need of justifcation, namely asymmetrical subjection. Te distinction between so-­called distributive and relational equality, which is now familiar in the literature about “relational egalitarianism” can be usefully applied to the idea of equal power over political outcomes. One way to think of decisional equality is in the case of a decision procedure over which each individual in some domain of people has, at least in virtue of the formal procedures, an equal probability, should they participate, to be decisive over the decision. In that respect, power over the decision is distributed equally, in an arithmetic sense: the power each possesses can be quantifed, and we can speak of 5  Lippert-­Rasmussen (2012) has a helpful discussion of epistocracy and elitism. 6  Or equal opportunity for power, or infuence, etc.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Epistocratic Paternalism  107 each having the same quantity.7 Call this decisional equality, and distinguish it from a diferent idea of equal power that is salient in the context of political decision-­ making, especially law.8 What we might call equal interpersonal power9 is where, over some range of matters and some set of people, no person has power over any other that the other does not also have over them. Equal power over outcomes—decisional equality—notably makes no reference to some people’s power over other people, only to power over outcomes of a decision procedure. Asymmetrical legal subjection is a case of interpersonal power, through law, of some over others that the others do not have in return. It might seem that power over the making of law is always a case of subjection— power of some people over others—but this is not so. Suppose there is a political procedure that decides on the names to be given by law to buildings and streets. (Tis is not always a small matter, as we know from recent debates over buildings that had been named for prominent fgures in the Confederacy or the Civil War.) Other similar examples include choosing the poet laureate, or the colors of the fag. If those decisions are made by voting and universal sufrage, then each has an equal chance of being decisive, should they participate. But this is not clearly any kind of power that any of the voters has over the others. It is not a power to coerce them, for example, or to spend coercively raised taxes (names are free, etc.). Tis contrasts with a majority voting procedure to decide speed limits on local roads, among many other cases. Tose decisions will come with coercively backed laws, a clear imposition of power over everyone subject to those laws. So, while power over the law is not always power over others, obviously it ofen is. In either case there is a question of equal or unequal decisional power, but in the latter cases there is also the question of symmetrical or asymmetrical subjection. An important case is where the salient inequality cannot be captured by the idea of unequal decisional power at all, but only by the idea of asymmetrical interpersonal power. Te possibility of this case assures us of the diference between the two kinds of political equality. In a democratic system of lawmaking, since laws apply to everyone (in a given domain) and many laws exercise power over people (unlike the case of naming decisions), anyone who has more power than others over the outcome—perhaps they have doubly weighted votes—also, and distinctly, has more power over the others than they have over her. One natural case is where the ones with more such power are also equally subject to the laws. Te present point is that even then some would have asymmetrical power over others even so. Another case would be if there were a set of people who are not subject to the system of laws, or less subject to it in some way. Ten, even if each had equal outcome power over the 7  Let it be a fraction of a fxed total, or an absolute quantity in a potentially changing total. 8  Here, of course, I am adapting some points that have become familiar in discussions of “relational egalitarianism.” I resist that name which might refer to either of two aspects of Anderson’s (1999) own view: First, that as a principle of justice it applies only to members in real social relations. Second, that the kind of inequality is itself a certain social relationship between people. 9  I do not call it “equal relational power,” because the idea of distributive equality is, in a well-­established respect, a “relational” (sometimes called “comparative”) standard—concerning how much some people have relative to or compared to how much others have. Some proponents of distributive equality interpret the idea non-­relationally or non-­comparatively as taking a prioritarian form (see Anderson (1999) for a discussion of a prioritarian version of luck egalitarianism).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

108  David Estlund laws, others would be subject to this subset of people in an asymmetrical way. Tere can be cases of both unequal outcome power and unequal subjection. For example, monarchs might, in some systems, not be subject to the laws of their own state, or at least not to all of them. Another example would be ofcials from a colonizing state having extra power in a colony. Te asymmetry of interpersonal power is obvious in those cases. However, conceivably even those who are outside the reach of the laws might have only equal outcome power in relation to the others—say, one person one vote. Nevertheless, they have power over the others that is not possessed by the others over them by virtue of being unequally subject to the law they have equal power over. Asymmetrical legal subjection, then, takes at least these two independent forms: diferential degree of subjection to the laws, and diferential degree of power over the laws. In cases where only those subject to the law have any power over it, the power over it, or the subjection to it, might nevertheless be unequal, in which case there is asymmetrical subjection. Te moral case for decisional equality, as such, whatever it might be, is not guaranteed to be the same as the case for symmetrical interpersonal power. For example, Tom Christiano (2008) defends equal decisional power as such, and not on the ground that decisional power is power over others. Te importance of equal power over outcomes, on his view, is as an essential public acknowledgment of the political system’s equal regard for everyone’s interests. Since a standing unequal decisional power over laws is, in large part, also a case of asymmetrical subjection, that latter fact raises the possibility of a distinct critique.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

7.  Te Recurring Pattern of the Epistocratic Challenge to Political Equality In this fnal section, I move past whatever illumination there might be in an analogy between epistocracy and paternalism. I want to mention briefy three other philosophical settings for the defense of democracy, which I will not address in detail, in order to see how the epistocracy challenge arises in a recurring pattern. In this way, we get a better idea of how a very broad family of arguments derived from some requirement of equality among persons can naturally be challenged—maybe not decisively, but that is what needs determining—on epistemic grounds.

7.1  Anti-­hierarchy Here is a simplifed form of argument, stated in my own terms, and adapted to my purposes, drawn from recent work of Niko Kolodny (2014). I mention it only to help in exhibiting the broad pattern of the kind of challenge to democracy on epistemic grounds of which it is an instance. 1. Tere is a right against broad social hierarchy. 2. Non-­democratic forms of political rule are, or produce, broad social hierarchy. ___

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistocratic Paternalism  109

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3. Terefore, there is a right against non-­democratic political rule. 4. Epistocracy is a non-­democratic form of political rule. ____ 5. Terefore, there is a right against epistocracy. Te interest here, it seems to me, is as an efort to avoid relying on a dogmatic assertion of a right to political democracy itself. Te thought seems to be that if we think about broad social hierarchy, it may be easier to see how and why avoiding it can have signifcant importance against the possibility of inegalitarian political procedures, even if they would perform better. Now, the right to any form of equality, such as the absence of social hierarchy could, in principle, be absolute. In that case, no matter how disastrously the entailed (suppose) democratic form of governance might perform, then even if there is an alternative political procedure that would perform far better—say, much better for every person, or with outcomes that are more just—it is required that the alternative be declined and those advantages be forgone. Tis makes democratic absolutism sound implausible. But of course, a challenge of the same form waits in the wings for an alleged requirement of democracy of any great strength, even if not absolute. Notice that even if, as in this example, the kind of equality being appealed to in such arguments is “relational” or anti-­ hierarchical rather than arithmetic (or “distributive”),10 there is something here reminiscent of the famous leveling down objection for arithmetic equality principles.11 Here, the counterpart question is this: does the sort of symmetry required by that kind of relational equality have value of such a kind and weight that it can outweigh even signifcant advantages in substantive justice of decisions and outcomes, or, even benefts to all, as in the classic leveling down problem? Must political status be leveled, on grounds of procedural justice, at the cost of a downward impact on the substantive justice of the laws? Returning to the sketched argument, and leveling down aside, a requirement, with some signifcant weight, of political democracy is held to be grounded in a requirement, with some weight, of social non-­hierarchy. Tat plausibly helps make a case for forgoing at least some outcome advantages of epistemically better political arrangements. Te bump in the rug—the recurring pattern—is this: How robust is the egalitarian requirement that is doing the work? It may depend partly on how pronounced a social hierarchy is necessitated by certain epistocratic arrangements. Te pattern, at any rate, goes like this: Is it so important to avoid that degree of social hierarchy that it is worth forgoing signifcant advantages in the substantive quality and even justice of political decisions? Indeed, couldn’t social hierarchy of some kinds be rendered legitimate if they are democratically authorized, and so under the authority of symmetrical legal subjection, chosen for what seem to voters to be good reasons? It is just a question, of course, and it does not answer itself.

10  I follow Anderson’s (1999) meanings for these adjectives. 11  I believe Parft (2002) coins the term.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

110  David Estlund

7.2  Priority of Liberties An additional instance of this same pattern, which may have occurred to the reader by now, is the supposed lexical priority of the equal basic liberties in Rawls. No inequality in people’s basic liberty is permitted, not even for the sake of benefts in the form of more goods for every person. Now, Rawls in particular argued that equality of at least the political liberties, and, indeed, also their fair substantive value for individuals, was essential for guarding against what would eventually be a profoundly unjust society. As he says (Rawls, 1993, pp. 327–8), “unless the fair value of these liberties is approximately preserved, just background institutions are unlikely to be either established or maintained.” Te so-­called “fair value of the political liberties” might be thought of as an implication of the idea of the procedural fairness of the political system, a version of what I have been calling decisional equality. It depends, though, on what is meant by procedural fairness. If that means to refer to a value that does not derive in any way from the quality (say, justice or injustice) of the procedure’s outcomes—call that outcome-­independent procedural fairness—then, while that is a common way of thinking about the importance of political equality, it does not seem to me to fully describe Rawls’s way since he has both reasons for “fair value.” Afer all, as he writes:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Te guarantee of fair value for the political liberties is included in the frst principle of justice because it is essential in order to establish just legislation and also to make sure that the fair political process specifed by the constitution is open to everyone on a basis of rough equality.  (Rawls, 1993, p. 330)

Next, it might look as though Rawls is condoning some compromise of political equality for the sake of better decisions. In the context of campaign fnance, Rawls (1993) argues that epistemic considerations could favor some inequality in freedom of political expression in defense of fair value of the political liberties. In this campaign fnance case, the fair value of the political liberties, whose point is largely instrumental on his view, is in tension with their strict formal equality. If the rationale is limited to the preservation of the basic liberties over time, then this would not seem to violate the lexical priority of that principle over the others. Unfair value of political liberties threatens equal basic liberty overall, and might need to be guarded against by certain formal inequalities of political expression in the form of political spending restrictions. If the right to political speech might need to be unequal, what about the right to vote itself? Rawls (1993, p. 204) takes a nuanced view in his treatment of, as he describes it, “Mill’s view that persons with greater intelligence and education should have extra votes in order that their opinions may have a greater infuence.”12 He argues that since the political liberties are especially important in their role of protecting the other equal basic liberties, the equal basic liberties principle does not automatically rule out political inequality of that specifc kind. He argues that if Mill is right that plural voting would better protect the overall package of equal basic liberties, then that is a strong case for that 12  Rawls (1993) is referencing Mill (1946) here.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistocratic Paternalism  111 kind of epistocracy, all of this within the strictures of the two principles of justice. Te equal basic liberties principle, along with its lexical priority over fair equality of opportunity and the diference principle, does not preclude giving more votes to those with more intelligence and education. Tis can be surprising, especially since, as we have seen, the political liberties are singled out by Rawls as having a special (also instrumental) importance that calls for a guarantee in the special case of their “fair value.” But this does not mean the political liberties are the most important. Indeed, they are subordinate in the sense that their primary importance is as means to the others. About Mill’s proposed epistocracy of the intelligent and educated Rawls writes, without trying to decide the matter,

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Now the ship of state is in some ways analogous to a ship at sea; and to the extent that this is so, the political liberties are indeed subordinate to the other freedoms that, so to say, defne the intrinsic good of the passengers. Admitting these assumptions, plural voting may be perfectly just.  (Rawls, 1999, p. 205)

For our purposes, the main point is that Rawls is explicit that, while there is also important non-­instrumental value to the equality of the political liberties, the two principles of justice are not to be understood as precluding unequal political liberties such as the epistocratic device of plural voting if the gains in the protection of the equal liberties overall is sufcient. Tis would be, in efect, an appeal to a competence gap sufciently great to override such weight as the principle of political equality may have, in support of a version of epistocracy. It may seem that it is too strong to call the idea of plural voting a case of epistocracy, since everyone is still entitled to at least one vote. Isn’t that a modifed form of democracy? Tere are two parts to the reply. First, I have defned epistocracy, for simplicity, in the more capacious way. Obviously, diferent versions can vary more or less from systems with equal sufrage. Second, while Rawls and Mill only consider the moderate case in which no one is utterly deprived of the right to vote, none of the considerations adduced tells decisively against the more radical case in which some or many are disenfranchised on the very same grounds. Rawls has sketched the form that the argument for such partial sufrage would have to take. Te case would need to be made that this was the best way to protect the other equal basic liberties from a sufcient danger of violation over time. So, even if one wished to limit the term “epistocracy” to cases in which only some citizens are allowed to vote, epistocracy would apparently not be ruled out by the equal basic liberties and their lexical priority, at least as Rawls sees the matter. Whether a strong enough case for it could be made would depend on considerations beyond the principles of justice, appealing to the various kinds of evidence that might bear on whether such an arrangement is necessary in particular circumstances in order to protect the equal basic liberties over time.

7.3  Qualifed Acceptability And now, in fairness, let me explain how the pattern shows up for my own preferred defense of democracy against epistocracy. Tis is not an efort to defend this defense

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

112  David Estlund of democracy, but only to lay out enough of it to see how a challenge arises that fts the pattern. Elsewhere, I posit a principle of a broadly Rawlsian kind, according to which at least fundamental elements of the imposed order are not permissibly imposed unless there is a justifcation that is acceptable to the wide range of views we should regard as so-­qualifed (Estlund, 2008). Te details won’t matter here, and I call it a qualifed acceptability requirement. So, even if some form of epistocracy (I especially consider an epistocracy of the educated) would indeed make better and more just laws, that kind of political inequality would not be permissible unless that case for it were beyond qualifed disagreement. Next, I argue that it would not be plausibly disqualifed—even if you don’t think it would be correct—to suspect that such things as selection efects might well skew the demographics of the “educated” in a way that throws into doubt their epistemic superiority with respect to making better and more just political decisions (Estlund,  2008, ch. XI).13 If so, then that form of epistocracy would be turned back. Tat gives the pertinent form of the argument. Te familiar pattern of the epistemic challenge to democracy appears again here, as follows: Te challenger asks, what is so important about justifcation to all qualifed views (at least many of which are mistaken, by the way) even if this means forgoing the signifcantly better and more just decisions that could be made by the true, but controversial epistocrats? Even as the pattern repeats itself, we might also start to discern the two main ways in which the challenge might be answered, and they are not exclusive. One, of course, is to establish that the moral importance of the relevant kind of manifest equality that equal sufrage involves indeed outweighs the moral importance of having the better and more just decisions. Te second, which I have not yet mentioned and will not pursue here, would be to argue that certain democratic arrangements that satisfy the relevant kind of equality have enough epistemic value of their own that the epistemic advantages of epistocracy are, to some extent, blunted. My own approach has been to doubt that without such an epistemic case for democracy itself, the appeal to the relevant kind of equality will be enough. Of course, that would be compatible with thinking that a strong case for the intrinsic value of the form of equality might also be a necessary part of any answer.14

References Anderson, E. S. 1999. “What is the Point of Equality?” Ethics 109(2): 287–337. Christiano, T. 2008. Te Constitution of Equality: Democratic Authority and its Limits. New York: Oxford University Press.

13  See Estlund (2011) for a response to critics. 14 I am grateful for discussion of earlier versions at the Political Epistemology Workshop, at the Kennedy Institute of Ethics, Georgetown University, October, 2018, the Workshop in Social Justice Teory, UC San Diego, May, 2019, and at University of Bayreuth, and University of Hamburg, both in June, 2019.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Epistocratic Paternalism  113

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Dworkin, G. 2020. “Paternalism.” Te Stanford Encyclopedia of Philosophy, E.  N.  Zalta (ed.), . Estlund, D. 2008. Democratic Authority: A Philosophical Framework. Princeton, NJ: Princeton University Press. Estlund, D. 2011. “Reply to Copp, Gaus, Richardson, and Edmundson.” Ethics 121(2): 354–89. Guerrero, A.  A. 2014. “Against Elections: Te Lottocratic Alternative.” Philosophy and Public Afairs 42(2): 135–78. Kolodny, N. 2014. “Rule Over None II: Social Equality and the Justifcation of Democracy.” Philosophy & Public Afair 42(4): 287–336. Lippert-Rasmussen, K. 2012. “Estlund on Epistocracy: A Critique.” Res Publica 18(3): 241–58. Mill, J.  S. 1946. On Liberty and Considerations on Representative Government. Oxford: Basil Blackwell. Parft, D. 2002. “Equality or Priority?” in M. Clayton and A. Williams (eds.), Te Ideal of Equality. New York: Palgrave Macmillan Ltd, pp. 81–125. Rawls, J. 1993. Political Liberalism, New York: Columbia University Press. Rawls, J. 1999. A Teory of Justice. Cambridge, MA: Harvard University Press. Ripstein, A. 2006. “Beyond the Harm Principle.” Philosophy & Public Afairs 34(3): 215–45. Ripstein, A. 2009. Force and Freedom: Kant’s Legal and Political Philosophy, Cambridge, MA: Harvard University Press. Rousseau, J. 2019. “Te Social Contract” and Other Later Political Writings. Cambridge: Cambridge University Press. Shifrin, S.  V. 2000. “Paternalism, Unconscionability Doctrine, and Accommodation.” Philosophy & Public Afairs 29(3): 205–50. Viehof, D. 2014. “Democratic Equality and Political Authority.” Philosophy & Public Afairs 42(4): 337–75.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

7

Te Basis of Political Equality

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Tomas Christiano

Recent critics of democracy pose an important challenge to democratic institutions from an epistemic standpoint.1 Tese criticisms have the efect in my view of challenging instrumental justifcations of democracy and intrinsic justifcations. Tey argue democracy is rule by the ignorant or by those who must appease the ignorant. Tey base this idea on an economic theory of information, which asserts that citizens do not have incentives to become adequately informed in a democracy. Tis is backed up with data suggesting that citizens in large democracies are unable to answer correctly basic questions about their society and its democratic institutions. Tis is a challenge both to the idea that democracy is instrumentally valuable and that it is an intrinsically fair way of making decisions because it suggests that the power that is supposed to be equally distributed to people is without much value. Tey argue either for radically diminishing the size of the state or for rule by experts. I argue that their pessimism about democracy is overstated and that there are good grounds for thinking that democracy can work well despite the necessity that it work in a context of low information decision-­making. Te reality of low information decision-­making need not undermine democracy’s promise to advance the interests and concerns of citizens in an egalitarian way. I argue that a careful and empirically grounded account of the institutional context of democratic participation can show us how citizens can properly carry out their roles in a modern democracy. In this chapter I will focus mainly on the epistemic criticism of democracy. I will start by laying out the argument for the instrumental and intrinsic values of democracy and I will show how it relies on the expectation that citizens can act on the basis of adequate information about politics. I will argue that there is a prima facie case for the thesis that participation is adequately grounded in good information. I will then lay out the crude version of the Downsian model that is meant to undermine the expectation that citizens are adequately informed. I show how this model is defective. I respond to this argument with a revised conception of how citizens think in a democracy, that can help us see both how democracy might work reasonably well and how citizens might not be very knowledgeable about politics.

1  See Pincione and Teson (2006); Caplan (2007); Somin (2014); Brennan (2016).

Thomas Christiano, The Basis of Political Equality. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Thomas Christiano. DOI: 10.1093/oso/9780192893338.003.0008

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Basis of Political Equality  115

1.  Te Critique of the Intrinsic Importance of Political Equality It is important to address, frst, whether an epistemic critique of democracy also damages arguments for the intrinsic importance of democracy. My thesis here is that the intrinsic worth of democracy depends on the expectation that citizens can act on the basis of adequate information about the democratic system and society in which they live. Te basic reason for this is that the intrinsic worth of democracy is dependent on the instrumental value of political power to citizens because the intrinsic worth consists in the equal distribution of instrumentally valuable political resources. Tis, in turn, is dependent on citizens’ being able to act on the basis of a good understanding of the society and politics. I will argue later that the skeptics’ pessimism regarding this expectation is ill founded, though I do not think that it has been refuted. But it is important to determine how serious the epistemic worry is.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Te Importance of Political Equality Equal political status implies that persons each have a set of equal political rights. Tese rights include the right to vote and the right to run for ofce, the rights to form and participate in political associations and express political opinions. Tese are rights to participate in political discussion and propose political solutions to problems. We are talking about status and not merely rights because the status comes with duties and liabilities in addition to claims and immunities. Equal political status gives one the duty to treat one’s fellow citizens as equal participants in decision-­ making and the duty to comply with collective decisions that have been made in an egalitarian way. One acquires equal political status by membership in the community and by meeting some publicly accepted standard of minimal competency. Tere is no further need to assess equal ability or capacity for good judgment beyond this minimal competency. And, to be sure, the means by which one determines minimal competence are very crude—such as a minimum age and some minimal mental abilities. Let us examine the grounds of this equal political status. To be clear, the intrinsic worth of democracy does not rest on, and in fact explicitly avoids resting on, any fact including the equal capacities of persons to have good political or moral judgment. Equal political status is founded on the idea that there is deep disagreement among persons about what the good is for persons and about how society ought to be organized. A normal concomitant to such frst order disagreement is that persons disagree about who are reliable sources of truth on these matters. Tere is little disagreement with the claim that some people are more reliable sources of truth than others, but there is a lot of disagreement on the relevant criteria for assessing this relative reliability. Te basis of political equality cannot be something that is inconsistent with each person’s judgment, so it is not to be founded on an assertion of the equal abilities or capacities of persons. Tis disagreement is, again, a normal consequence of the basic disagreements about the good and the right already

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

116  Thomas Christiano mentioned. Tis reality of disagreement is one of the most fundamental problems that a conception of political equality is meant to deal with. It is far too much ignored in the more skeptical discussions of the epistemic dimension of democracy. Equal political status is founded on the idea that there is a way in which we can treat each other as equals despite each one of us thinking that some are better than others in coming up with sound political judgments. Te basic idea is that society sets aside judgments of superiority, inferiority, or equality of judgment in deciding on basic rights. It gives persons equal political, economic, and liberal rights to fgure out for themselves who they trust or believe and who they do not. So, for example, even in cases where people properly defer to the judgments of other persons, we still leave it in signifcant part to the non-­experts to determine who they will trust and who they will not trust to do a good job. If we were to distribute rights to make these decisions (about who to trust and defer to) unequally, that would be a clear way of treating people as unequals. Equal political status is, to be sure, subject to a condition, which is that these people are minimally competent. Tis is why we do not assign full rights of this sort to young children and severely mentally disabled persons. To be sure, in the context of economic exchange we impose some limits on who is permitted to ofer certain services. And this depends on certifcation that the person has achieved a certain amount of knowledge relevant to their services. We impose licensing requirements on doctors, engineers, lawyers, and plumbers. In some cases, this limits the freedom of each person, to some degree, to choose the kind of person from whom to receive certain services. For example, only doctors and registered nurses can prescribe certain medicines or hospital care. Of course, patients are usually free to reject their prescriptions. And the license is a salient signal of who it is best to receive advice from. At the same time, each person may seek out advice from anyone she pleases. We do not legally impose even such minimal requirements on politicians, party ofcials or opinion leaders. To be sure, the opinion leaders tend to be reasonably highly regarded journalists, academics or religious leaders, and politicians tend to be lawyers or other kinds of professionals. Some degree of informal sorting of persons occurs in a political society for selecting those who are to occupy the more demanding places in the division of labor in politics. What accounts for the diference? I want to discuss this more later, but the basic reason is that the degree of disagreement in politics is normally greater than the degree of disagreement on medicine or plumbing. In politics, there are disagreements on basic values and on the theories of society which help us understand how these values are to be achieved.2 And this disagreement is signifcantly greater in scope than what we fnd in medicine, where we fnd signifcant agreement on the aims and science grounding medical treatment. Equal status in liberal and political rights is founded on the equality of importance of the fundamental interests of persons coupled with the reality of deep disagreement and confict of interests among persons. Te fundamental interests must be advanced despite the reality of disagreement. And they must be advanced in an egalitarian way. 2  See de Ridder, Chapter 12, in this volume for further discussion of deep disagreement.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Basis of Political Equality  117 Te basic way in which to advance people’s interests despite the reality of deep disagreement about what those interests are is to give people power to advance their own interests. Democracy, economic rights, and liberalism are three of the central ways in which we can distribute power in an egalitarian way to individuals and allow them to advance their interests as they see ft. Tis depends on an empirical idea, which is that giving people the kind of power involved in exercising liberal rights or economic rights and the power involved in exercising political rights will in fact advance their fundamental interests in society and that depriving a group of persons of such powers will set back those interests in a publicly clear way. Here we see a potential challenge to the justifcation of democracy (and liberalism). If democratic, economic, and liberal rights all work by giving people power to exercise and act on their judgments, we ought to worry about whether people are up to the task of exercising their power wisely. To be sure, the question is not merely whether people’s judgments are any good simpliciter, it is whether inequality and hierarchy are better at advancing the interests of people in society than democratic, economic, or liberal rights. No doubt giving power to people is a risky proposition because they can sometimes do very bad things with the rights and power that they have, whether these be liberal rights, economic rights, or political rights. And if a lot of people do very bad things together, this can have very bad consequences for everyone else. It is these risks that lead Plato to reject democracy and liberalism altogether and that lead Hobbes to favor monarchy without liberal rights. Plato and Hobbes think this problem applies equally to liberal and economic rights. In contrast, the core proposition at the root of liberalism and democracy is that society should be organized on the basis of the judgments of the people in the society acting as equals. It is a risky venture and constantly needs tweaking to avoid really bad outcomes. So far, though, it has worked reasonably well, while highly imperfectly, in both political and economic decision-­making. We need to carry out a comparison between the wide distribution of power characteristic of democracy and the highly concentrated distribution in epistocracy. Te frst thing to note is that critics of democracy are wrong to say that epistocracy has never been tried. Indeed, it seems to me that epistocracy has been the reigning idea behind governments since the fall of the Roman Republic. It is not until the later part of the nineteenth century that we have universal manhood sufrage and not until the middle of twentieth century that we have a full-­dress version of democracy in the nation-­state. Te exclusion of women was ofen justifed on the basis of epistocratic considerations. Te exclusion of the propertyless was likewise justifed. Te exclusion of religious, ethnic, and racial minorities has ofen been justifed on these bases. Epistocracy has been the standard theory. In the twentieth century, epistocracy was the standard justifcation of Communist and theocratic societies such as Iran and many authoritarian societies. Te results have everywhere been pretty similar. Tose who are deprived of power have their interests set back. Tis is not merely the consequence of selfshness on the part of the rulers. We cannot say this of Communist societies. We cannot say this about men’s willingness to exclude women. Tey thought they had the best interests of their charges in mind. We cannot put this down to blind and invidious discrimination either. Twentieth-­ century

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

118  Thomas Christiano authoritarianism avoided these in signifcant part. But in all these cases, epistocracy failed. And it failed in a way that has come to be seen as highly predictable now. Tey failed because they were unable to advance the interests of those who did not possess power. Is this characterization of epistocracy unfair? Perhaps. But the examples involve very diferent societies: communist societies devoted to equality, laissez-­ faire societies devoted to some version of economic liberty, societies devoted to one religious ideal or another. Tey involve very diferent institutional structures: hereditary aristocracies, party-­led systems, and electoral systems with restricted franchise or with unelected houses of experts of various sorts. One thing to be noted is that in each case, the epistocrats chose themselves in some important way. Either they simply took power on the grounds that they knew best, or they established systems with criteria that tended to favor themselves. And perhaps this is a poor way of selecting epistocrats. Maybe the people who chose themselves were always the wrong people. But certainly, the past history of eforts at epistocratic rule is a humbling one for those who hold out the hope that we can eventually fnd a way to put the true epistocrats into power.3 Te democratic idea and the democratic arguments that theorists have made are founded in the frst instance on the failures of epistocracy. Te recognition that only when people have power, whether it be liberal power or political power, are their interests advanced, is the frst fundamental idea of liberal and democratic thought. It is the idea that animates the drive for the inclusion of the working class, women, and minorities. Te worst of generally fare better when they are included. Tere is a strong relationship between the rise of universal manhood sufrage by the end of the nineteenth century and the era of progressive reform in the U.S.  and Europe culminating, afer the extension of the sufrage to women, in the creation of the modern welfare state with its protection of unions, workers, the elderly, and many others and its concomitant high rates of economic growth and increased economic equality in the second half of the twentieth century.4 Granted there may be disagreement on whether this was good or bad, there is certainly a sufciently large body of economic opinion that favors the development of the welfare state to suggest that reasonable people disagree here. But one thing it seems hard to argue against here is that the disadvantaged did know how to promote plausible conceptions of their interests. Another interesting piece of data in support of this view is the recent 3  Tese empirical theses do not undermine the claims of weaker, more egalitarian forms of epistocracy such as the lottocracy proposed by Alex Guerrero in Chapter 9 in this volume and Lopez-­Guerra (2014). We do not have much experience with these forms of rule. It is beyond the scope of this chapter to evaluate these ideas, but I do think that such theories dramatically underestimate the importance of competition for power and honors in inspiring politicians and political parties to listen carefully to ordinary citizens and to propose policies that genuinely advance the interests of citizens. Admittedly, the competitive process can be distorted by the need for money in campaigns and by demagoguery. But if we can limit the infuence of money and enhance the social bases of citizen understanding (see below), competition can be a major force for good. 4 For the argument that democracy signifcantly enhances economic growth, see Acemoglu et al., (2019). For a brief survey of evidence see Acemoglu and Robinson (2009, pp. 58–64).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Basis of Political Equality  119 work of economic historian Gavin Wright arguing that the Voting Rights Act of 1965 played a major role in boosting the economic fortunes of African Americans in the South as well as those of poor whites (many of whom were also burdened by the literacy tests and poll taxes). Wright argues that we can see that Black disenfranchisement afer the end of Reconstruction led to a precipitous loss in educational resources as well as other state resources previously devoted to African Americans and poor whites. And these resources were signifcantly restored fairly soon afer the passage of the Voting Rights Act (Wright,  2013). When we add to these kinds of cases the more general fndings that democracies tend not to go to war with one another, tend to protect basic human rights much better than non-­democracies, tend to produce public goods on a much greater scale, avoid famine in areas previously plagued by it, and many other results (Christiano, 2011), it is really hard to argue that the disadvantaged are not getting some signifcant bang for their votes. To be clear, the thesis is not that citizens vote their own self-­interest, though sometimes they do. It is that their conceptions of the common good are normally heavily skewed towards their own interests and experiences, not by design but because of the ordinary limitations of human cognitive functioning. Tough each has a duty to pursue the common good, each has an interest in participating in the process to make sure that her interests and those of similar persons are not ignored or marginalized. Of course, this logic applies primarily at the level of interests that are shared with many other people. Tat is where the argument for the intrinsic value of democracy hits the road. It starts with the observations of the importance of the possession of power to the advancement of interests and the recognition of deep disagreement and confict of interest. We need an inclusive distribution of power if we are to advance the interests of all citizens. But, given the reality of disagreement among citizens, there is no clear outcome standard by which to evaluate the diferent inclusive distributions of power. Te idea is that when there is a great deal of controversy and disagreement on how to advance the common good and justice, there is also likely to be controversy on ways to measure people’s reliability in coming up with good political judgments about how to advance the common good and justice. It is hard to see how we can demonstrate the superiority of a person’s capacity for political thinking over another’s without presupposing some conception or another, because, in the end, it must be that person’s superior abilities to come up with the right answers that settles the matter. But people are generally biased towards their own interests and experiences when they try, however conscientiously, to arrive at a reasonable political judgment. As a result, we have reason to think that favoring any particular conception of justice and the common good will have the efect of favoring some people’s interests over those of others. Consequently, eforts to arrive at standards for assessing competence cannot but be widely and reasonably understood as in violation of the public realization of equal advancement of interests (Christiano, 2008, ch. 3). So, we cannot use a further outcome standard to evaluate the diferent inclusive distributions of power. As a consequence, we must assign political status to persons without recourse to such standards either for assessing outcomes or competence. Hence, we must distribute power equally, otherwise we publicly express the lesser importance of the

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

120  Thomas Christiano interests of some relative to others in the society. We can only accept some minimum possession of political and economic competence and the equal importance of each person’s interests. Tis argument is for the intrinsic importance of equal distribution of power because it selects equal distribution among the many possible inclusive distributions, not on the basis of any outcome standard but directly on the basis of equality. Te intrinsic value asserted here is a conditional one, conditional on the facts of judgment and diversity as well as the idea that we are equalizing something that is highly valuable to the persons who receive it.5 One objection to the above questions the thesis that distributing radically unequal political power to diferent people symbolically expresses the inferior importance of those people with less or no power. Jason Brennan correctly observes that we do not think of a person as an inferior just because we accord his opinion on what we ought to do with our broken faucet less weight than that of a plumber or when we think the opinions of pilots on how to fy airplanes are superior to others’ opinions (Brennan, 2016, p. 122). In response, on the account I have been developing, once the basic distribution of power is set, each person is permitted, and expected, to choose some as better sources of understanding than others and to act on the basis of a judgment that is partly grounded in the deference he or she accords those better sources. Again, this holds for the distribution of political power and rights as well as for liberal and economic rights. Tere is no inconsistency in thinking on the one hand that one is right on a certain matter or even that one is better than others at thinking about that matter and on the other hand thinking that it is reasonable for the society not to base its distribution of rights of participation on one’s own assessment. A citizen can think that he or she has better ideas than fellow citizens and yet also assent to a distribution of power based on the importance of each person’s interests and the reality of fallibility, bias, and confict and not based on the idea that everyone’s capacities for judgment are equally good. What would be a serious abridgement of equality is if someone were not permitted to choose the plumber they want or the doctor they see or if they needed to go through some extra steps (relative to others) in making such a choice. Ten the person would be treated as if she were a child or a severely mentally disabled person. Analogously, in the case of politics, not allowing someone to choose their representative or their favorite opinion leader would also amount to a serious abridgement of equality. A second kind of argument is that what symbolically conveys the message of equal status is a highly contingent matter. It is a matter of a particular culture and particular history. So, it is concluded that there is no reason to think that equal political power is necessary to expressing the equal political status of persons in society. If epistocracy 5  I have argued with Will Braynen that equality in distribution is valuable only on condition that the objects distributed are valuable in the sense that more is better than less. An equal distribution of the letters in one’s last name is of no importance. Furthermore, we have argued that the proper understanding of equality as a principle of distributive justice rejects leveling down as a way to realize greater justice. See Christiano and Braynen (2008).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Basis of Political Equality  121 does better in advancing people’s interests generally, then we should abandon the idea that epistocracy necessarily symbolically expresses inequality of status (Brennan, 2016, p. 128). In response, one should grant that it is a contingent feature of democratic societies that equal political power expresses the equality of persons. But any grounding of a political system is dependent on contingent facts. Te question is what the facts are and how deep they are. And I want to say that the facts on which the argument above rest are very deep ones and that we are not in a position to change the meanings they convey. Tey involve a kind of social necessity. It is hard to see how any other inference could be defensible in the light of the facts about judgment I have pointed to and their implications for understanding the failures of previous eforts at epistocratic rule. To conclude this main part of my piece, the argument for democracy, including the intrinsic version some people have ofered, relies on the idea that the right to vote and other political rights must have signifcant instrumental value for the persons who possess it. Te idea that democracy has intrinsic value, afer all, depends on it being intrinsically valuable to distribute political power equally to persons of minimum capacity. And this depends on the idea that political power is actually valuable to the people who possess it. But, in fact, I think we have good reason to think that the possession of the vote has been a great beneft to those who previously lacked it and were otherwise disadvantaged. As a consequence, the rest of the argument for democracy can go through. Tese very same arguments show the weakness of epistocracy and why it is that people throughout the world have rejected it in favor of democracy.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Citizenship and Democratic Participation Something seems to be working in modern democratic societies to promote the interests of the members, but we do not have a good idea of what it is. Te main theme of the skeptics is that we cannot expect much good to come from citizen participation because citizens are either largely ignorant of political and economic issues or excessively partisan. How can political power be useful to people if they are ignorant? How can they be the governing element in the society under these circumstances? We do not have a clear idea of how citizens are the governing element in society and how they can be equal participants in it. Without this idea we cannot have a cogent account of the underlying ideal of political equality.

4.  Te Crude Downsian Model of Citizenship Te skeptics’ conception of citizens derives its support from a particular model of how citizens think about politics as well as empirical evidence that appears to support the model. Te basic model comes down from Joseph Schumpeter and Anthony Downs, though I will argue that the model is not really true to the Downsian picture (Schumpeter, 1950, ch. 21; Downs, 1957, part III). In what follows, I will lay out the

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

122  Thomas Christiano model on which the skeptics rely in a bit more detail and I will then criticize the model. I will defend a more sophisticated account of the nature of citizenship in an egalitarian society and show how it is possible to vindicate the ideal of political equality. Te root idea is very simple. Citizens have extremely little impact on policy through the vote, which implies that the value of becoming informed about policy is very low if its main purpose is to bring about good outcomes. To see this, note that the discount on the value of the vote is very high in large electoral constituencies because the chance that one’s vote makes a diference is very small. If the value of whatever information one collects is that one makes a positive diference to the outcome when one is well-­informed, then the value of the information itself is very small because discounted so heavily. A rational voter will not spend much time or resources to acquire such information, at least if they are primarily self-­interested and outcome oriented. On this account, there may be some value in becoming informed if we can use this information to get benefts directly from the political process itself. For example, we might fnd it enjoyable to form and participate in groups regardless of how efective they are in bringing about good political outcomes (Olson, 1965). We just want to win because it is fun to fght together. Tis produces low quality policy ideas and highly antagonistic relations with others who are not part of the group. So, it is even worse than simple ignorance. Te assumptions are: one, people do not place very much store in the common good, two, they act only individualistically to produce good outcomes and, three, only an appreciation of the frst-­order reasons for action is a suitable basis for actions such as voting. Tese three propositions amount to what I call extreme individualism. I think all three are false. It should be noted before we start that none of these assumptions are part of Downs’s picture. Tat is why I describe this as a crude Downsian model. Te frst two assumptions are added on and the third is clearly rejected by Downs’s account. Let us examine each one of these assumptions. First, let us examine the two motivational assumptions: the assumption that in politics people are primarily self-­interested and the assumption that, in politics, they do not act as contributors to a larger common project but merely as separate maximizers of good outcomes. On the frst assumption, citizens do not have to be completely self-­interested, indeed most voters must be concerned with the common good to some extent, otherwise they would not go to the polls at all.6 But the assumption of the primacy of self-­interest is necessary. For highly altruistic voters (who care a lot about the common good), the value of the consequence can be sufciently great (because so many people are afected) that, even if the probability that informed participation makes a diference is extremely low, it would still be rational to become informed about the values of outcomes (Parft, 1984). Second, the usual skeptical view is that persons are individualistic maximizers in politics. Tey do not think of themselves as participants in a collective project and so they do not think of themselves as doing their share of what is necessary to bring about a general good. Each citizen, on this view, simply fgures out how he or she can 6  Somin calculates how much of a concern with the common good is necessary to get voters to the polls while not being enough to get them to inform themselves.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Basis of Political Equality  123

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

do the most good independent of how others act. So collective projects, in which each person’s contribution is very small and inessential, are avoided in favor of individual projects. Te third component of the crude Downsian picture asserts a kind of individualism about knowledge and action based on knowledge. Te idea is that a person can act on the basis of a good understanding of the issues only if she possesses that understanding in a way that is directly available to her. Let us call this a kind of cognitive individualism with regard to theoretical and practical reasoning. My action is practically justifed to the extent that it is based on my own appreciation of the reasons relating directly to my action and its outcome. If I am not currently able to grasp the frst-­order reasons for my action, then I am not acting with practical justifcation. And one strong indicator of this, though not conclusive, is that I cannot articulate the reasons when I am asked for them. Tis last point is important. One of the central pieces of evidence ofered for the thesis that people vote in ways that are not efective in bringing about good outcomes is the evidence from empirical surveys of citizens’ knowledge of politics going back about 60 years.7 Te net result of these surveys has been consistently to afrm that people do not know the answers to basic questions about politics. Somewhere between 30–60 percent get the wrong answers to basic questions about politics. Tis evidence is sobering, without a doubt. It seems to confrm the model that people are rationally ignorant of politics. Tese are the three elements of the extreme individualist approach to the justifcation of action that undermines the idea that citizens can be efective participants in the political process. Te frst two assumptions support the model of the economics of information in which it is rational that citizens are almost entirely ignorant of politics. Te third assumption, that voters must act on their own frst-­order appreciation of the reasons for good action in order to be efective promoters of interests or the common good, implies that voters will not be efective.

5.  Te Basic Problems of the Crude Downsian Model But there are four very big problems with the crude Downsian model. Te frst big problem is that the crude Downsian model has a very difcult time explaining why democracies tend to do so well by the interests of their citizens. We observe not only that democracies do pretty well on the whole and much better than any self-­described epistocracy, we also observe that as previously disenfranchised groups get the vote, their interests and their ideas about the common good tend to be much better advanced. Here I simply draw attention to the evidence given in the frst half of the chapter. Tis frst problem already suggests that we should be looking to change the basic picture of the crude Downsian model. Te second problem is that people are not uniformly incapable of answering questions on surveys. Afuence and education tends to improve their answers 7  See National Election Surveys.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

124  Thomas Christiano signifcantly. One might think this should be obvious. But the problem for the crude Downsian model is that the afuent and the educated should not be more informed about politics than anyone else. Indeed, if they are well educated, one would think that they would follow the logic of collective action and rational ignorance with greater regularity than anyone else. Tey should be ignorant about politics because they know how little diference it makes to know something. Tey should use their education to develop knowledge that is more useful and more likely to make a diference. Yet we do not see this. And indeed, we not only see that they are better informed about politics but that politicians tend to be more responsive to them than to others. None of this makes any sense on the crude Downsian model. Tere should not be such a regular correlation between afuence, education, and informedness. I will return to this in a moment. Te third problem is that it is hard to credit the primacy of self-­interest view in the specifc context of politics. Tere is evidence that citizens are genuinely committed to pursuing the common good in the context of politics. First, their participation seems to suggest this. And second, surveys uniformly suggest that they are concerned with the common good in voting (Mackie, 2014, pp. 302–4). Moreover, the whole tenor of democratic politics is saturated with concerns about fairness and the common good. In those cases where politicians appeal specifcally to the interests of a particular group of voters, the common assumption is that their interests have been unfairly neglected. Furthermore, people tend to act not exclusively as individual maximizers (of their own or the common good). Tere is extensive evidence that people are willing to do their fair share in the pursuit of collective aims. Tey evaluate the group’s activity in terms of its outcomes, but they evaluate their own individual actions in terms of whether they are doing their fair share of the group’s action. Tis entirely changes the calculations about what they ought to do. Each person does not ask what the chances are she will make a diference individually, but what the chances are that the group, of which she is a member, will make a diference and how she can help. I will say more about this in Section 6. Te fourth big problem is with the individualism of justifcation. Te demand that people must know the frst-­order reasons for their actions and beliefs is simply not met in the modern world for much of the action that we engage in, which action nevertheless does efectively promote our ends. We depend essentially in our lives on other people’s appreciation of the reasons for action. We depend on doctors’, mechanics’, friends’, and others’ judgments about good action, and the empirical conditions of action, to be efective. We ofen do not know very much at all about the things we rely on in much of our lives. And this is a good thing too. If we could not rely on a division of labor in the development of knowledge necessary efectively to promote our interests, we would be in bad shape. We could not do our own jobs, take care of our families, and other things. Te kind of knowledge necessary to the efective promotion of our ends is for the most part held by other people. We depend on them. And the productivity of society depends on our relying on them. When people are asked about some of the basic features of their lives, such as the composition of the things that they eat or the functioning of all the machines they rely on, or the complications of tax law on which they rely, they generally answer

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Basis of Political Equality  125 these questions very badly. Te surveys are no better in the case of politics. Yet, they are able to do things reasonably well to further their interests (Mackie, 2014). Te non-­individualistic nature of our knowledge defeats the connection between the claim that there is widespread inability to articulate an appreciation of the reasons and facts that justify action and the ability of people efectively to promote their interests and other ends. It is quite possible that people give mistaken answers oneto two-­thirds of the time about the qualities of the products they buy and are still able to promote their ends with those products. Te key lies in the fact that they depend on reliable cues from others. So, the evidence culled from surveys of political knowledge does not show that individual citizens are not acting on good political knowledge in politics.8 To be clear, this does not show that the cues are always, or even normally, reliable; it simply short circuits the skeptical argument from the surveys to the thesis that citizens do not act on good information. Te conclusion to be inferred from the surveys must be, at most, unclear. In brief, the crude Downsian model is simply not up to the task of characterizing the basic motivations and epistemological condition of most ordinary citizens. We need to start anew. I will put together the elements of a more sophisticated Downsian model of citizens in what follows. Much is still lef to work out but it is a start.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

6.  Towards a Collaborative Conception of Democratic Citizenship Tere are four components of the view I will develop here. First, I lay out the essentially collective character of political participation. Second, I articulate a non-­ individualistic conception of motivation and knowledge. Tird, I develop a simple model of economics of information that makes citizens acting on good information possible. It can also explain the inequality we see and suggests how that inequality can be overcome. Fourth, I suggest how some institutions can help improve the epistemic condition of citizens on the basis of my sophisticated Downsian model of the economics of information. Tis will allow us also to think through the requirements of political equality for a democratic society. In politics, citizens mostly think of themselves as members of groups of various sorts. Tey identify themselves as citizens, progressives, free marketeers, conservatives, defenders of public morals, a badly treated minority, members of a particular political party, or whatever and they are members in part because they think that the aims of the group are, on the whole, good ones to try to achieve by means of policy and they think the group is more or less capable of competently

8  Tere is very substantial dissent from scholars of the political informedness of citizens about whether the surveys are giving a good picture of how informed citizens are. A large part of that dissent comes from appreciation of the fact that people use many diferent kinds of complicated methods including shortcuts to keep track of political events. For the classic statement of the importance of shortcuts, see Downs (1957, chs. 12 and 13). For the seminal application of the idea of shortcuts to political information see Popkin (1991). For a recent sophisticated questioning of the signifcance of survey data, see Lupia (2016).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

126  Thomas Christiano pursuing those goals.9 Tis involves two elements. As citizens, they have a concern for the common good and justice in society, which they pursue within the highly norm governed and collectivized processes of elections and referenda. As members of more particular groups, they want to make contributions to the pursuit of the particular conceptions of justice and the common good shared within their groups. To be sure, the more particular groups are fuid as people change their minds about the worth of the aims or the ability of the group to pursue them. Tese groups possess power in proportion to the number of members and fellow travelers and to the quality of organization of the groups. And it should also be noted that many individuals, though not a large proportion, are unafliated with politically oriented groups. Tese individuals, for reasons we shall see later, tend to be among the least well informed, politically speaking, in the society (Campbell et al., 1960, p. 143). Citizens think that they should do their fair shares within the group to help it promote the desirable aims. People do this, to think of a non-­political instance, when they contribute to recycling. Tey might think that the group or groups of which they are members have chances to afect outcomes and then proceed to do their fair share of helping the group bring about desirable outcomes. Te group is acting in an outcome-­oriented way, but each individual is only partly acting in an outcome-­ oriented way, they are also just doing their share to advance the group’s concerns.10 Moreover, there is substantial evidence that people are motivated to pursue the common good and justice in politics. Tis may not hold of all individuals, but it seems clear that this is the norm in political contexts. Tere is also substantial evidence that people are motivated by the idea of contributing a fair share to a collective project at least when they perceive enough others to be similarly motivated. Tey are motivated to be “strong reciprocators” in many diferent kinds of environments. And these motivations have been detected in numerous controlled experiments.11 Moreover, the activity of citizenship is an obvious context in which this kind of practical reasoning can take place. And there is evidence that people think of themselves as motivated by this kind of reasoning in politics. Of course, how active citizens are will depend on how much others are participating and how informed they are will depend on the norms they accept concerning how informed they should be. We do not have a clear model of how this kind of motivation works in the context of citizenship, but it would be reasonable to think that it is of great importance.12 When we combine these two points, we can see that the potential for citizens to think that they should be reasonably well informed, and that it might be rational for them to be so, is quite high. Much depends on how concerned with the common good they are and how much they are willing to do their fair share in promoting that common good as well as how much they think others are doing their part. Tese are very large questions about what citizens are motivated to do and without answers to 9  Tis sense of identifcation with a group is the kernel of truth in Achen and Bartels (2017) but it is not incompatible with a collectively instrumental approach by citizens. 10  Julia Maskivker (2019), articulates a conception of voting along these lines. 11  See Fehr and Fischbacker (2006, pp. 151–92) for a classic statement of this approach and the many sources of evidence. 12  See John Roemer (2010) for an efort to model this kind of behavior. See also Fischbacker (2006).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Basis of Political Equality  127 them, an essential part of the crude Downsian inference to the conclusion that citizens will simply be ignorant of politics is undefended and seems to be false in signifcant part. Te other essential element of the sophisticated Downsian approach starts with a kind of collectivism in practical and theoretical justifcation. What I mean here is that many times we act on the basis of frst-­order reasons that only others appreciate. I think my car is in good shape and drive it without concern because my mechanic thinks it’s fne. I think I am in good health because my doctor says so. I don’t really know the reasons for this, but to the extent that the mechanic and doctors I rely on are trustworthy, I act on the basis of good information. I cannot articulate the reasons and I do not have an appreciation of them. As I noted above, this kind of dependence on other people’s knowledge is ubiquitous in modern life, which depends heavily on the division of labor both cognitive and non-­cognitive.13 But it does not undercut the ability of people efectively to promote their interests, indeed the opposite is true. To put it in a slogan, for hunter gatherers, each of whom know an enormous amount about their environments, life is short and difcult, while for us, who don’t, life is long and easy. But let me fll this picture out with an economics of information, which displays the collective dimension of knowledge.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

7.  Economics of Information Let us put together a very simple model of the economics of information here. Every person collects information up to the point where the marginal beneft of the information equals the marginal cost. And if we take the springs of motivation to be those I outlined above, marginal beneft refers to the extent of my doing my share in promoting the collective aim and marginal cost refers to the other aims I do not contribute to and the things I cannot do to advance my interests and those of my friends and family. Tis is quite a heterogeneous set of things so it will be hard to give it a clear formal representation. In any case, what this implies is that people will economize on information gathering. Tey will not try to know everything because the marginal cost of acquiring a lot of such knowledge is greater than the marginal beneft. Te more time I spend collecting information, say, about politics or about my car, the less time I have to do other things like play with my child or do my job. At some point, the marginal cost is greater than the marginal beneft of new information. And in such a case, I do not collect the relevant information. Tis is what Downs means by rational ignorance. Tere are many things of which I am ignorant because it is not worth my while to have the information. Tere are two essential steps here. Te frst step is simply ignored by the standard skeptical account, but it is utterly obvious: I need to know some things (or at least believe some things) in order to fgure out what the marginal costs and benefts are of collecting information. Second, I need to know how to diminish the costs of information collection. 13  See Hutchins (1995) for the idea of distributed cognition in which people depend on each other’s cognition to act. See Hardin (2009) for a discussion of this in many diferent areas of life.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

128  Thomas Christiano Te frst step is achieved by what Downs calls the “free information system.” Tis is information that we get for free in the sense that we do not acquire it by expending efort, time, or money for the purposes of politics itself. We get the information while we pursue other concerns. We get information about politics from doing our jobs, from talking to colleagues, friends, and family. We get information in school, which we are required to absorb. And we get information just because it is thrust upon us in everyday life on television, radio, or social media. It is from this information that we can begin to form the judgments about what further information it is worth trying to get and how to get it. Without the frst step, there is no way we would have a concern about politics or even be interested in knowing about politics.14 Diferences in the reception of free information play a large role in explaining diferences in the extent of political knowledge and thus in explaining inequality of power. Downs thought that the relatively afuent, on average, are more likely to have jobs that interact with law and government in sophisticated ways and so these people are more likely to start from a base of “free information” acquired from work. And they live in neighborhoods with good schools. Furthermore, afuent people interact with other afuent people who have similar sources of “free information” in diferent jobs. Tus, they learn more from their work and they learn from their co-­workers and family members about other political important facts. Hence, they start from a good base, which both supplies them with initial beliefs about politics and enables them to engage in much more efective pursuit of information to supplement those beliefs. In contrast, many ordinary workers, especially in the U.S., where few are unionized, are not the benefciaries of free information about politics at work. Tey do not need to know about law and government to do their jobs since their jobs are low skilled or completely directed by managers. Moreover, the quality of schools in their neighborhoods is signifcantly lower. And to the extent that the society is divided into classes, their friends and family are in similar positions. Consequently, the afuent have a better sense of what the marginal costs and marginal benefts are from further information and methods of saving information than others who have jobs that require very little information about law and politics. Tis is, then, a powerful source of inequality. Not the only one, but an essential one. And this inequality is a by-­product of the way that the division of labor in society is currently set up. Te second step involves further collection of information. And here we rely heavily on various methods for reducing the costs of collecting information, in all walks of life. As Downs observes, we rely heavily on friends, co-­workers, newspapers, opinion leaders, television, political parties, interest groups, and social media to receive information. But we also rely on these crucially to avoid having to pursue further information. Many of us, completely rationally, rely just on party afliation to determine who to vote for and we efectively advance our ends by doing so. Tis is rational when the parties are pretty far apart in terms of social aims and one has the sense that the ofcials in charge of constructing policy are reasonably competent and 14  See Downs (1957, p. 221f.). Tis is a crucial dimension of the information gathering process that is missed in Jefrey Friedman’s probing discussion of shortcuts in Friedman (2020, esp. ch. 6). It solves the logical problem he raises about why we would try to fnd shortcuts in the frst place.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Basis of Political Equality  129 trustworthy. We also rely on each other and opinion leaders to help us decide who to vote for without knowing many of the reasons for which the opinion leaders or friends favor the candidate. As long as party, friends, and opinion leaders make recommendations on the basis of reasons that are important to us, we do not need to know more. Here, too, there is an important source of inequality since the afuent have more resources that help them in the pursuit of better information. All of this implied for Downs that the better of would be better informed and more efective than the others. At the same time, there is good evidence that representatives and senators in the U.S.  tend to be very much more responsive to the concerns of relatively afuent individuals in their political decision-­making than to the concerns of others (Bartels,  2010; Gilens,  2014). Hence, it is reasonable to go with the hypothesis that the better information seems to pay of. One main point to remember is that it simply does not follow from the fact that people do not answer questions very well about politics that they do not make good choices in politics (Lupia, 2016; Kuklinski and Quirk, 2001). Te fact that only one-­ third of a population can answer basic questions about politics may be a sign of great efciency. Why not have diferent people hold diferent pieces of information? Again, we do this kind of sharing in markets all the time, why not do it in the case of democracy? And so, the evidence of ignorance is not strong enough to defeat the evidence given above that the disadvantaged have benefted greatly from participation in the political system.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

8.  Institutions Matter Te fact that social exposure to free information and that shortcuts are necessary and pervasive parts of almost all aspects of modern life does not entail that democracy must work well. What it shows is that each person is deeply dependent on networks of persons to make good decisions. Te key question is whether the networks of people and institutions are set up to be reliable sources of free information and reliable sources of information shortcuts. And the further question is whether the networks operate in such a way that there is a kind of egalitarian distribution. Tese refections point to the importance of institutions, which can shape the networks people belong to, for democracy. Te above arguments are not meant to suggest that all is fne in modern democracies. Tey are only meant to suggest that there can be rational and efective ways for people to make good decisions even with low levels of information. Tey point to ways democracies have been able to achieve what they have achieved. But they can also point to how democracies can be improved.15 With regard to free information, for example, Downs argues that the dishwasher and many persons who perform relatively low skilled jobs in strongly hierarchically governed frms will not have much in the way of quality free information from work and so will be at a disadvantage relative to professionals, business managers, and 15  Tis is the most important element in the response to skeptics about the value of shortcuts such as Somin (2014) and Friedman (2020). Te solution is not to try to get people to know more but rather to establish institutions that make information shortcuts more reliable.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

130  Thomas Christiano government employees. Hence, as a result of the structure of the division of labor alone, there will be signifcant inequality in politics. Tere may be ways in which the division of labor can be altered so as to bring ordinary workers into a higher quality free information stream. Tere has been a spate of recent studies that suggest that there are institutions from which working class and lower middle-­class people can beneft both economically and politically. Tese studies suggest that unions have played a signifcant role in making representatives and senators responsive to working class and middle-­class people, bucking the tendency pointed to by Bartels and Gilens. Te results have suggested that the responsiveness of governments, representatives, and senators to working class concerns has declined appreciably with the decline of unions. And they suggest that districts with strong union presence, show signifcantly less inequality of representation than districts with weaker union representation. Te efect seems to be through the activity of voting and not primarily through union funding of campaigns (Becher and Stegmuller, 2020; Flavin, 2016). Furthermore, it appears that part of the reason for this increase in responsiveness is that union members are better informed than non-­union members in the same jobs and industry about matters of importance to them both. Tey tend to take more nuanced views of political issues (MacDonald, 2019; Kim and Margalit, 2017). One might worry that the reason representatives are more responsive to lower middle class and poorer people in districts with high union density is because unions organize these voters politically and that collective action is powerful in getting politicians to pay attention. It is not information but organization that counts, one might say. But this is a false opposition. Organization works because of the increased information among workers that unions create and that tends to promote increased aggregation of information among workers. Unorganized workers lose the information about each other. Tis information aggregation function of unions is well documented in the role unions play in making it possible for health and safety laws to be enforced (Donaldo and Walde, 2012). And it is a large part of what makes collective action possible and efective. Tis is relatively new work, but it has been growing rapidly in the last fve years. It requires further analysis. But it does ft with the sophisticated Downsian model in which unionized workers receive a great deal of free but high-­quality information at their workplace. Te union must make good use of law and government to advance worker interests in the workplace and workers need to take account of this in the workplace; this information is then available for free in political contexts. Furthermore, the shop foor provides a good context in which information is shared among diferently informed persons. Tis may be an example of an institutional structure that can enhance worker’s understanding of politically important information in a rational way. It does this by improving the information networks ordinary workers participate in. More generally, various forms of worker participation in the governance of frms, including though not limited to union participation, may help overcome the kinds of problems Downs saw. By worker participation, I mean some kind of worker participation in the management of the frm. Tis means the addition of workers’ voices to the decision-­making of frms. Tis may involve participation in unions, it may involve some kind of co-­determination as is required in German frms of over

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Basis of Political Equality  131 2000 workers, it may involve worker ownership with some kind of control rights. I mean for the category to be quite broad. I think there is evidence that each one of these forms of worker participation can be value enhancing for frms.16 But diferent forms may be desirable for diferent kinds of markets.17 Recall that the Downsian problem is that there is an inalterable confict between equality and a complex division of labor. And that a signifcant part of this confict is derived from the fow of free information to the diferent positions in the division of labor. Some positions in the division of labor beneft from very high-­quality free information that enables people to go on to make good decisions concerning what information to look for and where to look for it. And the key diference is that some persons are in positions of control in frms, which exposes them to the fne grain of law and policy, while others are not. One can see here that one promising solution to the Downsian problem, if there is one, is to alter the division of labor.18

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

9. Worries One worry to which my proposal can give rise and which requires further refection is that it may seem to violate the impartiality that democracy ought to take towards people’s preferences and values (within substantial limits). Tis is because my proposal is recommending an alteration of the division of labor on the grounds that this will enable workers or lower middle class and poorer citizens to gain greater informational abilities in the political process and thus greater political power. But this thesis is committed to the idea that greater informational abilities with regard to economic organization is an especially prominent concern that should be accorded a special place in political organization, special enough to warrant a major intervention in the division of labor. Does this not introduce a kind of bias in favor of economic issues into our thinking about the political system? And does this proposal not take too much of a stance on a set of issues on which democratic societies ought to have a say? I think this worry should be taken seriously. But I want to say several things to mitigate the worry. First, the proposal leaves a lot of room for diferent views of employees so it does not take a particular stand on these. And the particular proposal leaves substantial room for diferent ways in which worker voice can be enhanced in the market. Second, the proposal merely emphasizes something that is quite important to the vast majority of people in a society, namely the legal organization of economic life. Tird, the proposal does not preclude other issues from having great importance in politics. Questions of morality, religion, education are not likely to be crowded out by a more sophisticated approach to economic organization. Fourth, greater 16  For the most wide-­ranging and careful economic study of worker managed frms, see Dow (2003). 17  For the locus classicus on the value enhancing qualities of unions, see Freeman and Madof (1984) and more recently, Freeman (2007); Boeri and van Ours (2013). 18  One can see some similarity here to the thesis of Carole Pateman in her ground-­breaking work, (1970), where she defends the educative role of worker participation as a way of enhancing participation in political democracy. My thesis here complements hers. While she focusses on the efect of worker participation on the sense of efcacy of workers, I am focusing on the ability of participation to enhance the informational abilities of workers.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

132  Thomas Christiano sophistication in economic matters and on the issues that relate to these matters should be to the advantage of political thinking about other matters. Another worry is that I have not given a conclusive demonstration that the approach recommended here will work to enhance ordinary citizens’ participation in democracy. I readily agree with that. Philosophically guided refections on economic and political institutions must be completed by social science. I have ofered some social science to suggest that the task may be worth pursuing, but conclusive proof is not available at this point and cannot be given by unaided philosophy.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

10.  Knowledge and Citizenship How much do citizens know about politics? And, how much ought they know about politics? Now that we have gone through two diferent aspects of the problem of knowledge, that is, the facts of disagreement and the collectivism of justifcation, we can see that both these problems are very difcult to get a clear handle on. And it is not clear to me that we can get a full understanding of these. It is essential that citizens are in charge of the values or aims that guide the making of policy as well as the trade-­ofs between those values. But it is also important that citizens have knowledge that the political system is actually pursuing those aims to the best of the abilities of its members. Citizens must not only be able to choose the aims of the system, they must also have some way of gauging whether it is pursuing those aims. But it is hard to know how to measure their understanding in the light of disagreement and the collective form of justifcation necessary to a society with a complex division of labor. We do have some evidence that citizens know enough to use power reasonably efectively, that is the conclusion of the frst section. But how do we determine that they each have knowledge that gives them equal power? Here perhaps the best we can do is to look at what kinds of institutional arrangements give people reasonably secure ways of informing themselves and then make sure that this kind of institutional access is available for everyone. And I include in that the free information system so necessary to one’s ability to become oriented in the political system. It may be that we do not and cannot have a measure of knowledge that can serve the purpose of ensuring equal political power. But if we have some signifcant understanding of the institutional conditions of how knowledge originates and develops, we might be able to ensure that everyone is in the position to participate in the development of knowledge. My discussion of the economics of information is meant to provide some of the basic elements of the circumstances under which one can become informed in political society. Tis requires more empirical research, but if it is on the right track we may be able to develop an institutional conception of equal access to the relevant information as a component of political equality.

11. Conclusion To conclude, the issues raised by skeptics about democracy are important ones to discuss. Tey go to an oddly neglected part of democratic theory, which is the way in

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Basis of Political Equality  133 which citizens can become or not become able to make good decisions in a political society. I have tried to show that pessimism about democracy is overstated and that there may be good grounds for thinking that democracy can work well despite the necessity that it work in a context of low information decision-­making. Te problem of low information decision-­making need not undermine democracy’s promise to advance the interests of citizens broadly. A careful attention to the institutional context in which citizens operate is necessary to make good on this promise.19

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Acemoglu, D. and J. Robinson. 2009. Economic Origins of Dictatorship and Democracy. Cambridge: Cambridge University Press. Acemoglu, D., S.  Naidu, P.  Restrepo and J.  Robinson. 2019. “Democracy Does Cause Growth.” Journal of Political Economy 127(1). Achen, C. and L.  Bartels. 2017. Democracy for Realists: Why Elections Do Not Produce Responsive Government. Princeton, NJ: Princeton University Press. Bartels, L. 2010. Unequal Democracy: Te Political Economy of the New Gilded Age. Princeton, NJ: Princeton University Press. Becher, M. and D.  Stegmuller. 2020. “Labor Unions and Equal Representation: Te Impact of Labor Unions on Legislative Responsiveness in the US Congress.” Perspectives on Politics July. Boeri, T. and J. van Ours. 2013. Te Economics of Imperfect Labor Markets 2nd edition. Princeton, NJ: Princeton University Press. Brennan, J. 2016. Against Democracy. Princeton, NJ: Princeton University Press. Campbell, A., P/ E. Converse, W/ E. Miller, and D/ E. Stokes. 1960. Te American Voter. New York: Wiley and Sons. Caplan, B. 2007. Te Myth of the Rational Voter: Why Democracies Choose Bad Policies. Princeton, NJ: Princeton University Press. Christiano, T. 2008. Te Constitution of Equality: Democratic Authority and Its Limits. Oxford: Oxford University Press. Christiano, T. 2011. “An Instrumental Argument for a Human Right to Democracy.” Philosophy and Public Afairs Spring. Christiano, T. and W.  Braynen. 2008. “Inequality, Injustice and the Leveling Down Objection.” Ratio December. Donaldo, A. and K. Walde. 2012. “How Trade Unions Increase Welfare.” Te Economic Journal 122(563): 990–1009. Dow, G. 2003. Governing the Firm: Worker’s Control in Teory and Practice. Cambridge: Cambridge University Press. Downs, A. 1957. An Economic Teory of Democracy. New York: Harper and Row. Fehr, E. and U. Fischbacker. 2006. “Te Economics of Strong Reciprocity.” In H. Gintis, S.  Bowles, R.  Boyd, and E.  Fehr (eds.), Moral Sentiments and Material Interests: Te Foundations of Cooperation in Economic Life. Cambridge, MA: MIT Press. 19 Funding from the ANR Labex-­IAST is gratefully acknowledged. I also want to thank Elizabeth Edenberg, Michael Hannon, Emily McGill, Houston Smit, Marc Fleurbaey, and Alex Guerrero for helpful comments on previous drafs of this paper or useful discussion of its ideas.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

134  Thomas Christiano Fischbacker, U. 2006. “A Model of Strong Reciprocity,” in H. Gintis, S. Bowles, R. Boyd, and E.  Fehr (eds.), Moral Sentiments and Material Interests: Te Foundations of Cooperation in Economic Life. Cambridge, MA: MIT Press. Flavin, P. 2016. “Labor Union Strength and Equality of Political Representation.” British Journal of Political Science 48(4). Freeman, R. 2007. America Works: Critical Toughts on the Exceptional US Labor Market. New York: Russell Sage. Freeman, R. and J. Madof. 1984. What Do Unions Do? New York: Basic Books. Friedman, J. 2020. Power without Knowledge: A Critique of Technocracy. Oxford: Oxford University Press. Gilens, M. 2014. Afuence and Infuence: Economic Inequality and Political Power in America. Princeton, NJ: Princeton University Press. Hardin, R. 2009. How Do You Know? Te Economics of Ordinary Knowledge. Princeton, NJ: Princeton University Press. Hutchins, E. 1995. Cognition in the Wild. Cambridge, MA: MIT Press. Kim, S.  E. and Y.  Margalit. 2017. “Informed Preferences? Te Impact of Unions on Workers’ Policy Views.” American Journal of Political Science 61(3): 728–43. Kuklinski, J. and P.  J.  Quirk. 2001. “Conceptual Foundations of Citizen Competence.” Political Behavior 23(3), Special Issue: Citizen Competence Revisited (Sept.): 285–311. Lopez-Guerra, C. 2014. Democracy and Disenfranchisement: Te Morality of Electoral Exclusions. Oxford: Oxford University Press. Lupia, A. 2016. Uninformed: Why People Know So Little about Politics and What We Can Do about It. Oxford: Oxford University Press. MacDonald, D. 2019. “How Labor Unions Increase Political Knowledge: Evidence from the United States.” Political Behavior April. Mackie, G. 2014. “Rational Ignorance and Beyond.” In J. Elster and H. Landemore (eds.), Collective Wisdom. Cambridge: Cambridge University Press. Maskivker, J. 2019. Te Duty to Vote. Oxford: Oxford University Press. Olson, M. 1965. Te Logic of Collective Action: Public Goods and the Teory of Groups. Cambridge, MA: Harvard University Press. Parft, D. 1984. Reasons and Persons. Oxford: Oxford University Press. Pateman, C. 1970. Participation and Democratic Teory. Cambridge: Cambridge University Press. Pincione, G. and F. Teson. 2006. Rational Choice and Democratic Deliberation: A Teory of Discourse Failure. Cambridge: Cambridge University Press. Popkin, S. L. 1991. Te Reasoning Voter: Communication and Persuasion in Presidential Campaigns. Chicago: University of Chicago Press. Roemer, J. 2010. “Kantian Equilibrium.” Te Scandinavian Journal of Economics 112(1). Schumpeter, J. 1950. Capitalism, Socialism and Democracy 3rd edition. New York: Harper and Row. Somin, I. 2014. Democracy and Political Ignorance: Why Smaller Government is Smarter Stanford, CA: Stanford University Press. Wright, G. 2013. Sharing the Prize: Te Economics of the Civil Rights Revolution in the American South. Cambridge, MA: Harvard University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

8

Does Public Reason Liberalism Rest on a Mistake? Democracy’s Doxastic and Epistemic Problems

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Jason Brennan

Te theory of public reason liberalism was created to solve a particular problem. But what if the problem is largely an illusion, based upon a problematic and controversial theory of political psychology and voter behavior? In Jonathan Quong’s (2017) words, public reason liberals believe that people “are not naturally subject to any other person’s moral or political authority.” Public reason liberals maintain that people have a diverse range of moral, religious, and political views. Tey claim citizens deeply dispute the nature of the good and the just, and also dispute which institutions, policies, and practices best realize their normative goals. If these conditions obtain, this supposedly generates a normative problem: how could uniform moral or political rules be imposed upon all of us, without treating us as unfree or unequal, without running roughshod over our reasonable moral disagreements about the good and the just, and without running roughshod over our reasonable empirical disagreements about how to realize our moral goals? In response to this problem, public reason liberals generally defend some version of the “liberal principle of legitimacy,” a principle which constrains the imposition of moral rules, rules of justice, or coercive political and social policies. Here are three leading statements of the liberal principle of legitimacy: 1. Gerald Gaus’s (2004, p. 227) version: “A’s coercive interference with B is permissible only if there is a justifcation for it that B may reasonably be expected to endorse.” 2. David Estlund’s (2008, p. 33) version: “No one has legitimate coercive power over another without a justifcation that could be accepted by all qualifed points of view.” 3. John Rawls’s (1996, p. 137) version: “Political power is legitimate only when it is exercised in accordance with a constitution the essentials of which all citizens as free and equal may reasonably be expected to endorse in the light of principles and ideals acceptable to their common human reason”. Public reason liberals dispute both the broad and fne details of their theory and how to interpret the liberal principle of legitimacy (see Vallier, 2018). Still, public reason libevrals agree (a) that people have diverse political views and deep political Jason Brennan, Does Public Reason Liberalism Rest on a Mistake? Democracy’s Doxastic and Epistemic Problems. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Jason Brennan. DOI: 10.1093/oso/9780192893338.003.0009

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

136  Jason Brennan disagreements, but (b) in principle it is possible to implement at least some institutions, laws, norms, and policies without subjugating anyone. Notice the theory presupposes certain empirical claims about citizens’ beliefs and political psychology. It presupposes most people both have a robust set of politically-­ salient normative and political beliefs and that they tend to have signifcant disagreements with each other about those beliefs. In empirical political science, there are two major paradigms of political psychology and voter behavior, which Christopher Achen and Larry Bartels (2016) label “populism” and “realism.” In short, populists hold that voters are genuinely ideological, they have many politically-­salient normative and empirical beliefs, they vote on the basis of those beliefs, and politicians in some way tend to enact the policies which voters advocate. In contrast, realists hold most voters are not genuinely ideological. Realists hold that most citizens lack politically-­salient normative and empirical beliefs; they vote for other reasons. Political outcomes are not generally a function of citizens’ beliefs. I worry that public reason liberalism rests not merely on the populist model of democracy, but on a strong populist model which few empirical political scientists would defend. Tis chapter instead asks whether we can make sense of the public reason project in light of democratic realism. I do not argue here that democratic realism is correct, though I think it is. However, realism is an important and leading school of thought within empirical political science (e.g. Lippmann,  1922; Campbell et al.,  1960; Converse,  1964; Tajfel,  1982; Delli-­Carpini and Keeter,  1996; Caplan, 2007; Oppenheimer and Edwards,  2012; Somin,  2013; Achen and Bartels, 2016; Brennan,  2016; Kinder and Kalmoe,  2017; Mason, 2017, 2018). Public reason liberals may—sincerely or out of necessity—reject realism, but the foundations of their project are shaky if they rest on a controversial empirical model of political psychology, a theory which many leading political scientists reject. Indeed, contemporary populists ofen accept large swaths of the realist critique and have modifed their views accordingly. As far as I can tell, public reason liberals ofen ignore the relevant empirical work in political science; they write as if one of the leading views of voter behavior and political psychology simply does not exist or has no bearing on their work. In this chapter, I frst briefy summarize the realist theory of political psychology and voter behavior. I then show how it makes trouble for public reason liberalism. Finally, I conclude by examining some possible ways to reconcile the public reason project with realism, though none are promising.

1.  Te Realist Project: People Vote for Who Tey Are, Not What Tey Want Realism is in part a reaction to and a critique of populism or the popular sovereignty theory of elections. In this section, I will summarize the basics of both populism and realism. In order to heighten the basic contrast, I will describe both of them in rather extreme forms, though many actual populists and realists hold weaker and more nuanced versions of their theories.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Does Public Reason Liberalism Rest on a Mistake?  137 In a recent New York Times op-­ed, Kwame Anthony Appiah (2018) says, “People don’t vote for what they want. Tey vote for who they are.” Tis is an excellent two-­ sentence summary of the distinction between realism and populism. Populists say voters vote for what they want. Realists, like Appiah, say people vote for who they are. Note that political theorists and others sometimes use the words “populism” and “realism” to refer to unrelated ideas. My usage here follows Achen and Bartels, 2016, among others. But one should be careful when reading other papers using these terms, as they may be talking about entirely distinct issues and debates. Populism holds that (a) ordinary citizens have ideologies or preferences about policies, laws, and political outcomes, and (b) that representatives have a signifcant tendency to implement those ideologies and satisfy those preferences (Dahl,  1956,  1998; Pitkin,  1967; Tompson,  1970, 1987; Stimson et al., 1995). Populists claim that individual voters begin with their interests—whatever things they care about and believe are normatively signifcant. On the basis of their interests and the available empirical evidence, voters then form ideologies or stable political beliefs, including both empirical beliefs (e.g. “Capitalism causes inequality”) and higher-­ level normative beliefs (e.g. “Capitalism is unjust”). Voters examine the various political parties and candidates. Tey then vote for those they believe best serve their political goals or which share their political beliefs. Teir votes signifcantly determine what governments do; voting induces governments to implement citizens’ policy preferences and respond to voters’ normative goals. Tis is the folk theory of democracy schoolchildren learn in sixth-­grade civics classes. Populism is a highly cognitive, doxastic, and epistemic theory of citizens’ political psychology. It holds that citizens tend to have relatively stable and robust political and moral worldviews, if not necessarily something as robust as what Rawls (1996) calls a “comprehensive moral view.” Tis view holds that votes have epistemic value, indicating not merely what voters genuinely want, but also transmitting information about how they believe various policies would work. On the populist model, if voters are decently informed, then we would expect democratic decisions to track the truth, to show us which policies or politicians will best support voters’ underlying goals. Realists Christopher Achen and Larry Bartels, among many other political scientists, argue that few citizens ft the populist model. In contrast, they claim, identity dominates politics; nearly all politics is identity politics. Citizens vote largely on the basis of partisan loyalties. Tese partisan loyalties are grounded in their identities, but do not track ideology, sincere policy preferences, or their interests. Rather, partisan attachments usually result from accidental, historical connections between certain identity groups and certain political movements and parties. Certain groups become attached to certain parties, but not because they believe in what those parties do or because those parties tend best to serve their interests and goals. Most citizens simply lack underlying ideological commitments or signifcant political views—the very kinds of views public reason liberals regard as leading to the problem of public justifcation. Realists claim that most citizens are not ideological. Many citizens will, if surveyed, describe themselves as “liberal” or “conservative,” but most self-­described liberals and conservatives lack stable beliefs ftting those ideological self-­descriptions. Most citizens lack stable political beliefs at all. Only a minority of citizens have organized

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

138  Jason Brennan and stable sets of political beliefs. Most have few opinions and even fewer stable opinions, and what few opinions they have cannot be amalgamated into a coherent position. Contrary to what news reporters claim, there are few “single-­issue” voters (Converse, 1964; Barnes, 1971; Inglehart and Klingemann, 1976; Arian and Shamir, 1983; Bartels, 1986; Converse and Pierce, 1986; Zaller, 1992; McCann, 1997; Goren, 2005; Zachmeister, 2006; Muddle, 2007; Lewis-­Beck et al., 2008; Achen and Bartels, 2016; Kinder and Kalmoe, 2017; Mason, 2018). Realists claim that for most citizens, voting and political afliation are identity-­ based, rather than based on belief or opinion. Te best analogy here would be sports team afliation. A person from Boston becomes a Red Sox and Patriots fan because that is what Bostonians do. Tey wear Pats shirts to signal or express to one another they are a loyal part of a particular network, and so can be trusted as potential business partners, mates, or neighbors. Being a Red Sox or Patriots fan serves their interests not because the teams themselves help them, but because rooting for the team helps them in their personal lives. People are more trusting and friendly towards those with similar sports team preferences. It is not as though Bostonians frst form strong opinions about baseball and football, then examine the teams on ofer, and fnally select teams based on how well those teams respond to or realize their pre-­existing values. Rather, they root for the Red Sox and Pats because those teams are historically attached to their hometowns. In an alternative world, where the Yankees (including all historical Yankees players) played in Boston, Bostonians would wear pin stripes, root for the Yankees, and proclaim their love for Mickey Mantle, Derek Jeter, and C. C. Sabathia. Democratic realism holds that for most citizens, political afliation is psychologically equivalent to sports team loyalty. Teir political afliations are little more refective of their moral views, views of justice, or views of the good life than their commitment to the Red Sox or Patriots. Teir political behavior refects their demographic identities (such as “Boston Irish” or “evangelical Christian”). Most citizens possess some regulating self-­identity (which may be complex), such as “college professor” or “Boston Irish” or “Southern evangelical Christian.” Politically-­ active citizens learn how others with the same regulating identity vote, and then usually vote the same way (Somin,  2013; Achen and Bartels, 2016; Mason,  2018; Simler and Hanson, 2018). Realism holds that various identities or demographic groups become attached to particular political parties for largely accidental historical events or circumstances that have little to do with voters’ underlying values or interests (Achen and Bartels, 2016, pp. 213–66; Campbell et al., 1960; Tajfel, 1982; Greene, 1999; Gamm, 1989). For example, American Jews switched from Republican to Democratic loyalties between 1928 and 1940, not because of policy platform changes, the Wall Street Crash, or ideological changes, but because of reduced antagonism between Jews and Catholics in the 1930s (Achen and Bartels, 2016, pp. 236–40). On the realist model, when voters use identity to determine afliation, they are not making use of cognitive shortcuts or heuristics. If a Boston Irish person votes Democrat because other Boston Irish citizens do, this is not because Democrats have historically been better for the Boston Irish (or their moral concerns) than the Republicans. In the same way, Clevelanders root for the Browns, but certainly not because the Browns are better for them than, say, the Chiefs.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Does Public Reason Liberalism Rest on a Mistake?  139 On the realist model, being a Democrat serves voters’ interests not because the Democrats themselves help those voters, but because rooting for the Democrats helps those voters in their personal lives. Te benefts are much the same as publicly demonstrating sports team loyalty. For instance, consider how my fellow academics would treat me if I announced I vote Republican. (Don’t worry, I don’t.) On the realist picture, most voters stop here. Most voters mostly lack political beliefs. Non-­doxastic factors explain their political behavior. However, realists add, a sizeable minority of citizens do express stable commitments to ideologies and political beliefs. However, realists claim, for most of these seemingly ideological citizens, this amounts to a post-­ hoc, superfcial endorsement of their parties’ platform. Tese citizens engage in a mental process that “feels like thinking”; they vote for their party because that is what people like them do, but they later learn what the party stands for and claim they also stand for it. Tese voters have no deep, interest-­based commitment to their party’s ideology. Most of these citizens cannot quite be said to “believe” in their party’s ideology or platform. Tey will express commitment to it, but most will immediately express commitment to a new platform or ideology if their party changes, unaware that they have even “changed their minds.” Tey will claim they have always believed these new or opposite opinions (Achen and Bartels, 2016, pp. 267–96; see also Lenz, 2009, 2012). Tus, on the realist picture, while a minority of citizens express more stable ideological beliefs, most of these citizens are not deeply committed to their ideologies. To extend the sports team analogy, consider how a New England Patriots fan might subconsciously “reason” about his color preferences: “I’m from the Boston area” → “Bostonians root for the Pats” → “Te Pats’ colors are silver, blue, and red” → “I like silver, blue, and red.” Similarly, most ideological voters “reason” in the same way: “I’m a Boston Irish guy” → “People like me vote Democrat.” → “I’ll vote Democratic” → “Te Democrats want to increase regulation; so, I’m pro-­regulation.” But even this last part is misleading. Most of the citizens in question are not really pro-­regulation. For most citizens, saying “I’m pro-­regulation” is not expressing a belief about appropriate responses to market failure; it is instead expressing a commitment to seeing the Democrats win. It expresses one’s membership in and loyalty to various social groups. Te implicature, if not meaning, of “Let’s increase regulation” is “Yah, Democrats!” or “I’m one of you.” (Tese statements are still true or false, but the person uttering them utters them for performative reasons and does not really believe the statements.) Finally, realists hold that a small minority of voters are genuinely ideological, with sincere political beliefs and with political afliation based on such beliefs. Tis minority fts the populist model above. Democratic realism creates a doxastic challenge to the public reason project: Te Doxastic Challenge: Public reason liberals claim we must be able to in some way base any justifcation for coercive moral and political rules on reasonable citizens’ underlying beliefs. However, if realism is correct, most citizens have few real beliefs. Most who do have political beliefs do so in way that does not relate to or track their deeper concerns; they “wear” their political beliefs the way they wear sports colors, but they are not genuinely committed to such beliefs. So, there is nothing there upon which to base public justifcation.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

140  Jason Brennan Public reason liberals ofen claim there is an analogy between disputes in contemporary politics and the disputes which led to the European wars of religion (e.g. Rawls, 1996; Weithman, 2011; Gaus, 2004; Vallier, 2018). European principalities and states faced the problem of achieving legitimacy among people with diverse religious views, many of whom believed the state’s legitimacy depends on having a strong public connection to the “correct” Christian sect. Public reason liberals say that just as there was the problem of justifying a secular political order in the face of widespread religious disagreement, so there is the problem of justifying a political order in the face of widespread religious and secular disagreement about the good, the right, the just, and empirical matters about which means achieve which ends. But if realism is correct, this analogy may fail, because most people lack the right underlying political beliefs, and their seeming political disagreements are superfcial and inauthentic. What political liberals regard as widespread political disagreement is not real disagreement, because citizens’ public political expressions are not reports of their beliefs, and so we should not attribute to most citizens the beliefs they appear to endorse in public. Realists also hold that voters tend to be deeply ignorant, misinformed, and epistemically irrational. Te typical citizen, and even the modal voter, is ignorant of most basic political information (Campbell et al.,  1960; Converse,  1964; Delli-­ Carpini and Keeter,  1996; Friedman,  2006; Caplan, 2007; Somin,  2013; Brennan,  2016). American citizens generally cannot identify which party controls Congress, do not know who their senators and representatives are, do not know about major recent policy changes, do not know what is in the federal budget, cannot estimate major economic or social indicators (such as unemployment or crime rates) with any degree of precision, and have little idea what diferent parties or candidates have done or propose to do (Delli-­ Carpini and Keeter,  1996; Somin,  2013; Brennan, 2016). Tey are not just ignorant; rather, many have mistaken beliefs about both basic political facts as well as social scientifc issues (Bartels, 1996; Althaus, 2003; Caplan, 2007; Gilens,  2012; Caplan et al.,  2013). Further, voters generally do not reason about politics in a scientifc or truth-­tracking way. Rather, they engage in motivated reasoning; they tend to believe what they want to believe rather than what the evidence supports. Tey sufer from both conformation bias (they tend only to see out, to pay attention to, and to accept information that reinforces their current beliefs) and disconfrmation bias (they tend to reject or ignore information that undermines their current beliefs). (Kahneman and Tversky, 1973; Kahneman et al., 1982; Rasinki, 1989; Bartels, 2003; Arceneaux and Stein, 2006; Westen et al., 2006; Kelly, 2012; Haidt, 2012; Chong, 2013; Lodge and Taber, 2013; Taber and Young, 2013; Erison et al., 2014.) Widespread political ignorance and misinformation also create problems for the public reason project. Afer all, on many liberal conceptions of the “reasonable,” citizens must have a certain degree of epistemic rationality, which they may in fact lack. Further, to the extent that citizens endorse policies on the basis of what they ought to know are mistaken beliefs about facts, this creates a normatively signifcant degree of separation between citizens and the policies they endorse. Tat is the realist picture in a nutshell. We should ask: If the realist picture is correct, is there much work for public reason liberalism to do? Can we make sense of

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Does Public Reason Liberalism Rest on a Mistake?  141 public reason in light of this empirical paradigm? If public reason liberals are instead committed to the populist paradigm, then their theory can be refuted, or at least rendered irrelevant, by empirical work on voter behavior.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  X- and Y-­People Consider John Tomasi’s helpful “alphabet people” categorization of kinds of people among whom the public reason project is meant to facilitate compromise and consensus (Tomasi, 2001, p. 17). Tomasi’s “A-­people” value autonomy, creativity, and individuality; they are highly self-­aware, refective, deliberative, and ethically liberal. D-­people (such as Nazis or theocratic citizens) are the other extreme; they are radically communitarian and authoritarian, reject political liberalism altogether, and wish to impose their comprehensive religious and/or secular views upon others through government coercion, regardless of dissent. D-­ people are considered “unreasonable” or “unqualifed” in the public reason project, and so the liberal principle of legitimacy does not require that coercive principles be justifed to them. Next to D-­people are C-­people. C-­people may be “citizens of faith” (Rawls, 1996) or “reasonable romantics” (Larmore,  1990). Tey afrm comprehensive religious or traditionalist moral doctrines in which there are strong moral and status hierarchies, and may reject many ethically liberal ideas (e.g. by holding abortion is evil or that mothers should work at home). However, they are “reasonable” because they accept political liberalism and thus believe it is wrong to impose traditional moral rules or religious/communitarian ideas of personhood through politics. Finally, B-­people are ideologically between the As and Cs (Tomasi, 2001, pp. 17–19). Tey have weaker commitments to their not quite liberal, not quite traditional views. Tey fnd elements of both ethical liberalism and ethical communitarianism/traditionalism plausible. Like the A- and C-­people, they are also committed to the public reason project. Tomasi (2001, p. 19) asserts that “the great sweep of citizens” in modern democracies are B-­people. Te public reason project is designed to deal with A- through D-­people. It has seemingly intramural debates, of course: Who count as the D-­people, the ones who get excluded from public justifcation? How and in what ways must we accommodate reasonable citizens of faith? Public reason liberals dispute the answers, but the theory is explicitly about such belief-­flled people. Realism complicates this picture dramatically. Te realist contends that few people—maybe only about 10 per cent—in modern democracies are A-, B-, C-, or D-­people, though many more people talk like they are. Instead, we must add at least two more categories of letter people: X-­people: Citizens who are largely non-ideological. Tey have few political opinions, and lack what Rawls would call comprehensive moral, religious, or political views. What few opinions they have are largely ephemeral and unstable. Tey may be attached to a particular party, but that merely refects arbitrary historical connections between that party and their identity; they do not endorse their parties’ politics.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

142  Jason Brennan

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Y-­people: Citizens who post-hoc rationalize and assert that they agree with their party politics, but whose ideological commitments are superfcial. Like the X-­people, these citizens have a sense of identity, and their identities are connected to particular political parties for arbitrary, historical reasons. Unlike X-­people, Y-­people later learn what positions their parties advocate, and will claim that they too advocate it. However, their apparent ideology and political beliefs are feeting and shallow. When their party changes platforms, they will change “beliefs,” usually without any awareness that they have “changed their minds”. Like the X-­people, they also lack normatively signifcant moral, religious, or political views, despite how loudly they seem to advocate such views.

Te main diference between the A-, B-, C-, and D-­people versus the X- and Y-­people is doxastic: A- through D-­people hold the kinds of beliefs which the public reason project regards as normatively signifcant. X- and Y-­people do not. Tey are, in Lilliana Mason’s (2018) apt description, “ideologues without issues”; they are strongly committed to their parties despite not advocating anything or sharing their platforms or values. Te Y-­people seem to hold political beliefs, but it turns out on further inspection that their expressed commitment to such beliefs is insincere and superfcial. Tey express commitment to demonstrate membership in valuable social groups, not because they really believe what they say. Or, to paraphrase Kevin Simler and Robin Hanson (2018), for most people, politics is not about policy. Y-­people seem to have political beliefs, but they hold their “beliefs” in a weird way. Teir beliefs, if we can even call them that, do not refect real commitments about the good or just, or about how the world works. Instead, expressing political “beliefs” is largely equivalent to wearing sports team colors; they are a form of conspicuous display intended to show membership in what are, for that voter, socially advantageous groups. For them, advocating a policy is like wearing the Patriots’ blue and silver or waiving the Steelers’ terrible towel. Consider: Fans of the New England Patriots loudly proclaim that Tom Brady is the Greatest of All Time (GOAT), and loudly sneer at the Colts, Jets, Giants, or Bills. Tese conspicuous displays are part of Patriots fandom and demonstrate to others (and oneself) that one is indeed a fan. But now consider: While it seems like such fans are committed to certain beliefs and attitudes about the sport and their team, many of these commitments are superfcial. For instance, if the Pats change their colors to green and purple, then (afer some complaining), the fans will change their “favorite colors” too. When I wrote the frst draf of this paragraph, I predicted that if Tom Brady joined another team, fans who previously said he was the GOAT would now say he is overrated. While I have not done a scientifc survey, the Patriots fan Facebook pages indicate this behavior is indeed widespread. Tey are committed to the view that the Pats should win; most of their other proclamations and behaviors are just expressions of that view. Te realists say the same goes for most citizens in modern democracies. Te identity “southern evangelical Christian” is (for largely arbitrary reasons) attached to the Republican Party; the identity “Jew” is (again for largely arbitrary reasons) attached to the Democratic Party. A proper subset of such citizens—the Y-­people—become

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Does Public Reason Liberalism Rest on a Mistake?  143 ideologically-­expressive voters; they will claim post-­hoc that they share their party’s political beliefs upon learning what their party believes. However, their commitment is no deeper than Patriots fans’ commitments to blue, red, and silver; if the parties were to change policy platforms, most of their “ideological” voters would claim they agree with the changes, and some would claim they believed such views all along (Taber and Lodge, 2013; Achen and Bartels, 2016). Even apparent key issues, such as views for or against abortion, or for or against free trade, are largely expressive proclamations intended to demonstrate membership in the group. To illustrate, consider how many Republicans switched their “views,” seemingly without cognitive dissonance, on a wide range of economic subjects when Trump came to power.1 Tey went from pro-­free trade to protectionist almost overnight, without awareness that they switched. On the realist view, when a typical Democrat or Republican expresses her “beliefs” about justice or politics, we should regard this not as evidence of reasonable pluralism about the good and right, but instead regard their behaviors as expressions of commitment to seeing their particular political team win, without any deeper commitment to what that team stands for. What seems like the expression of belief— and thus what seems like evidence of widespread doxastic disagreement and value pluralism—is instead the conspicuous display of party afliation which generates external benefts for voters, by helping them form and maintain friendships, romantic attachments, and fnancial partnerships outside politics.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Te Justifcation Problem for X- and Y-­People Te public reason project’s central thesis is that coercive moral and political rules must be justifed to the people subject to those rules, in some way that those people can themselves recognize and accept in light of their current attitudes and beliefs. For the public reason liberal, to justify some political principle or policy P, it is not enough that P turns out to be true or that there exists a sound argument for P; rather, one must be able to show people they have sufcient grounds to accept P in light of their extant beliefs and commitments (Quong, 2017). Gaus says that public justifed norms must in some way be “intelligible” to reasonable agents “by their own lights” (Gaus, 1996, p. 114, 2004, p. 106, 2010, p. 29). As Kevin Vallier (2018) summarizes, pubic reason liberals share a “core assumption”: A core assumption of public reason liberalism, the political theory associated with the idea of public justifcation, is that people’s reasons for action and belief can difer substantially. Consequently, a reason must in some sense be relative to an agent’s beliefs, values and other commitments.  (Vallier, 2018)

While public reason liberals dispute the exact details of this problem, we can state the problem as follows: 1  https://www.vox.com/science-­and-­health/2017/11/17/16585982/psychology-­memory-­polls-­trump.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

144  Jason Brennan Te public reason problem: For any potential political policy or moral rule P, how do we show individual reasonable citizens that they have sufcient reason to endorse P, in light of their current doxastic states? How do we show that P hooks on to what they already believe and endorse?

Public reason liberals dispute what counts as sufcient reason, which beliefs matter, and to what degree we may idealize actual citizens. Tey also dispute whether the arguments or reasons for P must be accessible or understandable to citizens, or whether the arguments must simply be grounded in premises the citizens already accept. But this is the basic framing. Te realist, however, introduces a new spin on the problem: Te realist version of the public reason problem: Most citizens lack sufciently robust doxastic states upon which to “hook on” a justifcation for P. If citizens lack the right kind of underlying beliefs, how can we justify P to them?

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Realism signifcantly complicates the public reason project. Justifying moral or political norms requires ofering public reasons. Tese reasons must be relative to an agent’s beliefs, values, and commitments. Yet since many or most agents lack the right kind of beliefs, values, or commitments, there may be no sense in which we can justify policies and norms to them. We cannot make norms intelligible to them “by their own lights” if they have no lights. We cannot ground an argument for P on shared premises if citizens accept no relevant premises. We cannot ofer them reasons that hook onto their beliefs if they have none. In this section, I discuss some possible responses the public reason liberal might give to this problem. Since there are dozens of variations of public reason liberalism, I paint with broad strokes.

3.1  Proclaim that X- and Y-­People Are Unreasonable Public reason liberals say that policies must be acceptable to all “qualifed points of view” or to all “reasonable people.” All public reason liberals then exclude various ideologies, points of view, and people from counting as “qualifed” or “reasonable.” Some liberals—the more liberal ones—are more inclusive than others. One potential way to handle X- and Y-­ people would be to declare them unreasonable or unqualifed. Tis response has ample precedent. Indeed, certain public reason liberals are committed to making this very move. For instance, recall John Tomasi’s categorization of the A- through D-­people. Tomasi regards the the A-, B-, and C-­ people as reasonable because they endorse the liberal principle of legitimacy. For Tomasi, what makes a person reasonable just is sufciently strong commitment to the public reason project. John Rawls and Samuel Freeman at times endorse the same position; they ofen write as if what makes views or people reasonable (and, in their view, liberal) is that they have some strong enough commitment to some version of the public reason project (Rawls, 1996, p. 217; Freeman, 2001). For instance, this explains why Freeman regards some libertarians—Robert

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Does Public Reason Liberalism Rest on a Mistake?  145 Nozick and Murray Rothbard—are unreasonable but thinks others—John Tomasi and Gerald Gaus—are reasonable (Freeman, 2001, 2017). X- and Y-­people—the vast majority of citizens—are not public reason liberals. Tey lack stable commitments to more basic political beliefs such as “free trade is good,” let alone the complicated and esoteric beliefs which constitute public reason liberalism. Tis should be no surprise. Even populist-­leaning empirical political scientists would fnd it surprising if citizens understood, let alone were committed to, a highly sophisticated political meta-­ideology like public reason liberalism. Even contemporary populists agree most citizens have relatively basic political attitudes. Asking whether citizens endorse or reject public reason liberalism is a bit like asking whether they endorse or reject the second fundamental theorem of calculus. Instead, a well-­informed contemporary populist might hold that most citizens have stable and sincere commitments to certain liberal rights, such as a free speech and free association, and/or to general norms of tolerance. But few citizens advocate public reason liberalism per se. Few have beliefs which logically imply or presuppose public reason liberalism. So, even on the most plausible version of the populist view, let alone the realist view, it is tempting—given Tomasi’s or Freeman’s view of “the reasonable”—to conclude most citizens are not reasonable in the appropriate sense. Citizens are not doxastically sophisticated enough to endorse public reason liberalism, explicitly or implicitly. Tis move—calling X- and Y-­people unreasonable—has the virtue of rendering the X- and Y-­people a non-­problem for public reason liberalism. Public reason liberals will merely have to modify their position somewhat, and admit their theory is about justifying norms and policies to the A- through C-­people, a small minority of the democratic public. However, this response trivializes the public reason project. Indeed, it is a reason to avoid holding, as Tomasi or Freedom do, that the “reasonable” or “qualifed” just are those people explicitly or implicitly committed to public reason liberalism. Te fundamental motivation behind public reason liberalism is to respect individuals as free and equal, and, as a consequence, to put up justifcatory barriers to imposing rules and policies on such people. If it turns out that these justifcatory barriers protect only a limited class of professional philosophers or laypeople who endorse the controversial theory of public reason liberalism, then the theory is implausible. As Quong (2017) complains, if the public reason project excludes too many people, then it may seem to be a “sham”; it is not “an impartial or neutral method of moral or political justifcation, but it is, in fact, a form of sectarian secularism or modern liberalism masquerading as something more inclusive.”

3.2  Hold that Public Reason Need only Defeat Objections Te realist worries, how can we publicly justify policies if most citizens lack the right kind of underlying beliefs? One way to deal with this question is to weaken the requirements of justifcation.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

146  Jason Brennan Gaus (1996, p. 31) claims that to justify policy P to some person, this person (a) must have strong enough positive reasons to endorse P, and (b) must lack undefeated objections to P. In other words, justifcation requires both (a) the presence of strong reasons to endorse P and (b) the absence of undefeated objections to P. But one possible way to overcome realism’s doxastic challenge is to adopt a weaker view of public justifcation, which requires B but not A. Tat is, one might hold that to justify some policy, we must merely show that citizens lack strong reasons to reject that policy, by their own lights. More formally, we might interpret public justifcation as follows:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Public justifcation as overcoming reasonable objections: A policy P is sufciently justifed to an individual when that individual has no good undefeated objections to P, in light of that citizens’ doxastic states and attitudes.

Tis position has some intuitive plausibility. Afer all, the abstract idea of public reason is that we are not supposed to ram our favored policies or norms down people’s throats. Diferent reasonable people, whoever they are, will have diferent objections to any norm or policy we may wish to justify. If we can defeat their actual objections in some suitable way, then, plausibly, we have done all we need to do to justify the norm or policy. If they have no standing objections, then they have no complaint, even if they lack a positive reason to endorse the policy or norm in question. Consider, for instance, a policy restricting immigration. A realist (and indeed many populists) might say that half the country does not really care either way. Another 20 per cent will strongly express support for the restriction, while another 20 per cent will express disapproval, but they are not genuinely committed to such views. (For them, saying “We must build a wall” really just implicates “Hooray, Republicans!.”) So, for around 90 percent of the citizenry, overcoming objections is easy. Te problem of overcoming objections applies only to the minority. Like the last proposal, this new proposal partly trivializes the problem of justifying coercive policies to citizens. Since most citizens lack the relevant beliefs, they lack objections, and so therefore lack undefeated objections. On this proposal, it becomes unclear why public reason liberalism is such an important theory of justifcation or why the liberal theory of legitimacy matters as a requirement for a free, democratic society. However, unlike the last proposal, this new proposal at least does not render public reason liberalism a sectarian, exclusivist view. Regardless, many public reason liberals would not want to endorse such a limited view of justifcation, even if it helps them overcome the problems realism presents. Many, such as Stanley Benn (1988), Rawls (1996), Gaus (1996,  2004,  2010), and Vallier (2016,  2018), hold that coercion is inherently morally problematic, as it involves an exercise of authority by some over others. Indeed, public reason liberals usually begin with the thought that coercion is bad in itself. Te fact that a policy or norm is coercive is by itself a reason not to adopt a policy. For many public reason liberals, the lack of other objections to coercion is never sufcient to show a coercive policy is justifed. Rather, there must always be some strong positive reason to implement a coercive policy, a reason which outweighs the primitive and inherent

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Does Public Reason Liberalism Rest on a Mistake?  147 badness of coercion. Tis positive reason must itself hook onto citizens’ underlying moral and political beliefs. So, for Benn, Rawls, Gaus, and Vallier, among others, the doxastic problem of realism remains intact. Tey would not want to overcome the realist problem by weakening the requirements of public justifcation.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.3  Ground Justifcations on even More Basic Moral Views Gaus would reject the last proposal. Recall, he holds that to justify a policy P to a reasonable individual, that individual must both (a) have good enough positive reasons to endorse P and (b) lack undefeated objections to P.  As we just saw, the realist claims the second half of this conjunction is easily—maybe too easily—met for X- and Y-­people. However, Gaus’s view still requires that X- and Y-­people have sufciently strong positive reasons to endorse P, in light of their current beliefs. Tese positive reasons are relative to individual reasonable citizens’ doxastic states. It is not enough that there exists—say in academic journals or in Platonic heaven—some sound social scientifc and moral argument for P. Rather, the reasons to endorse P must be suitably accessible or intelligible to the citizens subject to P. As we discussed in Section 3, though, X- and Y-­people lack sufciently robust political beliefs that it remains unclear what we can “hook onto” to justify P to them. Te realist agrees that many of them, on the surface, may seem to have ideologies or at least stable sets of political and moral beliefs—the kinds that could be used as premises in a political argument. But they “wear” those beliefs the way sports fans wear team colors; they have no actual commitment. Most citizens do not even get that far, according to the realist. Tey might vote this way or that, but they do not even wear their team’s ideology. However, the public reason theorist might respond, it is not as though 90 percent of the public is completely agnostic about basic moral concerns. Sure, most lack ideologies or political theories. Most do not have comprehensive moral or philosophical views. Most lack opinions about most social scientifc issues or matters of basic political knowledge. However, they are not complete moral agnostics. Tey probably have a wide range of genuine beliefs about day-­to-­day ethical concerns, such as about the morality of stealing, hurting others, or keeping promises. Most citizens have moral principles and intuitions which they employ and rely upon in daily life (Haidt, 2012). So, perhaps the public reason liberal could respond to democratic realism by exploring how to justify various political policies on the basis of citizens’ extremely basic moral commitments. Perhaps public reason liberals could succeed in such an endeavor. I cannot prove otherwise, though I can ofer reasons to think it will be difcult. Most citizens have a wide range of moral beliefs and endorse various mid-­level moral principles. Political theorists and philosophers ofen attempt to ground their political theories on just such beliefs. However, these basic moral principles, on their own, probably radically underdetermine politics. Te major moral theories—such as Kantianism, rule

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

148  Jason Brennan consequentialism, Rossianism and so on—appear to be compatible with a wide range of political views and ideologies, but do not in themselves provide strong justifcation for or reasons to endorse any of these views or ideologies. Getting to any political policy from a moral principle will nearly always require, at bare minimum, ancillary empirical claims about how institutions function and what diferent policies are likely to achieve. Further, it may require intermediary normative political commitments which are not themselves directly derived from more basic moral principles. For instance, we see Aristotelians both defending and attacking liberalism, democracy, or capitalism. Tere are Kantians arguing Kantianism requires social democracy and those arguing it forbids it. Tere are utilitarian anarchists and authoritarians. And so on. Now, it may be that some of these philosophers are right and others wrong. Perhaps utilitarianism really does commit one to a neoclassical economist view of the politics, and so all the utilitarian socialists are just doing it wrong. But more plausibly, they reason why people who agree on a moral theory nevertheless go every which way in politics is that moral theories radically underdetermine political theories. To illustrate, consider all of the papers and books you know which attempt to derive some political philosophy from some underlying moral theory. For any moral theory, such as Kantianism, you can fnd hundreds of papers and books arguing it implies socialism or capitalism or neither, lef-­ liberalism or libertarianism or communitarianism, egalitarianism or meritocracy, anthropocentrism or biocentrism, and a whole host of other incompatible views. You can fnd people using it to defend or critique abortion rights or the permissibility of live organ sales. Similarly, for any theory, such as egalitarianism, or any particular view in applied ethics, such as that it is permissible to hire sweatshop labor, you can fnd hundreds of philosophers defending it on diferent and incompatible background moral theories. Now, not all of these papers or arguments are equally good. Nevertheless, there are strong arguments for each of these positions. Te reason so many philosophers can go in incompatible directions with the same starting points, or arrive at the same ending point from incompatible beginnings, could simply be that most everyone is guilty of serious error. But, more plausibly, it seems that high-­level, abstract moral theories, along with the basic, day-­to-­day moral principles most of us live our lives by, greatly underdetermine any particular political philosophy. Te major moral theories are probably compatible with a wide range of political views and ideologies, but do not in themselves provide strong justifcation for or reasons to endorse any of these views or ideologies. A Kantian or an act utilitarian can go any number of ways politically. Getting to any political policy from a moral principle will nearly always require, at bare minimum, ancillary empirical claims about how institutions function and what diferent policies are likely to achieve. Further, it may require intermediary normative political commitments which are not themselves straight derivations from more basic moral principles. If even highly detailed, robust, and sophisticated moral theories radically underdetermine politics, then the more minimal and basic moral beliefs most citizens possess must underdetermine politics even more. While most people have some strong moral beliefs, there is far too much logical space between their (a) basic moral attitudes and (b) any particular set of policies,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Does Public Reason Liberalism Rest on a Mistake?  149 coercive institutions, or coercive social norms to justify (b) on the basis of (a). Getting from (a) to (b) requires adding a wide range of intermediary moral and political principles, plus a wide range of empirical claims about basic political facts and higher-­level social scientifc principles. Tese intermediary principles and facts may be compatible with X- and Y-­citizens’ basic moral outlooks, but citizens do not positively endorse these principles or facts and have little access to them. Teir basic normative outlooks do not imply or even strongly suggest these intermediary principles. Remember from Section 3.2 that to justify some coercive policy in public reason liberalism, it is not enough that the reasonable citizens lack an objection to that policy, or that their beliefs are compatible with it. Rather, coercion is presumed wrong unless there is a strong enough positive case for coercion, based on the reasons the citizens subject to coercion possess. Accordingly, it still remains unclear how a public reason liberal can claim coercive policies are justifed to X- and Y-­citizens, even though such citizens have some moral beliefs.

3.4  Idealize Citizens

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

One possible way to “solve” the problem would be to idealize X- and Y-­citizens to a certain degree. We might ask what such citizens would endorse if only they were more strongly opinionated about political matters. However, idealization already runs into familiar difculties, as David Enoch (2013, 2015), Chad van Schoelandt (2015), and others (Raz, 1990) have argued. As Quong (2017) summarizes their objections: if the degree of idealization is very substantial, this creates two diferent worries. First, it’s no longer clear in what sense the resulting rules are justifable to the real persons bound by the rules. Te whole apparatus of public justifcation might seem superfuous; it would be simpler and more accurate to simply present certain principles or reasons as true, and declare that anyone who refuses to acknowledge their truth is making an error. Second, and relatedly, too much idealization may implausibly entail that almost all real persons are excluded from the constituency of public reason.  (Quong, 2017)

Idealization seems to tell us which people would have reason to accept certain principles or policies if only those people were, well, diferent from how they are. Further, the more one idealizes, the more one seems to be saying that people would have reason to accept those principles or policies because they are, well, true. Te other familiar problem with idealization is that it means we are not publicly justifying principles and policies to the actual public subject to those policies, but to a hypothetical public with diferent beliefs than the actual public. To that degree, the public reason project fails on its own terms; it means that in the real world, the people in power impose rules and policies upon citizens without justifying those rules or policies by citizens’ own lights. But, in addition, the realist can add that idealization does not rescue public reason from realism because there are too many equally valid but incompatible ways to “idealize” actual X- and Y-­people. X- and Y-­people are not moral agnostics. Tey are

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

150  Jason Brennan normal people. Tey have some basic moral principles and intuitions, but they do not subscribe to robust moral theories, political theories, or ideologies. Teir normative and empirical beliefs underdetermine whatever moral theory, political theory, or ideology they “would” adopt if they were more opinionated, interested, or informed. When we “idealize” them, we insert in our models of such citizens a number of opinions, attitudes, and beliefs they do not have and which are not entailed by their underlying moral beliefs.

4.  Te Justifcation Problem for I- and M-­People A-, B-, C-, D-, X-, and Y-­people are defned doxastically, by the kinds of beliefs they have—or lack. A-, B-, and C-­people endorse public reason liberalism. D-­people reject it in favor of an illiberal traditionalist view. X- and Y-­ people lack the background political beliefs which would commit them either way, and they seem to lack the kind of beliefs necessary for public justifcation to occur. Y-­ people superfcially look like A- through D-­people, but appearances are deceiving. But even with these six categories, we have not yet covered everybody. Consider these three new categories: I-­People: Citizens who are ignorant. Tey may hold various political beliefs, but the lack the information necessary to justify those political beliefs. J-­People: Citizens who are well-­informed. Tey may hold various political beliefs, and they have the information necessary to justify those beliefs.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

M-­People: Citizens who are misinformed. Tey may hold various political beliefs, but they are mistaken about the facts and/or endorse false social scientifc theories.

A- through D- and X- and Y-­people are defned in terms of what they believe. I-, J-, and M-­people are defned in terms of how they believe it and what information they have to justify whatever beliefs they have. Tus, a person can be both a B-­person and an M-­person at the same time. Te empirical evidence overwhelmingly shows that most voters are ignorant and misinformed about a wide range of salient political issues. For instance, Brexit “leave” voters were systematically mistaken about nearly every relevant fact (Brennan,  2017, pp. vii–viii). Tey dramatically overestimated the number and percentage of EU immigrants in the UK and the amount the UK sends to Europe for various kinds of welfare programs. Tey dramatically overestimated foreign investment from China and elsewhere and dramatically underestimated foreign investment from the EU. We do not have direct survey evidence about their beliefs about the efects of trade and immigration, but insofar as they had such beliefs, their economic beliefs likely went against the overwhelming evidence economists have produced (Van der Vossen and Brennan, 2018). Public reason liberalism provides a good model for justifying policies when there is underlying epistemically and morally reasonable disagreement over some matter. But what if the underlying disagreements result from bad epistemic states? Consider this common case: Suppose citizens advocate protectionist policies because they are ignorant and misinformed. If they were at all informed about the economics of trade, they would—like nearly all economists—reject protectionism. In this example,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Does Public Reason Liberalism Rest on a Mistake?  151 protectionist policies are justifable in light of citizen’s beliefs about policy, while liberal trade policies are incompatible with their beliefs. Public reason liberals claim we should not ram coercive policies down people’s throats when they have reasonable opposition. However, when there is overwhelming evidence for a particular policy, it is puzzling why opposition based on ignorance and misinformation would disqualify a view. Tis is especially puzzling when the citizens in question are making mistakes about means rather than ends. Most protectionist citizens support the ends free trade will produce, but they oppose free trade because they are misinformed about how it works. In response to these worries, public reason liberals idealize citizens to some extent. Early public reason liberals, such as Rawls or Habermas, ofen imagined that the citizens to whom justifcation is owed possess something close to perfect information. However, this leads to many of the problems of idealization that we previously discussed. It is unclear why perfectly informed citizens would not converge on the same beliefs, and so it is unclear why there would be any reasonable disagreement public justifcation could resolve. It seems to contradict the spirit of the public reason project, because we are no longer concerned to avoid ramming policies down actual citizens’ throats, but instead are asking what fully informed hypothetical people would support. To avoid this problem, public reason theorists more recently seem to recommend more moderate forms of idealization: they imagine citizens meet some moderate standard of information, if not perfect information (Gaus, 2011, pp. 232–60; Vallier, 2014, pp. 145–80). But it remains unclear how this solves the problem. Empirical work generally fnds that most citizens are ignorant not merely about more complicated matters of social scientifc knowledge, but about the basic political facts. Misinformation and systematically mistaken beliefs are widespread; somewhere between a third and two-­thirds of citizens are political know-­nothings (Caplan, 2007; Caplan et al., 2013; Achen and Bartels, 2016; Brennan, 2016). Te dilemma remains: If we make the epistemic standards low enough to qualify most I- and M-­people as “reasonable,” then public justifcation becomes beholden to culpable ignorance and misinformation. We must implement bad policies and fail to implement good policies, despite overwhelming evidence, because citizens make culpable, basic epistemic mistakes. If, on the other hand, we keep the bar higher than that, then we end up excluding the majority of citizens from the reasonable. Most citizens are I- and M-­people, not J-­people. Public justifcation is possible and necessary for only a small percentage of the best-­informed citizens. Public reason liberals end up saying we may force policies upon most citizens who do not endorse those policies, based on reasons that do not hook onto those citizens’ beliefs. It thus fails on its own terms.

5.  Summary and Conclusion Te public reason project appears grounded in the populist paradigm, which sees the vast majority of citizens as rather opinionated about politics. Even on the populist paradigm, the public reason project faces plenty of trouble, as evidenced by the interminable intramural debates among public reason liberals. But the populist paradigm is highly controversial. A massive literature in political science claims that most citizens are not very ideological. Tey have few political beliefs

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

152  Jason Brennan and their beliefs are unstable. Tey are ignorant, misinformed, and biased. Even contemporary populists largely accept these points. Realists add that political behavior—including expressing commitment to certain political platforms and beliefs— is based on who one is, not what one believes or what one wants. It remains unclear how we can make sense of the public reason project if these fndings are correct.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Achen, C. and L.  Bartels. 2002. “Blind Retrospection: Electoral Responses to Draught, Flue, and Shark Attacks.” Prepared for presentation at the annual meeting of the American Political Science Association, Boston. Achen, C. and L.  Bartels. 2016. Democratic Teory for Realists. New York: Princeton University Press. Althaus, S. 2003. Collective Preferences in Democratic Politics. New York: Cambridge University Press. Appiah, K. A. 2018. “People Don’t Vote for What Tey Want. Tey Vote for Who Tey Are.” Washington Post, August 30, URL: https://www.washingtonpost.com/outlook/ people-dont-vote-for-want-they-want-they-vote-for-who-they-are/2018/08/30/ f5b7e44-abd7-11e8-8a0c-70b618c98d3c_story.html Arceneaux, K., and R. M. Stein. 2006. “Who Is held Responsible When Disaster Strikes? Te Attribution of Responsibility for a Natural Disaster in an Urban Election.” Journal of Urban Afairs 28: 43–53. Arian, A. and M.  Schamir. 1983. “Te Primarily Political Functions of the Lef–Right Continuum.” Comparative Politics 15: 139–58. Barnes, S. H. 1971. “Lef, Right, and the Italian Voter.” Comparative Political Studies 4: 157–75. Bartels, L. 1986. “Issue Voting under Uncertainty: An Empirical Test.” American Journal of Political Science 30: 709–28. Bartels, L. 1996. “Uninformed Votes: Information Efects in Presidential Elections,” American Political Science Review 40: 194–230. Bartels, L. 2003. “Democracy with Attitudes.” In G.  Rabinowitz and M.  B.  MacKeun (eds.), Electoral Democracy. New York: Oxford University Press. Benn, S. 1988. A Teory of Freedom. New York: Cambridge University Press. Brennan, G. and L.  Lomasky. 2003. Democracy and Decision. New York: Cambridge University Press. Brennan, J. 2016. Against Democracy. Princeton, NJ: Princeton University Press. Brennan, J. 2017. “Moral Pluralism.” in Powell and Babcock 2017, pp. 301–38. Campbell, A., P. E. Converse, W. E. Miller, and D. E. Stokes. 1960. Te American Voter. New York: John Wiley. Caplan, B. Te Myth of the Rational Voter. Princeton, NJ: Princeton University Press. Caplan, B., E. Crampton, W. A. Grove, and I. Somin. 2013. “Systematically Biased Beliefs about Political Infuence.” PS: Political Science & Politics 46: 760–7. Carter, J. and S.  Guerette. 1992. “An Experimental Study of Expressive Voting.” Public Choice 73: 251–60.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Does Public Reason Liberalism Rest on a Mistake?  153 Chong, D. 2013. “Degrees of Rationality in Politics.” In David O. Sears and Jack S. Levy (eds.), Te Oxford Handbook of Political Psychology, pp. 96–129. New York: Oxford University Press. Cohen, G. 2003. “Party over Policy: Te Dominating Impact of Group Infuence on Political Beliefs.” Journal of Personality and Social Psychology 85: 808–22. Converse, P. 1964. “Te Nature of Belief Systems in Mass Publics.” In D. E. Apter (ed.), Ideology and Discontent. London: Free Press of Glencoe. Converse, P. 1990. “Popular Representation and the Distribution of Information.” In John A. Ferejohn and James H. Kuklinski (eds.), Information and Democratic Processes. Urbana: University of Illinois Press. Converse, P. and R.  Pierce. 1986. Political Representation in France. Cambridge, MA: Harvard University Press. Dahl, R. 1956. A Preface to Democratic Teory. Chicago: University of Chicago Press. Dahl, R. 1998. On Democracy. New Haven: Yale University Press. Delli-Carpini, M. X and S. Keeter. 1996. What Americans Know about Politics and Why It Matters. New Haven, CT: Yale University Press. Enoch, D. 2013. “Te Disorder of Public Reason.” Ethics 123: 141–76. Enoch, D. 2015. “Against Public Reason.” In Oxford Studies in Political Philosophy 1, ed. D. Sobel, P. Vallentyne, and S. Wall, pp. 112–42.Oxford: Oxford University Press. Erison, C., M.  Lodge, and C.  S.  Taber. 2014. “Afective Contagion in Efortful Political Tinking.” Political Psychology 35: 187–206. Estlund, D. 2008. Democratic Authority. Princeton, NJ: Princeton University Press. Freeman, S. 2001. “Illiberal Libertarians: Why Libertarianism Is not a Liberal View.” Philosophy and Public Afairs 30: 105–51. Freeman, S. 2017. “Liberal and Illiberal Libertarianism.” In Jason Brennan, Bas van der Vossen, and David Schmidtz (eds.), Te Routledge Handbook of Libertarianism, pp. 108–26. New York: Routledge Press. Friedman, Jefrey. 2006. “Democratic Competence in Normative and Positive Teory: Neglected Implications of ‘Te Nature of Belief Systems in Mass Publics’.” Critical Review 18: i–xliii. Gamm, G. H. 1989. Te Making of New Deal Democrats: Voting Behavior and Realignment in Boston, 1920–1940. Chicago: University of Chicago Press. Gaus, G. 1996. Justifcatory Liberalism. New York: Oxford University Press. Gaus, G. 2004. Contemporary Teories of Liberalism. Tousand Oaks: Sage. Gaus, G. 2010. Te Order of Public Reason. New York: Oxford University Press. Gaus, G. 2011. Te Order of Public Reason. New York: Oxford University Press. Gilens, M. 2012. Afuence and Infuence. Princeton, NJ: Princeton University Press. Goren, P. 2005. “Party Identifcation and Core Political Values.” American Journal of Political Science 49: 882–97. Green, D. and I. Shapiro. 1994. Pathologies of Rational Choice Teory. New Haven, CT: Yale University Press. Greene, S. 1999. “Understanding Party Identifcation: A Social Identity Approach.” Political Psychology 20: 393–403. Haidt, J. 2012. Te Righteous Mind. New York: Pantheon.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

154  Jason Brennan Healy, A. and N. Malholtra. 2010. “Random Events, Economic Losses, and Retrospective Voting: Implications for Democratic Competence.” Quarterly Journal of Political Science 5: 193–208. Hibbing, J.  R. and E.  Teiss-Morse. 2002. Stealth Democracy. Cambridge: Cambridge University Press. Inglehart, R. and H. Klingemann. 1976. “Party Identifcation, Ideological Preference, and the Lef–Right Dimension among Western Mass Publics.” In I. Budge, I. Crewe, and D. Fairlie (eds.), Party Identifcation and Beyond. London: Wiley. Iyengar, S. and S.  J.  Westwood. 2015. “Fear and Loathing Across Party Lines: New Evidence on Group Polarization.” American Journal of Political Science 59: 690–707. Iyengar, S., G.  Sood, and Y.  Lelkes. 2012. “Afect, Not Ideology: A Social Identity Perspective on Polarization.” Public Opinion Quarterly 76: 405–31. Kahan, D., E. Peters, E. Cantrell Dawson, and P. Slovic. 2013. “Motivated Numeracy and Enlightened Self-Government.” Behavioral Public Policy 1: 54–86. Kelly, J. T. 2012. Framing Democracy. Princeton, NJ: Princeton University Press. Kinder, D. and N. Kalmoe. 2017. Neither Liberal nor Conservative: Ideological Innocence in the American Public. Chicago: University of Chicago Press. Larmore, C. 1990. “Political Liberalism.” Political Teory 3: 339–60. Lenz, G.  S. 2009. “Learning and Opinion Change, Not Priming: Reconsidering the Priming Hypothesis.” American Journal of Political Science 53: 821–37. Lenz, G.  S. 2012. Follow the Leader? How Voters Respond to Politician’s Policies and Performance. Chicago: University of Chicago Press. Lewis-Beck, M., W.  Jacoby, H.  Norpoth, and H.  Weisberg. 2008. Te American Voter Revisited. Ann Arbor: University of Michigan Press. Lippmann, W. 1922. Public Opinion. New York: Penguin. Lodge, M., and C.  Taber. 2013. Te Rationalizing Voter. New York: Cambridge University Press. Mason, L. 2017. Uncivil Agreement: How Politics Became Our Identity. Chicago: University of Chicago Press. Mason, L. 2018. “Ideologues without Issues: Te Polarizing Consequences of Ideological Identities.” Public Opinion Quarterly 82: 280–301. McCann, J. A. 1997. “Electoral Choices and Core Value Change: Te 1992 Presidential Campaign.” American Journal of Political Science 41: 564–83. Muddle, C. 2007. “Te Single-Issue Party Tesis: Extreme Right Parties and the Immigration Issue.” West European Politics 22: 182–97. Mutz, D. 2006. Hearing the Other Side. Cambridge: Cambridge University Press. Mutz, D. 2008. “Is Deliberative Democracy a Falsifable Teory?” Annual Review of Political Science 11: 521–38. Oppenheimer, D. and M.  Edwards. 2012. Democracy Despite Itself: Why a System that Shouldn’t Work at All Works so Well. Cambridge, MA: MIT Press. Quong, J. 2017. “Public Reason.” In Stanford Encyclopedia of Philosophy, ed. E. N. Zalta, URL: https://plato.stanford.edu/entries/public-reason/ Pettit, P. 2012. On the People’s Terms. New York: Oxford University Press. Pitkin, H. F. 1967. Te Concept of Representation. Berkeley: University of California Press. Rasinki, K. A. 1989. “Te Efect on Question Wording on Public Support for Government Spending.” Public Opinion Quarterly 53: 388–94.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Does Public Reason Liberalism Rest on a Mistake?  155 Rawls, J. 1996. Political Liberalism. New York: Columbia University Press. Raz, J. 1990. “Facing Diversity: Te Case of Epistemic Abstinence.” Philosophy and Public Afairs 19: 3–46. Simler, K. and R.  Hanson. 2018. Te Elephant in the Brain. New York: Oxford University Press. Somin, I. 2013. Democracy and Political Ignorance. Stanford: Stanford University Press. Stimson, J., M.  Mackuen, and R.  Erikson. 1995. “Dynamic Representation.” American Political Science Review 89: 543–65. Stokes, S. C. 1988. “Pathologies of Deliberation.” In J. Elster (ed.), Deliberative Democracy, pp. 123–39. New York: Cambridge University Press. Taber, C. and M.  R.  Lodge. 2006. “Motivated Skepticism in the Evlauation of Political Beliefs”. American Journal of Political Science 50: 755–69. Taber, C. and E. Young. 2013. “Political Information Processing.” In Huddy, Sears, and Levy, pp. 525–58. Tajfel, H. 1981. Human Groups and Social Categories. New York: Cambridge University Press. Tajfel, H. 1982. Social Identity and Intergroup Relations. Cambridge: Cambridge University Press. Tajfel, H. and J.  C.  Turner. 1979. “An Integrative Teory of Intergroup Confict.” In W.  G.  Austin and S.  Worchel (eds.), Te Social Psychology of Intergroup Relations. Monterey, CA: Brooks-Cole. Tompson, D. 1970. Te Democratic Citizen. Cambridge: Cambridge University Press. Tompson, D. 1987. Political Ethics and Public Ofce. Cambridge, MA: Harvard University Press. Tomasi, J. 2001. Liberalism Beyond Justice. Princeton: Princeton University Press. Tversky, A. and D. Kahneman. 1973. “Availability: A Heuristic for Judging Frequency and Probability.” Cognitive Psychology 5: 207–33. Vallier, K. 2014. Liberal Politics and Public Faith: Beyond Separation. New York: Routledge Press. Vallier, K. 2016. Liberal Politics and Public Faith: Beyond Separation. New York: Routledge. Vallier, K. 2018. “Public Justifcation.” Stanford Encyclopedia of Philosophy, ed. E. N Zalta, URL: https://plato.stanford.edu/entries/justifcation-public/#PubJusPri Van der Vossen, B. and J.  Brennan. 2018. In Defense of Openness. New York: Oxford University Press. van Schoelandt, C. 2015. “Justifcation, Coercion, and the Place of Public Reason.” Philosophical Studies 172: 1031–50. Weithman, P. 2011. Why Political Liberalism? New York: Oxford University Press. Westen, D., P. S. Blagov, K. Harenski, C. Kilts, and S. Hamann. 2006. “Te Neural Basis of Motivated Reasoning: An fMRI Study of Emotional Constraints on Political Judgment during the U.S. Presidential Election of 2004.” Te Journal of Cognitive Neuroscience 18: 1947–58. Zachmeister, E. 2006. “What’s Lef and Who’s Right? A Q-method Study of Individual and Contextual Infuences on the Meaning of Ideological Labels.” Political Behavior 28: 151–73. Zaller, J. 1992. Te Nature and Origins of Mass Opinion. New York: Cambridge University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

9

Te Epistemic Pathologies of Elections and the Epistemic Promise of Lottocracy Alexander Guerrero

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1. Introduction Tere are many ways of evaluating legal and political institutions. We can ask about whether the institutions do well by norms of political legitimacy, political equality, individual autonomy (freedom, liberty), and non-­domination. We can focus on outcomes, asking about what the institutions bring about in terms of welfare promotion, distributive justice, retributive justice, promotion of autonomy, and promotion of egalitarian values. We can ask about the stability and popularity and responsiveness of the political institutions, and the extent to which the people living under institutions consent to them, authorize them, and accept them. In this chapter, I introduce a new way to evaluate legal and political institutions: in terms of their sensibility. I understand sensibility as the ability to appreciate and to respond to the world as it is. Tus, there are two distinct components of sensibility: (1) appreciating (or understanding or knowing) the world as it is, and (2) responding to the world in light of this appreciation. Te frst of these concerns the epistemic capacities of institutions. Te second concerns the agential capacities (rationality, morality, steadfastness) of institutions. Epistemic capacities concern the ability and propensity of the institutions to gather and generate relevant evidence (evidence relevant to the decisions that need to be made); to engage with and draw from diverse sources of knowledge, including extant technical, esoteric, and expert knowledge; to accurately and appropriately assess, weigh, and evaluate evidence; and to organize and disseminate evidence and knowledge so that it is readily available and appropriately salient for decision-­making purposes. Agential capacities concern the ability and propensity of the institutions to make decisions in light of the evidence they possess; to make decisions in light of—and that are supported by—the best available evidence more generally; to make decisions that are coherent, rational, and morally appropriate in light of the evidence and the interests and values at stake; to act quickly and decisively when necessary; to change direction or respond to changed circumstances when necessary; to consider both short-­term and long-­term consequences of a decision; and so on. Accordingly, sensibility can be imperiled through epistemic error or agential error or a combination of those two. In this chapter, I will focus on one dimension of the relative sensibility of two broad categories of political systems—electoral representative systems and what I call lottocratic systems. I focus on these two systems because Alexander Guerrero, The Epistemic Pathologies of Elections and the Epistemic Promise of Lottocracy. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Alexander Guerrero. DOI: 10.1093/oso/9780192893338.003.0010

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Epistemic Pathologies of Elections  157 I think that (unlike many other systems) both arguably pass many of the plausible tests of political morality.1 Tis chapter will focus on just the epistemic dimension of sensibility. Discussion of diferent forms of political institutions rarely focus on epistemic considerations— although arguably the main justifcation for using elected representatives rather than direct democratic decision-­making is an epistemic one. Tere has been some shif in this recently, with epistemic pressure on electoral representative democracy coming from technocratic, meritocratic, and epistocratic directions.2 I will raise some worries in this vein, but most of my concerns are distinct from the familiar ones about the broadly ignorant general public and the need to move political decisions to the experts, the wisest among us, or those chosen through some sort of meritocratic selection. Whatever the epistemic merits of these responses, I think they run afoul of requirements of political morality. And I worry, too, about these responses on epistemic grounds—something I will say more about later. I do not diagnose the central epistemic issue with electoral representative democracy to be one of the fundamental incompetence of ordinary people, nor do I suggest that we move away from democracy understood as egalitarian rule by the people. I do suggest, however, that we need to reconceptualize democracy so that it is not simply equated with electoral representative democracy, and that we should notice the ways in which elections introduce distortions, biases, and other epistemic problems—bringing out the epistemically worst in us, rather than putting us in a position to be our epistemic best. It is worth stressing at the outset that epistemic capacities are related to other, more familiar political values: responsiveness, welfare promotion, justice, and so on. We should expect them to travel together. For example, if welfare promotion is a signifcant role for political institutions, then relevant evidence will include evidence about how some policy would afect the well-­being of all citizens, and political institutions may do better or worse at gathering this evidence, weighing it, and ensuring that it is available to the relevant institutional actors when needed. Similarly, a policy to regulate the environmental harms of some industry—through coercion, if necessary—is only going to be justifed or a good policy if those regulations actually succeed in addressing the environmental harms. And this requires having good evidence and true beliefs regarding that harm and what might address it. Indeed, it is plausible that we should care about epistemic capacities of institutions only derivatively. But I will not wade deeper into those waters here. Ultimately, we will need some account of what the proper purpose or function of political and legal institutions is supposed to be, before we can determine what evidence is relevant to the decisions being made, so that we can assess whether those institutions are doing well or poorly in epistemic terms. For now, let me remain as 1  Sensibility as a political value is a scalar value. Other things being equal, it is better for a political system to be more rather than less sensible. But other things might not be equal, in which case the sensibility of a political system must be weighed against other political values. Perhaps the most sensible political system in some context would be an advanced technocratic system (although I doubt this, for epistemic reasons, which will become apparent later), but that system might fail other tests of political morality. 2  See, for example, Bell (2015), Brennan (2016), and Somin (2013).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

158  Alexander Guerrero ecumenical as possible, maintaining only that some outcome-­ related purposes regarding justice, welfare promotion, and autonomy promotion are among the proper purposes of political and legal institutions, and that accomplishment of these purposes requires some substantial epistemic and agential quality at the level of legal and political institutions. In this chapter, I will consider a comparison of two diferent institutional arrangements—(1) electoral representative government (roughly, as practiced in the U.S.), and (2) lottocratic government as introduced in this chapter—in terms of their epistemic quality or expected epistemic quality. I will begin by drawing attention to several concerns about the sensibility of electoral representative institutions, focusing particularly on epistemic pathologies of those institutions. Te second part of the chapter discusses an alternative kind of political institution, which I call a lottocratic political institution, and argues that we might well expect these institutions to be more sensible alternatives, at least under some conditions, on epistemic grounds. It is difcult to compare a known and existing institutional form with a largely unknown and non-­existent institutional form. Te negative part of the chapter, then, can be seen as a series of concerns about or challenges to the epistemic merits of electoral representative institutions. Te positive part of the chapter can be seen as a suggestion for future thinking, empirical study, and experimentation.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Te Epistemic Pathologies of Electoral Representative Government Institutions and practices—whether social, legal, political, etc.—can have a number of diferent kinds of epistemically signifcant efects. Tey might improve or fail to improve the general knowledge of a population of people, afecting how many true and false beliefs individuals have about various questions. Tey might do this directly, through education or miseducation. Or they might undermine or reinforce various biases that prevent people from acquiring true beliefs, weighing evidence appropriately, or relying on good epistemic sources. We can see these capacities veritistically, concerning the production and promulgation of important and relevant true beliefs, and the avoidance of error (false belief) and ignorance (the absence of true belief). We can ask, then, in a Goldmanian vein: which political institutions and practices “have a comparatively favorable impact on knowledge [understood in the weak sense of true belief] as contrasted with error and ignorance”?3 But institutions and practices can also afect whose beliefs—whether true or false—matter, and how much they matter, and this can lead institutions to have something like collective or “efective” epistemic capacities, even if they do not afect whether particular individuals in the community have true or false beliefs. For example, institutions can help minimize the signifcance of false belief by preventing those beliefs from infuencing the decisions that are made and the actions that are taken, or by signifcantly reducing the weight of such beliefs. Or they might exacerbate the 3  Goldman (1999, p. 5).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Epistemic Pathologies of Elections  159 signifcance of false belief by giving the most power to those with systematic false belief or entrenched ignorance. In this section, I will consider a number of diferent pathologies of electoral representative government, with these falling into both of these categories of institutional epistemic efects. In discussing the epistemic pathologies of electoral representative government, I will focus on the case of political ofcials who are elected through relatively inclusive, majoritarian elections, conducted freely and fairly, where the ofcials chosen are elected to be a representative for a geographically defned district. For example, one might think of the U.S. House of Representatives or the U.S. Senate, or most state legislatures within the governments of the states of the U.S.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.1  Elections and Ignorance: Garbage In, Garbage Out One theme of political science research over the past 50 years is the remarkable extent of the ignorance of citizens in modern democracies, particularly in the U.S., across almost every politically relevant domain.4 Tis ignorance is both well-­ documented and, from a certain vantage point, unsurprising. As many have noted, it is not rational for individual voters to expend time and energy in becoming well informed about politics, given how unlikely it is that any one of their votes will be decisive. Furthermore, modern policymaking is incredibly technical and complex. Tis is important because although we might be generally ignorant, there may be some issues about which people are not ignorant, at least in a broad sense. So, for example, if there is a terrorist attack in a country, people in that country may not be ignorant of that fact. Or if there is a widespread famine in a country, people may not be ignorant of that fact. Te difculty comes in knowing more than these bare facts: what ought to be done? Is this a good idea? How should we respond? Will this be good for me, for our country, for the world? Is this the right thing to do? Te fact of modern policymaking being technical and complex should also afect our views about the potential of mass education as a possible response. Here is an initial dilemma, with a challenge to the epistemic quality of electoral representative institutions on either horn: either (a) the elected representative institutions are tightly responsive to the (very ignorant) views of the citizens or (b) they are not. If (a), then mass ignorance is guiding our political institutions in a way that is straightforwardly troubling, epistemically speaking. Tis is a familiar story; the original fear regarding electoral democracy and expansion of the franchise. Even if citizens might have the mental capability (and some who ofer this critique might even be skeptical of this) to gather evidence efectively, to think intelligently about policy questions, and to monitor their representatives, they do not have the time or the inclination to do this. Tey vote, instead, based on misinformation, simplifed versions of the policy problems, and on the basis of epistemically irrelevant 4  See Bartels (1996), Somin (2013).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

160  Alexander Guerrero considerations. If this is our situation—and the above evidence suggests that it might be—then the epistemic peril is obvious. Te familiar phrase from the world of computer programming—garbage in, garbage out—would be an apt description of our situation. It has been suggested that even if people are ignorant of much that seems relevant in terms of policy detail and basic facts of politics, they can still make epistemically responsible decisions by using proxies, signals, and heuristics of various kinds to overcome their ignorance.5 Tese strategies amount to a kind of deference to the monitoring and evaluation done by some other individual or group. For example, membership in a political party, endorsements from activist organizations or media institutions, and contributions and public endorsements from particular individuals might all seem to help individuals overcome personal ignorance to enable them to make decisions that are well supported by the evidence, even though they do not personally possess all the relevant evidence. But there are problems with strategies of this sort. First, the proxies may either be too coarse-­grained to help for particular issues or too fne-­grained to save individuals any efort. Second, it can be difcult and time-­consuming to determine which proxies are credible, particularly if one wants to fnd reliable but specifc proxies for many diferent issues. Tis can take almost as much efort, and be as challenging, as doing the research oneself. Finally, for some issues, there may not be good proxies or signals. Tere may be issues that are low profle or do not attract well-­funded individuals or groups to do the necessary investigative work, and there may be issues for which powerful interests have a lot at stake and do everything they can to shape the available information and to obscure the nature of their interests and eforts. Ultimately, it is an empirical question of whether elected representatives do hew closely to what their constituents believe and prefer. Te evidence is mixed but suggests that they pay attention to some of their constituents more than others.6 Te route to epistemic trouble is short and straightforward if the garbage in, garbage out story is correct. Perhaps it is not. For the sake of argument, and to explore the possible “best case” for electoral representative government under conditions of widespread voter ignorance, let us consider ways in which electoral democracy might manage to do tolerably well, epistemically, despite widespread voter ignorance.

2.2  Elections and Ignorance: Te End of Accountability and the Epistemic Disaster of Capture Tat brings us to the second horn of the dilemma: the possibility that elected representatives do not hew closely to what ordinary citizens believe or prefer in deciding what to believe or do. For those who favor electoral representative democracy over direct democracy, one of the central motivations for doing so is the expected improvement in epistemic quality—something one only achieves if elected 5 See, e.g., Ferejohn and Kuklinski (1990). Other work is more pessimistic about what can be accomplished by way of heuristics and signals. See Kuklinski and Quirk (2000). 6  See Gilens (2012).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Epistemic Pathologies of Elections  161 representatives do not just defer to the beliefs and preferences of the ignorant masses. Some worry about this from a perspective of democratic control or concerns about elite domination. I want to stress that epistemic peril lies this way, too. Te argument in this section goes against the standard justifcations for systems of electoral representative government. Te use of elected representatives is typically defended on epistemic and agential grounds. Elected representatives embody a kind of compromise: allowing for the “refning and enlarging” of constituent views and preferences, while having political institutions that are not completely untethered from what is in the interests of the citizens who are represented. Trough the mechanism of electoral accountability, systems of elected representatives require political ofcials to pay attention to the interests and beliefs and preferences of those people on whose behalf they are supposed to be governing. Te problem is that for electoral representative systems of government these epistemic and agential capacities are only going to be present if there is what I call meaningful accountability. Responsiveness is tied to accountability—we expect electoral democratic systems of government to do relatively well by responsiveness because those systems have the particular mechanisms of accountability that they do. But responsiveness is tied only to meaningful accountability. Meaningful accountability is distinct from accountability simpliciter in that the former, but not the latter, is connected to informed monitoring and evaluation practices. Furthermore, without meaningful accountability, we should expect to see high levels of political capture, which will in turn imperil both epistemic and agential capacities. Let me fll in this story a bit more.7 Accountability through elections requires—at least—free, regular, competitive, and fair elections. Candidate A runs on a platform of doing X, Y, and Z, in opposition to some Candidate B, who runs on a platform that is at least somewhat diferent from A’s. If A’s platform is more popular, she will likely win the election. Afer being elected, she will have many decisions to make while in ofce. Tese decisions will be monitored and evaluated by her constituents, perhaps aided by news media of various kinds, and the candidate will be held accountable for decisions made while in ofce when she next comes up for re-­election. If elections are not free, regular, competitive, and fair, these mechanisms of accountability will fail. Even in well-­established electoral democracies, there are familiar concerns about electoral systems on the grounds that they are not adequately free, competitive, or fair. But even if these concerns were addressed, serious problems would still arise. Meaningful accountability requires not just the ability to “vote them out,” but also the ability to do this based on good information, on actual evidence that bears on the quality of representation. Tis requires informed monitoring and evaluation. Tis monitoring of representatives can be thwarted by ignorance about what one’s representative is doing or ignorance about a particular political issue. And even if one knows what one’s representative is doing with respect to some issue, one may have no idea whether what one’s representative is doing is a good thing in general or whether what she is doing will be good for oneself. Each of these kinds of ignorance 7  Te argument in this section draws on Guerrero (2014).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

162  Alexander Guerrero can undermine the ability of ordinary citizens to engage in meaningful monitoring and evaluation of the decisions of their representatives. If I don’t know what you’ve done, I can’t hold you accountable for it. If I don’t know anything about the issues or how to evaluate what you have done, I can’t hold you accountable for voting yes, rather than no, or vice versa. Here is the basic concern: elected political positions for which the elected ofcials are not meaningfully accountable to their constituents will be used to advance the interests of the socioeconomically powerful. Let us refer to this phenomenon as capture: an elected ofcial is captured if he or she uses his or her position to advance the interests of the powerful, rather than to create policy that is responsive or good (when doing so would confict with the interests of the powerful). Te suggestion is that in the absence of meaningful accountability, we should expect to see high levels of political capture. Political capture is bad from an epistemic and agential vantage point. Te agential worries are clear enough and a familiar source of concern and disapproval: whatever elected ofcials believe about issues and policy options, they will be inclined to act so as to beneft the powerful interests who can keep them in power. On this view of capture, it is entirely possible that the politically powerful know exactly what they are doing, who it will harm, who it will beneft, and they are going ahead and acting anyway. Doubtlessly this does describe some captured elected ofcials. But a diferent worry—one that is perhaps more pernicious and difcult to detect and address—is that captured elected representatives really do come to believe that the best policies are X, Y, and Z—where X, Y, and Z are also the ones preferred by the elite. One route to this result is through motivated reasoning of a kind that everyone is subject to—we are very good at coming to rationalize and justify the actions we take. But another route to this result, one that is not incompatible with the frst, is through systematic epistemic distortion that results from capture. Focusing just on the epistemic side of things, captured representatives and institutions will typically have a perverse set of priorities which lead them to fail to obtain or to generate relevant evidence; to seek out testimony only from certain groups of people; to engage in distorted and selective reliance on and attention to experts; to receive and disseminate misinformation if doing so is to the advantage of the capturing entities (as it ofen is); to discount or ignore relevant bodies of evidence and knowledge; to ignore evidence and knowledge when acting if doing so better suits the interests of the capturing entities; and to act with an unduly limited focus on the issues and problems that are most signifcant to the capturing entities, rather than to the broader political community. Tere will be powerful incentives to ignore or just not seek out relevant evidence and sources of possibly relevant evidence, to generate and disseminate misinformation that serves the interests of the capturing entities, to consult and invoke expertise only asymmetrically (when doing so serves the interests of the capturing entities) and to otherwise ignore or undermine expert knowledge, and to act to advance the interests of the powerful even in those cases in which relevant evidence inclines toward other decisions. And we should expect that technocratic and purportedly epistemically useful institutions within the broader system—legislative hearings with expert testimony, legislatively created administrative agencies or task forces—will also be efectively captured and turned into engines of

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Epistemic Pathologies of Elections  163 ignorance as a result. Rather than improving the epistemic functioning of these institutions, they will mostly serve the ends of justifying the policies and decisions favored by the capturing entities, providing a veneer of epistemic respectability or inevitability to the decisions being made. Te basic argument is simple. Voter ignorance undermines meaningful electoral accountability. An absence of meaningful electoral accountability results in capture. And capture results in what might well be described as epistemic disaster. Tis argument might be contested. Te hope, however, is that the argument articulates a familiar set of concerns about electoral representative systems, albeit in more explicitly epistemic guise. Tese concerns are brought to the fore when one thinks about how little one knows about most of what one’s elected ofcials do, what they spend their time investigating, who they spend their time listening to, who drafs the legislation they end up supporting, who has their ear. Or when one thinks about how complex some issues are, how much of what one believes about those issues is a result of information provided by a few powerful media institutions, how much money powerful interests have at stake, and how hard it is to create rules to adequately monitor the infuence of these powerful interests and the way in which their actions and the practices of elected representatives might be distorting the epistemic environment. Tis is one central concern about the sensibility of electoral representative institutions, although it is not usually framed as an epistemic concern. We are ignorant, so there is no meaningful accountability, so we should expect high levels of capture, which imperils the sensibility of electoral representative institutions. Tey are perhaps sensible and rational if we focus only on the interests of the most powerful members of the political community (although I doubt even that). Tey are insensible and irrational if we include the interests of the rest of the political community and the rest of the world. But there are also many other threats to the epistemic capacities of electoral representative systems of government. Many of these require mass ignorance as a background condition, but they connect in diferent and specifc ways to other features of elections as well. Let me turn to these.

2.3  Electoral Incentives and Short-­term Focus Elections lead elected ofcials to focus on the near term—on those problems that they can get immediate credit for addressing, or which they will be blamed for not addressing before the next election. Given that the urgency of political problems is not simply a function of their near-­term proximity, this fact about elections has signifcant consequences for the sensibility of electoral representative government. It afects which problems get attention and are the focus of political action. But it also leads to signifcant epistemic distortion in what voters and elected representatives come to believe about the problems that we face. Begin by noticing an important background fact about the modern world: many of the most urgent problems we face in the contemporary world are ones that we face together, at least in the sense that there is a general acknowledgment that individual

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

164  Alexander Guerrero action will be insufcient or is unlikely to occur without some efort at group or collective coordination through political and legal institutions, whether at the national or international level. Modern political problems are also complex and information intensive in a way that makes it difcult for members of the political community to have informed beliefs and preferences about those problems, given their limited time and knowledge. Political problems with long time horizons— problems the adverse consequences of which will not be apparent in the near future—will ofen have the following additional features:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1. it is opaque to most people whether actions taken now have solved or prevented the problem; 2. it is possible for people to deny the existence of the problem without fouting evidence that is readily available to all (including non-­experts); 3. it is possible for people to suggest that some as-­yet unidentifed solution to the problem will emerge over time without anything diferent being done now; 4. present costs will be salient, but future benefts will not be. If there is a problem with these features, then mechanisms of electoral accountability will fare poorly at producing good decisions with respect to addressing that problem. Electoral representative government will show very low levels of sensibility in their response to such problems. Here is why: elections lead elected ofcials to focus on those problems for which they can get or claim credit for addressing (or for which they will be punished if they fail to address them), and to ignore or put on the back burner those problems with a longer horizon or those solutions for which it is harder to get credit. Arguably, the most urgent issue we face is climate change, and it is a problem that demands political solutions in order to address what appears to be a complex collective action problem. But many of the worst efects of climate change will not be realized for decades, and so elected politicians are unlikely to pay the short-­term political cost (due to unpopular taxes on fossil fuels, limits on vehicles, etc.), given that they will not see the longer-­term political benefts. So, even if there are clear steps that need to be taken, many elected ofcials will not take them. Tis might be an instance of faring poorly by sensibility due to agential failings— the problem is known to the elected ofcials, they believe it is a problem, and yet they do nothing because of the short-­term costs to their electoral chances of doing something. But, along with this possibility, there is the real possibility that motivated reasoning, eforts at misinformation and disinformation, and even just the investigative priorities of elected representatives lead to signifcant epistemic failings. On the disinformation front, for example, it is important to note that for problems with long time horizons, it is comparatively easy to deny the existence of the problem, or to question its reality or etiology in other ways, even when the best evidence suggests otherwise. Te best evidence may ofen be technical and complex, and—as bearing on a somewhat distant projection—far from certain in its implications. Furthermore, even if the whole population believes that the problem is a serious one, it will be possible to try to convince people that one is doing something to address the problem, or that other non-­political action will be sufcient to address the

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Epistemic Pathologies of Elections  165 problem. If there are costs to actually addressing the problem, costs that will be salient to all, then elected ofcials will have electoral incentives to compete by avoiding incurring these costs, even if this will make everyone worse of in the long run. One very efective way to compete on this front is through disinformation and epistemic pollution: spreading false information, undermining reliance on actual experts, propping up pseudo-­experts and junk science, manufacturing controversy where none should exist, and so on. Elections generate powerful incentives for political representatives to focus on the short term, at great cost to the overall sensibility of our political institutions.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.4  Electoral Dynamics and Emotional Distortion Elections introduce distinctive ways of thinking and talking about political issues. As  noted above, there are concerns about misinformation introduced by the most powerful in order to convince us of what is most useful for us to believe, given their interests. Tere are also distinctive problems of elections that are made possible by widespread voter ignorance. One such problem is that certain issues—crime, war, terrorism—are such that they easily and powerfully engage our emotions through our concerns about violence and safety. For these issues, it is relatively easy to manipulate the electorate by manipulating our emotions, our attention, and our corresponding beliefs—particularly using television advertising and television programming, charged memes, and various forms of social media. Furthermore, when a person does not have a lot of information about a political problem, it is easier to manipulate that person into believing something through a combination of misinformation and emotional manipulation. Tese policy problems—such as criminal justice policy and national defense policy—generate strong emotional reactions, claim to have a certain kind of urgent or emergency status, and have truly vivid and terrible worst-­case outcomes. For these problems, emotional manipulation is both particularly easy and particularly efective. Tis manipulation can make us believe that something is a bigger concern than it really is, can make us focus on a few issues while ignoring many others, can shape and direct our investigative and research priorities, can lead us to rely and seek out the testimony of some while ignoring the testimony of others—all leading to epistemic distortion and error. Of course, elections are not responsible for our vulnerability to emotional manipulation and the error and epistemic distortion that might result from that manipulation. But elections do provide a powerful incentive for candidates to exploit that vulnerability. Because candidates for elected ofce know that we are broadly ignorant of policy details and that we fnd detailed policy discussion boring, they aim to connect with readily graspable, emotionally resonant issues and claims. We have limited attention spans, and they want to capture our attention to maximum electoral advantage. Emotional responses, rather than dispassionate consideration of evidence, appear to drive much of our thinking about politics.8 Tus, issue discussion 8  For a particularly provocative discussion, see Westen (2007).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

166  Alexander Guerrero is skewed and distorted in ways that are not justifable epistemically, and which lead to skewed and distorted policy priorities once elected ofcials are in ofce. Tis results in bad, insensible policy choices, but it also results in a warped information environment, in which highly charged cases receive an undue proportion of political attention and discussion, leading to widespread false beliefs about the severity of various problems, the actual harm done from various causes, and so on. It also encourages the creation and success of political information and news as entertainment and titillation. If there is a relatively free market in sources of information, and if some of them rely on emotional engagement, entertainment, and focus on stories that may be non-­representative, but which are highly charged, then those sources will crowd out other, more epistemically responsible sources of information. Tis is true generally, a point made by Neil Postman (1985), among others, but it is plausibly exacerbated when a signifcant portion of the news is driven by electoral politics and the purported need for information relevant to electoral choice.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.5  Electoral Dynamics and In-­group/Out-­group Tinking As noted above, as voters, we are ignorant of the detailed policy issues and even of the detailed stances of various candidates on those issues. It is time-­consuming, difcult, and tedious to pay attention to the details. On the other hand, we enjoy and easily comprehend the character drama, the horse race, the scandals, and the inspirational candidate. Tose are entertaining, and we all can have a relatively informed opinion about at least some aspects of a person, just from paying a bit of attention. And we are highly susceptible to in-­group and out-­group thinking. So, we identify our candidate or our party, and they become our team. We root for them as we might a sports team and we adopt whatever positions are adopted by our team, rather than considering the evidence or making a decision about which position to support in light of the evidence. We antagonize the other side, reduce them to a caricature of their worst elements, and come to view them as the enemy—meaning it is much less likely that we will take them seriously as epistemic sources. We flter evidence, experts, and media consumption through this lens, letting in (mostly) only what we agree with. Group attachments and social identities drive our thinking about politics, rather than the other way around.9 Tis basic story is the one ofered—and supported in empirical detail—by Christopher Achen and Larry Bartels (2016). If this account is correct, then the electoral process itself is contributing to distorting political discussion and the way in which individuals process what evidence they do encounter. Couple this with the fact that elected ofcials then will respond to these features of the electoral process, and there is a real concern about the sensibility of the attendant political discussion, processing of evidence, response to expertise, and political action based on this epistemic process. 9  For further discussion of these issues, see Michael Hannon and Jason Brennan’s contributions to this volume in Chapters 16 and 8, respectively.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Epistemic Pathologies of Elections  167 It is worth stressing the epistemic consequences of this kind of in-­group/out-­group sorting that is the direct result of elections. Political candidates competing for votes want to rally us to vote for them. Tey can do this by ofering popular votes and by getting us excited enough to actually vote. Importantly, they need to make it clear why we should bother to vote, and why we should vote for them. To do this, it helps to promote the view not just that one has better policy ideas, but also that the other side is fundamentally misguided—morally and epistemically. One hypothesis is that, in response to this electoral incentive, candidates—and thus, indirectly, elections— encourage us to focus on “deep disagreements,” disagreements that are clustered, not rationally resolvable, borne of underlying disagreement about normative principles and the epistemic quality of various sources, evidence, and methods.10 Focus on these kinds of disagreements might plausibly lead to what Michael Lynch calls “tribal” or group-­based intellectual arrogance: an implicit or even explicit commitment to the epistemic “unimprovability” of one’s own worldview by the evidence or experience of those in the opposing group.11 We see those on the other side as not just in disagreement with us, but also as fundamentally out-­of-­touch with reality, relying on bogus experts and unreliable testifers, and equipped with a deeply-­fawed moral compass that leads only to immorality and systematic moral error. If the other side is really like that, then we really must make sure that our side wins. But if we come to believe the story—that the other side is fundamentally misguided, both epistemically and morally—then half of the political community becomes almost epistemically useless to us, discounted as testifers, potential sources of knowledge and evidence, along with anything they might be associated with. And this is not just a theoretical or anecdotal worry—there is empirical evidence of this kind of epistemic efect.12

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.6  Elections, Epistemic Diversity, and Standpoint Teory Aristotle said that elections were oligarchic, not democratic. Here is something that he did not explicitly say, but which is true: to succeed in electoral contests, a person will generally need to be some combination of famous, of high social status, and/or wealthy. Elected representative institutions will not be a microcosm of the general population. Tey will be much closer to a microcosm of the elite. We should not be surprised, then, to see that 53 of the 535 members of Congress have a net worth of over $7 million (as of 2015); 130 of the 535 members of Congress have a net worth of over $2 million; 80 per cent are male; 84 per cent are white, and more than half are lawyers or businesspeople.13 Tis has signifcant epistemic implications. Te epistemic implications are due both to (a) the investigative interests and priorities of the elite and (b) the limitations of the epistemic position of the elite. Members of the elite will have little personal interest in or experience with many of 10  See Jeroen de Ridder’s discussion of these issues in his Chapter 12 “Deep Disagreement and Political Polarization” in this volume. 11  See his “Political Disagreement, Arrogance, and the Pursuit of Truth,” Chapter 13 in this volume. 12  See, e.g., Marks and Sharot (2019). 13  See Petersen (2012).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

168  Alexander Guerrero the urgent problems faced by the non-­elite. Tey will not know about these issues, they will be less inclined to learn more about these issues, and they may not know where to begin to learn about them (whose opinions should be sought, to whom they should listen, and so on). Tey may also be overconfdent in thinking that they do understand these political issues, even when they do not. Drawing on the work of Gloria Anzaldúa (1987), Nancy Hartsock (1987), Patricia Hill Collins (1990), Sandra Harding (1993), Charles Mills (2007), and others, we might invoke forms of standpoint theory to note that what one is able to notice and explain may be partly a function of one’s social position. For example, those who are oppressed ofen have to learn in much greater detail about their oppressors than the reverse—at least in terms. It is worth stressing that the knowledge possessed need not be exhaustive social scientifc knowledge regarding the full complex structural causes of oppression and social inequality. Te knowledge may be much more microscale, situational, social, and emotional, concerning how people do or are likely to behave under various circumstances, how certain actions make people feel, what subtle obstacles might exist to limit a policy’s efectiveness, and so on. Furthermore, the better epistemic position of the relatively oppressed and marginalized might be merely comparative. Mills, for example, discusses ways in which whiteness in America can be associated with systematic misperception, due to cultivated “white ignorance” about all manner of things, including historical facts about race and discrimination and injustice and oppression. Harding, Hartsock, and others make similar points with respect to understanding gender-­based oppression. If there is merit in the standpoint theory argument, and if elections lead to routine selection of the elite and against the selection of the relatively disempowered, then we should expect signifcant epistemic distortion in the views of those elected. If we combine those epistemic errors with a general lack of concern or regard for the interests of the disempowered—either because they just do not identify with those interests or share them, or because the disempowered ofer less electoral payof— then we should expect to see insufcient investigation and bias in whose testimony is sought and heeded. Even leaving aside standpoint theory, it is clear that there is something epistemically limited about a legislature composed of people who, disproportionately when compared to the population as a whole, have no frsthand knowledge of being an engineer, police ofcer, nurse, construction worker, waitress, cashier, plumber, social worker, scientist, single-­parent, non-­native English speaker, disabled person, openly gay person, community college student, welfare recipient, woman, African-­American, Hispanic-­American, Muslim-­American, frst-­generation immigrant, and so on. Tis is simply a point about the knowledge one is likely to have come by based on one’s life experiences, the evidence one is likely to have encountered. By using elections that disproportionately select for a white, male, wealthy, lawyer/businessperson elite, we are losing out on much of the available knowledge about the world.

2.7 Summary Above, I suggested that sensibility was a signifcant political value, and that there were two distinct components of sensibility: (1) appreciating (or understanding or

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Epistemic Pathologies of Elections  169 knowing) the world as it is, and (2) responding to the world in light of this appreciation. In this section, I have raised a number of distinct but interrelated (and arguably, compounding) concerns about the sensibility of electoral representative systems, particularly focused on epistemic concerns. It is also worth stressing that although there is reason to be worried about the sensibility of electoral political institutions in the U.S., this does not mean that everything is awful. It is not. Modern electoral democratic governments do many things well, even if imperfectly. It is also true that modern electoral representative governments collect an extraordinary amount of money in taxes, so it should be no surprise that some things get done. Still, it would be a mistake to think that electoral representative democracy is a disaster from a perspective of sensibility. Tat concession made, I think we can do better. In the next part of the chapter, I will introduce and discuss what I call a “lottocratic” system of representative government. In discussing that system, I will focus in particular on potential epistemic advantages of lottocracy.

3.  Considering the Sensibility of Lottocratic Representative Government

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.1  Te Basics of the Lottocratic System Te kind of institution that I am interested in is unusual in that it uses selection of political ofcials by lottery, rather than by election. Tere is some historical precedent for this kind of method, referred to commonly as “sortition.” For example, in ancient Athens, three of the four major governmental institutions were populated by people chosen by lottery. More recently, Citizens’ Assemblies (in which citizens were chosen at random to serve on the assembly, and in which citizens heard from experts prior to coming up with their own proposals) have been involved in formal structures of legislative and constitutional decision-­making in Canada, Iceland, Ireland, Mongolia, and California. Te kind of institution that I want to consider as a possible improvement is an instance of what I call a “lottocratic” legislative institution.14 Te key features of lottocratic legislative institutions are these: 1. Single Issue: there are many single-­issue legislative bodies, with each legislative institution focussing just on one policy area or sub-­area. Each could have a standing role in addressing an issue (perhaps as one node in a network, of, say, 20 such single-­issue legislatures, covering each of, say, agriculture, immigration, health care, trade, education, energy, etc.), or it could be a one-­of institution, brought into existence to make a specifc policy decision. 2. Lottery Selection: the members of each single-­issue legislature are chosen by lottery from the relevant political jurisdiction.

14  Tis part of the chapter draws on Guerrero (2014) and Guerrero (2021).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

170  Alexander Guerrero

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3. Learning Phase: the members of the single-­issue legislature hear from a variety of experts and stakeholders on the relevant topic at the beginning of (and perhaps at various stages throughout) each decision-­making session. 4. Community Engagement: the members of the single-­issue legislature spend some structured time talking to, interacting with, and hearing from members of the public, including activists and stakeholders afected by proposed action. 5. Direct Enactment: the members of the single-­issue legislature either (a) have the capacity to directly enact policy or (b) have the capacity to enact policy if it is co-­authorized jointly with other single-­issue legislatures. Tere are many ways in which one might implement a political system that had institutions with these features. For shorthand, I refer to these single-issue, lottery-­ selected legislatures as “SILLs.” Tere are many questions regarding the details of these systems, their scale, scope, size, and the way in which they will operate in the details. Tese are very important; indeed, I get into them at much greater length in a forthcoming book.15 In what follows, let us consider one version of a lottocratic system, imagining it in one particularly vivid and relatively simple instantiation in the U.S. government, so as to help us consider the comparative question looking at the epistemic merits of standard electoral representative systems in the U.S.  So, here are some of the additional basic features of the lottocratic system I will consider. Imagine that there will be 20 diferent SILLs, divided by issue area (agriculture and nutrition, education, energy, health, transportation, military and defense, environmental protection, communication, regulation of markets, trade, immigration, science and technology, workplace safety, etc.). Tis SILL network replaces the U.S. Congress in the functional role of creating most law and policy, but with the possibility of delegation and supplemental regulation and enforcement through legislatively created administrative agencies that are overseen by a combination of courts and SILLs themselves. Each SILL consists of 450 people, chosen at random, to serve 3-­year terms, with 150 new people starting each year and 150 people fnishing their term each year. All adult citizens in the political jurisdiction would be eligible to be selected. For most issues, a truly random lottery would be conducted. For a select few issues, eforts would be made to ensure a demographically representative selection, using stratifed sampling along various dimensions if necessary. People would not be legally required to serve if selected, but the fnancial incentives would be considerable (perhaps something like 120 per cent of a person’s normal yearly income, with a foor of $100,000), eforts would be made to accommodate family and work schedules (including providing relocation expenses and legal protections so that individuals or their families are not penalized professionally for serving). Tis signifcant salary would be contingent upon a SILL member not having prohibited contact with potentially interested people or entities while serving on the SILL, not receiving money or other forms of infuence or beneft 15  Guerrero (2021).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Epistemic Pathologies of Elections  171 before or afer SILL service (as well as agreeing to be monitored for continued compliance). Tere would be some mechanism of removing people for bad behavior—failing to attend meetings, speaking out of turn, showing up intoxicated or otherwise incapable of participating fully—but this mechanism should be structured so as to protect those who simply are unlikable or who have divergent views. SILL members would be instructed to do what they think is best with regard to the particular policy question, afer having heard from experts, stakeholders, members of the community, and other SILL members. Tey would not be required to see themselves as “representatives” of any particular community or group. Each SILL would meet for two legislative sessions each calendar year, and the structure for each session would be something like this: agenda setting, learning phase with expert presentations, community consultation, deliberation/discussion, drafing, revising, and voting.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.1.1  Agenda Setting Te SILLs will decide the agenda for the next session by a process of agenda setting. Tis process should have some balance of input from those already involved with the issue (experts, stakeholders, activists) in addition to the general public, perhaps through sophisticated deliberative-­ polling and political party organization. Te members of the SILL will take this combination of in-­person proposals and polling information and vote for those items to be put on the agenda for the next legislative session. 3.1.2  Learning Phase For each item on the agenda, the SILL will hear from experts, providing general background and specifc information relevant to the issue. Accordingly, there will be a process by which a person is allowed to speak to a SILL as an expert. Tis requires both a process to determine whether a person counts as an expert (the qualifcation assessment process) and a process to determine which qualifed experts are given an opportunity to speak (the expert selection process). Expertise might be recognized based on advanced degrees; years of professional experience; formal professional credentials from institutions with national or international accreditation; publication of research in independent, peer-­reviewed journals; and so on. A diferent, but also important kind of expertise is the expertise that comes from experience, including occupational experience or lived experience, such as the experience of being a disabled person (particularly in the context of making policy that primarily afects disabled people). Whatever specifc process is used, experts will need to explain the basis of their expertise, describe their credentials (if relevant), and disclose any actual or possible conficts of interest due to sources of funding or employment. Tere are signifcant concerns and complications here. I discuss these issues at length elsewhere.16

16  Guerrero (2017), Guerrero (2021).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

172  Alexander Guerrero

3.1.3  Community Consultation, Deliberation, Drafing, Voting Afer hearing from experts, SILL members will begin the process of developing and deciding upon legislative proposals, and possibly eventually enacting a proposal. For most issues, this process should include consultation with non-­members, either virtually (online) or through having the members return to the geographic area from which they came, and to hold town-­hall style meetings, in which individual members or multi-­member panels talk through the items on the agenda, talk about what the experts told them, and solicit questions and comments from those in attendance. Tere are two main purposes to this: (1) to inform non-­members about the issues and proposals under discussion, and (2) to gather information from members of the community. Te details of the deliberation and consultation phases will matter a great deal. Tere is good evidence that group deliberation can go awry in a number of ways. Deliberation in the full-­group and sub-­groups would take place at various stages, but in a carefully structured way to ensure equal levels of participation, to avoid groupthink (through the use of red-­teaming and other counter-­advocacy measures), and to prevent pressure toward consensus. Tere is a considerable amount of empirical work on how to structure deliberation so as to avoid group polarization and to encourage the maximal epistemic contribution from all of the members of the group.17 A full discussion of the epistemic capacities of lottocratic institutions would require a detailed discussion of the structure of the deliberation and consultation phases, but I will not go further into those issues here. SILL members will then work together to draf proposals. Some of this might be modeled on how the drafing of legislation happens in other legislative bodies, with initial drafs or competing drafs written by diferent committees within the SILL. As with other legislatures, there might be drafing aides and consultants on hand who have expertise in drafing legislation, and who can help spot concerns of the formal, rather than substantive, variety. Tere could also be a period during which drafs are made public and comments are solicited from the broader community. Tere would then be a process by which proposals were put to a vote. In most cases, the votes would be aggregated to determine the result. In some cases, the votes would be put into a hat and ten would be chosen at random, with the winning result being determined by seeing which option was most supported by the ten randomly chosen ballots. Tere is no executive veto of SILL supported legislation.

3.2  Reasons to Expect Comparative Epistemic Merit of Lottocratic Institutions With these admittedly sketchy details of the lottocratic system in view, we can consider some of the reasons why this kind of institution might do comparatively well in epistemic terms. 17  For discussion, see Karpowitz and Mendelberg (2014), Myers (2017).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Epistemic Pathologies of Elections  173

3.2.1  Addressing the Problem of Mass Ignorance Perhaps the central problem for systems of electoral political representation is that voters do not know enough about political issues and about what their political representatives are doing. As noted above, one set of responses simply gives up on robust, egalitarian democratic norms, and suggests moving in an epistocratic or technocratic direction. Te lottocratic system attempts to retain the core democratic elements, but also to address the problem of mass ignorance. Tis is done both through the single-­issue focus, lowering the epistemic burden, and by having certain randomly selected citizens go through a learning phase where they become comparatively better informed about the relevant topic. Tis would substantially improve the epistemic position of those ordinary citizens involved in political decision-­making, while not giving up on core egalitarian political values (everyone would have a literally equal chance of having political power). Indeed, the learning phase by itself, focussing on a single-­issue domain, would be a huge improvement over the way in which expertise currently enters democratic politics. A worry some have is that randomly chosen citizens would simply be of inadequate intellectual capacity to make epistemically responsible policy. Tere is a concern— felt more powerfully by some than by others—that entrusting signifcant policy decisions to a randomly selected body of citizens would be a disaster, much worse than delegation to elected representatives, and a disaster because of the epistemic limitations of ordinary citizens. Perhaps electoral politics has its problems, but at least those selected have to be at least somewhat intelligent, well-­educated, diligent, competent, hard-­working—or so the thought goes. Tere are diferent possible responses to this concern. Te extent to which one is worried about citizen competence may well depend on one’s life experiences and background. Here’s a conjecture: people who have spent all of their time in “elite” institutions tend to think that elites are capable and competent and that others are not; not just that non-­elites have not had the education and training, but that even if they had, they would not be capable or competent. I don’t share this worry, but I think that is a result of my personal history. I grew up attending unexceptional public schools in Washington State. Two-­thirds of the people in my high school did not go on to any kind of college or university; almost no one lef the state for college. But I feel confdent that the vast majority of my classmates would do just fne serving on a SILL. Indeed, having then attended Harvard for college, I would say that the top third of the students in my unexceptional high school—around 200 people, many of whom did not go on to college—could have been swapped in for the bottom ffh of the students in my class at Harvard, without any signifcant diference in general competence or aptitude, perhaps afer a semester or a year of adjustment and training to make up some of the diference in previous preparation. But one might be skeptical of this. For those who remain concerned about competence, one kind of response is to present the many institutional solutions that might increase competence: • creating incentives for the full range of citizens to participate (so that one does not get a skewed sample as with juries)

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

174  Alexander Guerrero

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

• setting reasonable minimum thresholds for the particular policymaking context (e.g. requiring a high school diploma or the equivalent, or even much more education for certain technical domains) • improving public education so that the “worst of ” from a competence perspective are relatively more competent. If we have done these things and the competence question remains, there are a few other responses. First, one can treat the question as a quality threshold question and argue that most citizens who might be selected actually would be competent. Second, one can focus on the question as a comparative question, and attempt to highlight the incompetence of elected ofcials. Tird, one can argue that elected ofcials are perhaps more competent in some sense, but that this is outweighed or undermined by the ways in which they are otherwise epistemically undermined as a result of the various considerations enumerated above. Fourth, one can argue that randomly chosen citizens might actually do better than elected representatives, even if they are of lower average competence, because the randomly chosen members of SILLs are likely to include individuals with a greater range of life experiences and vocational skills than an electoral representative system, which may improve the quality of decision-­making due to improvements in the evidential and intellectual diversity of the group. Fifh, a possible side-­beneft of the lottery-­selection is what we might call the “humility of the randomly chosen.” As Barbara Goodwin (1992) puts the idea: “those allotted high ofce would comport themselves more humbly . . . no one could boast of his/her elevation or advancement as being personally merited.” Tis might provide a reason to think that those randomly selected feel some responsibility to demonstrate epistemic humility, to pay more attention to the issue at hand (and to leave political posturing to the side), and to engage more fully with the questions of what would be best and what people really care about. In the comparative assessment with elected ofcials, all of these might suggest greater relative quality in terms of the competence of the randomly chosen citizens. All these strategies have promise, and they do not compete; indeed, the fve together may be more plausible than any one in isolation. A full efort to develop these responses requires empirical investigation, but it is worth noting that it is not obvious, for example, that the average member of Congress is better able to understand technical policy issues.18 Finally, a salutary side efect of using lottocratic institutions is that it makes evident the need for good public education for all citizens, not just for the wealthy or politically connected members of the citizenry. It may be true that, at the moment, a U.S. public high school education does not prepare one to be a helpful and engaged citizen in the creation of law and policy. But this is something that should itself be the object of reform, not a reason to reject a proposed reform to the political system.

18  See Grifn (2013).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Epistemic Pathologies of Elections  175

3.2.2  Te Relative Difculty of Capture A second thing to stress is that the lottocratic system makes capture considerably more difcult. If one of the signifcant problems regarding the epistemic quality of electoral representative systems is that they are consistently captured, leading them to behave in epistemically unjustifable ways, then preventing or reducing capture may lead to a considerable epistemic improvement. Tere are several reasons to expect that capture will be more difcult. First, random selection, rather than election, makes a huge diference to preventing capture, by eliminating a central point of potential infuence. In general, lotteries, if conducted fairly, excel at preventing undue infuence in the selection of representatives. SILL members are chosen at random and do not need to run for ofce, so there will be no straightforward way for powerful interests to infuence who becomes a SILL member or to ensure that the only viable candidates are those whose interests are congenial to their own. Second, because there is no need to raise funds for re-­election, it would be easier to monitor and restrict members of the SILL to ensure that they are not having contact with or receiving funds from powerful interests either during or afer their service. At least, if this is possible with juries in high profle cases, it should be possible in the case of SILLs. Tird, by paying SILL members a high salary but conditioning that salary on their not being bought of, this makes the price that would be required to buy of a SILL member much higher (much more than $1 million per year, for example), and should dramatically limit the people who might be interested in taking a bribe (since they have to be willing to risk not receiving a defnite $500,000 a year for a chance at receiving some larger amount). Fourth, since SILL membership rotates regularly, the cost of “buying of ” particular SILL members would be higher, even if it could somehow be accomplished. It would not be possible to capture politicians who were virtually unbeatable (from partisan districts with incumbency advantages) and count on them being an ally for 30 or more years. A concern about lottocratic institutions and capture is that powerful interests might try to infuence who is identifed as a qualifed expert and who is selected as an expert to speak. Tis is a concern. Tere are a few possible responses. First, if there are non-­political hurdles to becoming an expert in a particular feld (advanced degrees from nationally and internationally accredited educational institutions, peer-­ reviewed publication, and so on) and if there are disclosure requirements mandating that experts disclose sources of funding, employment, and so on, this concern might be lessened. Second, there can be institutional mechanisms that make capture of experts more challenging—such as having the expert identifcation and selection processes happen in part by the accredited community of experts nominating or certifying some individuals as candidate experts for the SILL process. To achieve capture, then, would require not just buying of an individual, but an entire feld. Tird, there could be institutional mechanisms, that would lead to the creation of an expert database, with the experts who speak at any particular SILL learning phase being chosen at random from this pool of experts. Of course, there is a worry about

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

176  Alexander Guerrero the politicization of expertise under a system that uses experts in this way, or in any way. Tis is a battle that is important for any political system, whether lottocratic, electoral representative, or technocratic. With this kind of system, there is at least some hope that the process will be relatively transparent and non-­partisan, although it is important that those are just comparative claims.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.2.3  Te Elimination of Electoral Incentives that Result in a Short-­term Bias A third reason we might expect to see comparative epistemic virtue from lottocratic institutions comes from removing electoral incentives and the short-­term thinking they encourage. Since lottery-­selected individuals do not have to worry about re-­ election or being able to claim credit, they can take a longer view and research and implement good ideas that might not bear fruit until much further down the road. Of course, they still might not do so, since we can all be prone to near-­term biases. But at least they would not have an additional incentive in the wrong direction, leading to incentives to cultivate false beliefs and misinformation. 3.2.4  Deeper and More Extended Engagement with the Issues Undermines the Efcacy of Purely Emotional Appeals and Rhetoric One of the concerns about the epistemic quality of electoral representative systems was that elections, conducted against a background of inevitably high levels of voter ignorance, tend to focus on simple appeals attempting to resonate with powerful emotions and attitudes: anger, fear, disgust, distrust, hope, pride, and so on. Emotions can, of course, be epistemically very useful—they can be important guides and indicators regarding crucial features of the world, including the moral world. Tat said, they can also be overpowering and are subject to manipulation. One advantage of the lottocratic institutions is that the individuals chosen at random will have a chance to engage with the relevant issues at considerable length. Tis engagement may include powerful emotions and emotional discussions, but it will also include signifcant stretches of less “hot” engagement, allowing for more extended thinking and processing alongside whatever emotional engagement takes place. Tis kind of engagement will be fundamentally diferent than what one gets from television advertisements and three-­minute news segments, all of which are intended to excite, enrage, and entertain, rather than to inform or to result in deeper, more nuanced, and more sophisticated understanding of the issues. Te epistemic benefts of this engagement should be clear. And SILL members do not have to worry about looking “sof” on crime or national defense or military policy, nor do they have to worry about doing something—even if that something makes no sense—in response to high salience events. 3.2.5  Focus on Single-­issues and Extended Engagement Might Reduce In-­group/Out-­group Dynamics Te structure of modern politics in the U.S. is framed around the candidates of two dominant political parties. As Achen and Bartels (2016) demonstrate, partisan loyalty and in-­group/out-­group thinking deeply afect almost every aspect of the electoral process and the political participation of citizens. In particular, it dramatically

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Epistemic Pathologies of Elections  177 afects how we evaluate evidence, what we believe about politicians and political issues, and what issues we take to be most important and in need of urgent response. Tere are a number of reasons to think that the lottocratic system would result in an epistemic improvement due to the reduction in in-­group/out-­group dynamics. First, without elections, we lose the horse-­race element, the explicit confrontation of us versus them, the sense that “our team” will either be stably dominated or dominantly in power for four or however many years. We would not have clearly defned teams, at least not in the same way. Second, the focus would shif away from candidate personalities and toward policy issues and policymaking. We would no longer have to respond to our policy ignorance by trying to pick our favorite person of the bunch. Tis is arguably a rational response to electoral politics in the face of almost complete ignorance of the issues, but it is made unnecessary once the focus is not on deciding which candidate to entrust with power and is instead on deciding which policy would be a good one to enact. Tird, moving from a generalist legislative process to a single-­issue legislative process opens up places for us to identify issues on which we agree, moving us out of the situation where all political and electoral attention is concentrated on those few issues which most deeply divide us. Tis also will help reduce the introduction of misinformation relating to these issues, as there will be no incentive to maintain and reinforce our political divisions. Fourth, if lottocratic institutions make it possible for us to move beyond elite capture and control of political institutions, then we may see other benefts in terms of in-­group/out-­group dynamics. If part of the story of our apparent division is a story of manufactured confict, where the most powerful members of society keep us from working together by creating this sense of two teams (and handing each team a set of policy positions and political candidates that are basically agreeable to the most powerful), then lottery-­selection might be a way of breaking down these divisions. Tis is good for political community, certainly, but it also is good for repairing our epistemic community, allowing us to relearn how to trust and rely on each other, removing the incentives to denigrate the rationality and evidential sources of others in our community, and helping us work together to build the investigative and research institutions that can help us understand and address the most urgent problems we face.

3.2.6  Random-­sampling Leads to Better Descriptive Representativeness which Leads to Improved Epistemic Quality of Decision-­making In this chapter I argued that there were signifcant concerns about the epistemic capacities of electoral representative institutions due to the elected representatives being a highly non-­ representative sample of the whole. Tis meant both that knowledge from diversity of experience and background would be more limited, and that the set of interests of the group would align poorly with the set of interests of the whole political community. In the lottocratic system, things are exactly the reverse. Because individuals are chosen at random from the jurisdiction, they are more likely to be an ideologically,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

178  Alexander Guerrero demographically, and socioeconomically representative sample of the people in the jurisdiction than those individuals who are capable of successfully running for ofce. Unlike with electoral representative systems, a descriptively proportional number of people in the room really will have frsthand knowledge of being an engineer, police ofcer, primary caretaker of children, nurse, construction worker, waitress, cashier, plumber, social worker, truck driver, scientist, single-­parent, non-­native English speaker, disabled person, working-­class person, openly gay person, community college student, welfare recipient, woman, African-­American, Hispanic-­American, Muslim-­American, frst-­generation immigrant, and so on. Better descriptive representativeness does not ensure that SILLs will create policy that is on better epistemic foundations, but it does mean that the range of perspectives involved in making policy will be more similar to the range of perspectives of the polity as a whole, which makes an improved epistemic vantage point likely. It also makes it more likely that those in the room will be thinking about the likely outcomes in terms of their full efects on well-­being, distributive justice, and autonomy—not just for the elite, but for the whole of the political community. Even if each individual SILL member is thinking just of the likely efects of policy on people like them, they are a much more representative sample of all the people, and so this is a comparatively better method.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4. Conclusion In this chapter, I have identifed several potential epistemic pathologies of systems of elected political representation, and I have suggested some reasons to think that an alternative kind of democratic political arrangement—a lottocratic system—might be comparatively better in epistemic terms. Of course, what I have said here is at most suggestive of some reasons to consider using lottocratic institutions, or at least to consider reforming electoral representative institutions. Tere are other considerations beyond epistemic ones that I have not considered, and even those epistemic concerns I have discussed are potentially contentious. Perhaps some of the same worries I raised regarding electoral representative institutions would emerge within lottocratic institutions. Doubtlessly there would be epistemic worries about lottocratic institutions that I have not considered. My hope here is simply to have drawn attention to some unnoticed or inadequately attended to epistemic problems introduced by the use of elections, and to have suggested that perhaps—with some imagination and ingenuity—we might be able to design epistemically better political institutions, while retaining our foundational commitments to egalitarian democratic values.

References Achen, C. and L.  Bartels. 2016. Democracy for Realists: Why Elections Do Not Produce Responsive Government. Princeton, NJ: Princeton University Press. Anzaldua, G. 1987. Borderlands/La Frontera: Te New Mestiza. San Francisco, CA: Aunt Lute Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Epistemic Pathologies of Elections  179 Bartels, L. 1996. “Uninformed Votes: Information Efects in Presidential Elections.” American Journal of Political Science 40(1): 194–230. Bell, D. 2015. Te China Model: Political Meritocracy and the Limits of Democracy. Princeton, NJ: Princeton University Press. Brennan, J. 2016. Against Democracy. Princeton, NJ: Princeton University Press. Ferejohn, J. and J.  Kuklinski, eds. 1990. Information and Democratic Processes. Champaign: University of Illinois Press. Gilens, M. 2012. Afuence and Infuence: Economic Inequality and Political Power in America. Princeton, NJ: Princeton University Press. Goldman, A. 1999. Knowledge in a Social World. Oxford: Oxford University Press. Goodwin, B. 1992. Justice By Lottery. Exeter: Imprint Academic. Grifn, D. 2013. “Citizens, Representatives, and the Myth of the Decision-Making Divide.” Political Behavior 35(2). Guerrero, A. 2014. “Against Elections: Te Lottocratic Alternative.” Philosophy and Public Afairs 42: 135–78. Guerrero, A. 2017. “Living with Ignorance in a World of Experts.” In R.  Peels (eds.), Perspectives on Ignorance from Moral and Social Philosophy. Abingdon: Routledge. Guerrero, A. 2021. Lottocracy: A New Kind of Democracy. Oxford: Oxford University Press. Harding, S. 1993. “Rethinking Standpoint Epistemology: ‘What is Strong Objectivity?’” In L. Alcof and E. Potter (eds.), Feminist Epistemologies. Abingdon: Routledge. Hartsock, N. 1987. “Te Feminist Standpoint: Developing the Ground for a Specifcally Feminist Historical Materialism.” In S.  Harding (eds.), Feminism and Methodology: Social Science Issues. Bloomington: Indiana University Press. Hill Collins, P. 1990. Black Feminist Tought. Abingdon: Routledge. Karpowitz, C. and T.  Mendelberg. 2014. Te Silent Sex: Gender, Deliberation, and Institutions. Princeton, NJ: Princeton University Press. Kuklinski, J. and P.  Quirk. 2000. “Reconsidering the Rational Public: Cognition, Heuristics, and Mass Opinion.” In A.  Lupia, M.  McCubbins, and S.  Popkin (eds.), Elements of Reason: Cognition, Choice, and the Bounds of Rationality. Cambridge: Cambridge University Press. Marks, J. and T.  Sharot. 2019. “Epistemic Spillovers: Learning others’ political views reduces the ability to assess and use their expertise in nonpolitical domains.” Cognition 188: 74–84. Mills, C. 2007. “White Ignorance.” In S.  Sullivan and N.  Tuana (eds.), Race and Epistemologies of Ignorance. New York: SUNY Press. Myers, D. 2017. “Interests, Information, and Minority Infuence in Deliberation.” Journal of Politics 79(3): 804–22. Petersen, E. 2012. “Representatives and Senators: Trends in Member Characteristics Since 1945.” CRS Report for Congress at: http://www.fas.org/sgp/crs/misc/R42365.pdf. Postman, N. 1985. Amusing Ourselves to Death. London: Methuen Publishing. Somin, I. 2013. Democracy and Political Ignorance. Stanford, CA: Stanford University Press. Westen, D. 2007. Te Political Brain: Te Role of Emotion in Deciding the Fate of the Nation. New York: PublicAfairs.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

10

Policy, Ignorance, and the Will of the People Te Case of “Good Immigrants” Kristofer Ahlstrom-­Vij and Jennifer R. Steele

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1. Introduction Te fact that most of us have, at best, a superfcial understanding of politically relevant matters is well established (see, e.g., Delli Carpini and Keeter, 1996, Somin, 2016, and Achen and Bartels, 2016). Indeed, as pointed out by Jefrey Friedman (1998, p. 397), it is “one of the strongest fndings that have been produced by any social science— possibly the strongest.” Or as Scott Althaus (2003) puts the point in relation to the American portion of these results: “If ignorance is bliss, then the pursuit of happiness seems alive and well in American society.” (2003, p. 12). Tere are a number of concerns that we might have in light of this fact. One concern is that a given person’s mistaken beliefs might have negative efects on the rest of us (Brennan, 2011). Tis worry will be proportional to the number of people that, like that person, lack sufcient knowledge. A separate worry, however, is a worry for that person: if they lack sufcient knowledge on relevant political topics, that person’s political choices might fail to refect what they truly want. Tat, at any rate, is the claim we will be defending in what follows. In so doing, we focus on a policy area in the UK where such uninformed preferences are particularly rife: immigration policy (see, e.g., Dufy, 2018). Large-­sample studies tend to suggest that the public want to see the number of immigrants reduced (Ipsos MORI, 2018; Dufy et al., 2017; Blinder and Allen, 2016; Ford and Heath, 2014). However, in Section  2, we report on a recently conducted focus group study that suggests that desires for reduced numbers might ultimately express a desire for an immigration system controlling for immigrants of (perceived) high quality, not a reduction in their overall quantity, where quality is defned in terms of fscal impact. Moreover, the manner in which quality judgments are made and justifed indicates a clear hierarchy of diferent types of evidence for claims about quality, with anecdotal evidence ofen taking priority over statistical evidence. Furthermore, we argue that preferences for an immigration system controlling for such “good immigrants” are problematic. Specifcally, in Section 3, we argue that, if a preference can be expected to shif in one direction or the other, even if the only thing that changed about the person was that some relevant beliefs on her part were corrected—that is, if she in all other relevant respects (consistent with that correction) remained the same—then that preference can be reasonably assumed to rest on Kristoffer Ahlstrom-Vij and Jennifer R. Steele, Policy, Ignorance, and the Will of the People: The Case of “Good Immigrants”. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Kristoffer Ahlstrom-Vij and Jennifer R. Steele. DOI: 10.1093/oso/9780192893338.003.0011

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  181 those mistaken beliefs. In Section 3, we also make the case that preferences resting on mistaken beliefs can fail to refect what the person truly desires. Finally, in Section 4, we construct and deploy a number of counterfactual models, the estimates of which suggest that the relevant immigration preferences ft this pattern, in that they can be expected to shif in the relevant way, had we been more informed, and as such arguably do not capture what we ultimately want. Consequently, whatever else might be said in favor of restrictive immigration policy, the idea that they refect public preferences is questionable. It should be noted that, while our focus throughout will be on public preferences on immigration policy in a UK context, the implications of our arguments go beyond that specifc case. As pointed out by Althaus (2003) in relation to political opinion more generally, “strong and stable majority opinions can ofen be the fgment of the public’s misinformation” (2003, p. 22), which “call[s] into question many of the ways opinion data are used by political leaders, journalists, and the mass public” (p. 25). Indeed, given the widespread public ignorance on a variety of key political issues, what follows serves to highlight the ofen-­overlooked problem that public policies based on popular sentiments in many cases may not truly capture “the will of the people.”

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Quality over Quantity, Anecdotes over Statistics Te literature on immigration attitudes has two primary methodological strands: large-­sample, online surveys (e.g. Ipsos MORI, 2018; Dufy et al., 2017; Blinder and Allen, 2016; Ford and Heath, 2014) on the one hand, and experimental studies on the other (e.g. Hopkins et al., 2018; Grigorief et al., 2016; Lawrence and Sides, 2014; Esses et al., 2011; Deaux et al., 2010; Sides and Citrin, 2007). More recently, a third strand has been gaining traction. It employs focus groups in an attempt to complement aforementioned approaches by engaging participants in extended, face-­to-­face conversations about the reasons for their attitudes and preferences (e.g. Rutter and Carter, 2018; Katwala et al., 2014). In an attempt to better understand the interplay between the reported desire for reduced numbers that, as already noted, tends to crop up in large-­sample surveys, and a policy focus on control (e.g. Migration Advisory Committee, 2018), we convened twelve focus groups with a total of 97 participants in Sittingbourne, UK.1 We chose Sittingbourne because it has a number of features—for example, in regards to local results in the 2016 referendum on the UK’s membership of the EU (62.5 percent for “Leave” compared to a national result of 51.9) and levels of unemployment and economic activity—that makes if fairly representative of the type of “lef behind” area that is ofen referred to by those wishing to explain patterns of “Leave” voting in those terms (e.g. Ford and Goodwin, 2014). At the point of recruitment, standard demographic information was collected on each potential participant, to ensure roughly balanced groups along demographic 1  For a complete account of recruitment and methodology, see Rolfe et al. (2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

182  Kristoffer Ahlstrom-Vij and Jennifer R. Steele variables. Participants were also screened to exclude people who were either highly positive or highly negative about immigration, following the standards set by previous focus group research on immigration (e.g. Rutter and Carter, 2018). In terms of attitudinal variables, participants overall tended to want the government to use Brexit as an opportunity to cut down on EU immigration, which is in line with aforementioned, national survey data on the issue (e.g. Ipsos MORI,  2018; Dufy et al., 2017; Blinder and Allen, 2016; Ford and Heath, 2014).2 However, when this preference was elaborated on in the focus groups, it was ofen framed in terms of how there should be increased controls on immigration, rather than directly in terms of a reduction in numbers. Of course, the way to reduce numbers would be through the imposition of some kind of control—in the limiting case, by closing the country’s borders completely (a suggestion that only came up once across the twelve focus groups). But in virtually all cases where it was discussed, controls were invoked, not as a means to reduce the quantity of immigrants, but rather as a tool for ensuring the (perceived) quality of immigrants coming in. What makes for a “good” (i.e. “high-­quality”) immigrant? Based on focus group discussion, we defne a “good immigrant” as someone who (a) is immigrating to work, and (b) does not take out more than they put in with respect to tax-­funded services and benefts, such as the National Health Service (NHS) and welfare benefts.3 Te two aspects are intertwined: “good immigrants” beneft the UK (by working), and do not cost the country anything (by not being net recipients). Indeed, in general, there was a very pronounced form of instrumentalism underlying this sentiment about high- and low-­quality migrants in the focus groups. As one participant put it: Prove your worth and then you can stay. When it comes to tax-­funded services and benefts, this instrumentalism can be captured in terms of two forms of access: earned access and default access. Earned access is what immigrants have; they have access, provided they have already paid in to cover their share. As one participant explained, if you’re not putting anything into the system you shouldn’t be able to take anything out of the system. Default access, by contrast, is the type of access had by citizens, who can take out even if they have not paid anything in. Consider welfare benefts, for example. Several discussions about young British adults on welfare benefts made clear that participants felt that citizens may take out without having put in. Disapproval of those who do so was registered, not by suggesting that the relevant claimants had no right to claim to begin with, but by the suggestion that they needed to be encouraged or pushed to work. Moreover, the manner in which our participants discussed quality suggested a clear hierarchy of diferent types of evidence for claims about quality, with anecdotal evidence in the form of personal experiences or experiences of family and friends at the top, and statistical evidence at the bottom. For our purposes, anecdotal evidence can be understood as evidence framed in terms of a single or a very small number of 2  See Rolfe et al. (2018) for a more details on the sample. 3  Tis is similar to Hainmueller and Hopkins’ (2014) notion of a “preferred immigrant,” someone who “is well educated and in a high-­status occupation, with plans to work, good English skills, and no prior unauthorized entries” (2014, p. 239).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  183 instances, while statistical evidence can be characterized as evidence framed in terms of a (typically numerical) summary of a relatively large number of instances, ofen in the form of a summary statistic like a mean or average. Participants had a store of anecdotal evidence from the accounts of friends, family, and acquaintances, which they used to support specifc and popular themes around migration impacts. Of course, beliefs grounded in anecdotal evidence are not necessarily false, nor are they necessarily unjustifed. However, beliefs that (a) are grounded in anecdotal evidence, (b) concern generalities (i.e. not just the individual instances mentioned in the anecdote), and that moreover (c) confict with the statistical evidence on the matter are on epistemically thin ice. So, for example, if I believe that EU immigrants tend to constitute a drain on the UK’s public fnances on account of having heard about a case of an EU immigrant that seemingly is, but the statistical evidence on the matter suggests that EU immigrants tend to make a positive net contribution (Oxford Economics, 2018), then my belief does not rest on solid ground. Tis, moreover, was the general format of the type of anecdotally based beliefs that we came across in the focus groups. In light of this, we ofer the following, two-­pronged hypothesis:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Anecdotes about Quality (AQ): Popularly cited desires for reduced numbers of immigrants may ultimately refect a perceived need to screen for ‘good immigrants,’ and judgments about the quality of immigrants are typically made with reference to anecdotal (as opposed to statistical) evidence of their fscal impact.

Two things should be noted here: First, people’s preferences for anecdotal over statistical evidence might be the result of the availability heuristic, whereby information that is vivid or otherwise easily retrievable has more cognitive sway than other types of information (e.g. Kahneman, 2011). However, AQ does not assume that statistical evidence is less persuasive than anecdotal evidence all else being equal. Tat claim is controversial (see, e.g., Hornikx, 2005). In particular, AQ is silent on whether statistical evidence is less persuasive than anecdotal evidence under the relevant circumstances because it is statistical, as opposed to (say) because people are skeptical about the motives of the entities that typically provide them with such statistical evidence (e.g. the government). Second, as it stands, AQ is compatible with so-­called “aversive racism” (e.g. Gaertner and Dovidio, 1986; Dovidio and Gaertner, 2004). According to some aversive racism research, when racist or xenophobic sentiments against particular groups fnd expression, they will tend to be framed in more socially acceptable terms, including economic ones. Tis could explain why participants in our focus groups framed anti-­immigration sentiments specifcally in fscal terms, and moreover were willing in particular to express greater negativity towards unskilled immigrants as compared to highly skilled immigrants, the former of whom are easier to call into question on economic grounds without coming across as xenophobic. We will return to the relationship between AQ and aversive racism in Section 4.3. below.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

184  Kristoffer Ahlstrom-Vij and Jennifer R. Steele

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Te Problem with Preferences Based on Mistaken Beliefs Let us assume that AQ is correct: people generally want to see the imposition of immigration controls to select for “good immigrants,” even if their judgments about whether any given type of immigrant will tend to be of high quality might rest on questionable epistemic foundations. In light of this, it might be suggested that what policy should do is honor the public’s desire for a system selecting for quality, while disregarding their particular, anecdotally grounded (and as such potentially misguided) judgments about where quality can be found. Tis seems to be the approach of the Migration Advisory Committee—the independent public body that advises the UK government on migration issues—in their fnal report on European Economic Area (EEA) migration to the UK (Migration Advisory Committee, 2018).4 Te report reinforces previous fndings on the absence of any substantial labor market impact, a likely small but positive net fscal impact, and the absence of any adverse efect on public services of EEA immigration to the UK. At the same time, the recommendation ofered by the Committee is to impose a tiered immigration system for EEA migrants that attempts to protect the UK from any adverse economic and fscal impact—presumably because that might be in line with popular sentiments about control. However, in what follows, we will argue that—if AQ is correct—it is not just that people’s judgments about quality are problematic due to their anecdotal grounding, as discussed in the previous section. People’s very preference for a system that selects for “good immigrants” is problematic. Specifcally, we will argue that such preferences can be reasonably assumed to rest on false beliefs. What is the problem with preferences based on false beliefs? If we follow Hume (2003/1740), we might say that they are unreasonable, since they are “founded on false suppositions”; if we follow Brandt (1979), we might say that they are irrational. But whatever term of disapproval we opt for here, the fundamental problem with preferences grounded in false belief is that it is not obvious that they correspond to what the person truly wants, in some deeper sense.5 A hypothetical example will help here: Imagine that we are voting on a new water supply. Your preference is for an option—call it option A—that, unbeknownst to you, would involve drawing the water from a contaminated source. Let us assume that, were you to fnd out about the contamination, your preference on the matter would shif. Specifcally, assume that such a shif would have occurred even if the only thing that changed about you was that your belief concerning the safety of the water source was corrected; in every other respect consistent with that correction, you would have remained exactly the same. Note that this is not to suggest that the only thing that changes about you is that you shed a belief.6 Afer all, in the process of correction, you might take on

4  Te European Economic Area includes all EU countries as well as Iceland, Liechtenstein, and Norway. 5  Harsanyi (1997) defends such a view and Goodin (1995) either defends or comes close to defending it. See also Delli Carpini and Keeter (1996, p. 5), who suggest that “the real interest of an individual—and by extension of a group and of the polity as a whole – are refected in the choices one would make if he or she were fully informed about the consequences.” 6  Tanks to Robert Talisse for making this point.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  185 board new beliefs, including ones that call into question your grounds for believing what you previously believed, and as such do the cognitive work in ridding you of that belief. Additionally, you might shed not just that one belief, but potentially also beliefs that supported it but are no longer compatible with the way you now take the world to be. We can think of this in terms of there being some partition of your entire doxastic structure that, prior to the correction, contains some false belief together with some grounds, and that, afer the correction, potentially contains a new belief with some new grounds. When we say that the frst belief is being corrected, what we are saying is that this partition is being revised in a manner that goes beyond simply shedding the false belief. Returning to our example, we can now ask: what is the relationship between your preference and your false belief? Consider the following:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Counterfactual test: If someone’s preferences for option O would have changed, even if the only thing that changed about them was that their false belief B were corrected, then it is reasonable to assume that their preference for O rests on a false belief, namely B.

Te idea behind this test is that it concerns certain modal correlations, on which certain preferences and (false) beliefs co-­vary across modal space.7 Of course, we cannot infer from the mere fact of two things being correlated that one is caused by the other. But correlations can of course provide evidence that is consistent with causation—particularly in cases where confounds are in efect controlled for by requiring, as in this case, that the belief correction is the only thing that changes—and the idea is that the type of modal correlations tested for through our counterfactual test do just that. So, returning to our case from earlier: under this test, we have reason to believe that you prefer A because of your mistaken belief about the safety of the corresponding water source. Moreover, if that is so, we may ask: do you really want option A? It is clear that you want it in some sense; if you vote for A, you are doing what you want to do in the sense that, from a purely psychological point of view, the action expresses a desire on your part that, together with your mistaken belief, helps explain or otherwise make sense of what you are doing. At the same time, there is also a sense in which, were you to vote for A, you would not be doing what you wanted to. Te best way to show that is to perform what we might call a retrospection test: Retrospection test: Imagine someone having registered a preference for option O subsequently coming to realize their mistake about belief B on which that preference rests, and being asked about their original preference ‘So, was O really what you wanted?’. If they can reasonably be expected to answer ‘No,’ then their (misguided) preference for O does not refect what they want.

7  Tanks to Alex Grzankowski for the suggestion to frame the object of the test in terms of modal correlations.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

186  Kristoffer Ahlstrom-Vij and Jennifer R. Steele Running this test on the case of your vote for A, we imagine that you came to realize your mistake about the safety of the water, and that we then asked you about your original preference “So, was that really what you wanted?” You can reasonably be expected to answer “No,” in which case your (misguided) preference does not refect what you want. Let us refer to someone who can be expected to answer “No” under such circumstances as someone who fails the retrospection test.8 Someone might object that, in the type of case imagined, it is not that the agent’s original choice failed to be in accordance with their real preferences, but rather that their (real) preferences changed. We may call into question this alternative reading through what we will call the argument from regret (see Ahlstrom-­Vij, 2020). Return to the case of your misinformed vote for A. Let us assume that you come to realize your mistake. It would be appropriate for you now to feel regret. But such regret would not make sense had your original expression been in accordance with your real preferences on the matter. Afer all, on the alternative reading under consideration, the relevant regret simply comes down to this: you regret that your less informed self supported a policy that you now consider a bad one. But that cannot be it. If it were, there would be no diference between how you feel about your own past support, and how you feel about some other person who might have supported that same bad policy. In both cases, you fnd it lamentable that a bad policy has received support. But you will feel more troubled by your own past support than about that of some other person, and that is what the alternative reading under consideration cannot account for. Because of your false belief about the merits of the policy in question, you also let yourself down in a very specifc sense: you acted contrary to what you actually preferred. Tat is why it is appropriate for you to feel regret, and for you moreover to answer the question “Was that really what you wanted?” in the negative. If this is correct, it creates problems for any attempt to do public policy on the basis of preferences grounded in mistaken beliefs. But there are two worries that need to be kept separate here. One is a worry for us, in that a given person’s mistaken beliefs might have negative efects on the rest of us (e.g. Brennan, 2011). Tis is the well-­known problem of public ignorance in the political arena (e.g. Ahlstrom-­ Vij, 2019). But as noted already at the outset, this worry will be proportional to the number of people who, like that person, lack sufcient information. A separate worry, however, is a worry for that person: if he or she is not well informed, and on that account fails the retrospection test, there is a sense in which that person’s desire on the matter fails to be registered in the process. Returning to our example, you do not want the water to be drawn from a contaminated source, but that preference of yours does not get factored in; even worse, a vote expressing a contrary (or at least incompatible) preference does get factored in. Crucially, even if you are the only one to whom this happens, this hardly compensates for your lack of a voice on the matter.

8  Goodin (1995, p. 137) has in mind similar types of cases, when he notes that, under circumstances of ignorance, “we can serve a person’s ‘real’ preferences only by censoring the misleading indication of his preferences that is revealed in his choices.”

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  187

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.  Preferences for a System Selecting for Quality Likely Rest on Mistaken Beliefs We argued in the previous section that, if people’s preferences can be expected to shif one way or the other, even if the only thing that changed about them was that some relevant set of beliefs was set right, we can reasonably infer that the preferences in question rest on (those) false beliefs. Moreover, the problem with preferences that rest on false beliefs is that they fail to refect what people want, on the deeper level revealed through the retrospection test. We will now argue that the aforementioned preferences for an immigration system that imposes controls selecting for quality immigrants ft this pattern: they can be expected to shif even if the only thing that changed about the person is that they became more politically informed—a process that we are assuming would involve having at least some (and quite possibly several) false beliefs corrected. From this we can reasonably infer that those preferences rest on mistaken beliefs, and as such fail the retrospection test. To begin with, note that, if AQ is correct, then (stated) concerns about the numbers of immigrants are ultimately about controlling for the (perceived) quality of immigrants, which in turn means that a weaker (reported) preference for a reduction in numbers likely means a weaker preference for such control. Ten, consider that a fairly large number of studies have found that more educated people tend to report more liberal (i.e. less restrictive) preferences regarding immigration numbers (see, e.g., Braakman et al.,  2017; Dustmann and Preston,  2007; Card et al.,  2005). Tis might be taken to ofer some reason to believe that the relevant preferences are liable to shif as people become more informed, as higher degrees of education would typically be taken to be associated with greater amounts of information. At the same time, the problem is that education is associated with a whole host of other factors as well. As Hainmueller and Hopkins, (2014, p. 241) note, while “[n]umerous studies have shown that education is perhaps the most powerful predictor of pro-­ immigration attitudes, . . . the interpretation of this correlation is contested because of the multiple mechanisms through which education may act.” For example, it might be that education, in addition to (hopefully) rendering the person more informed, also inculcates a variety of distinctly liberal values, such as tolerance and anti-­racism, that in turn explain the shif towards less restrictive attitudes on immigration. For that reason, we need some way of isolating the efect that the degree to which people are informed might have on immigration attitudes. Here, it is worthwhile to consider the literature on political knowledge tests, and in particular the type of scales developed by Delli Carpini and Keeter (1993,  1996), and subsequently deployed by (among others) Althaus (2003), Caplan (2007), Oscarsson (2007), Blais et al. (2008), Hansen (2009), Bhatti (2010), and Ahlstrom-­Vij (2020), to explore the role of knowledge in political attitude formation and choice. Against the background of this tradition, large-­scale political surveys, such as the American National Election Survey (ANES) in the U.S. and the British Election Study (BES) in the UK, typically include a number of knowledge items in their national surveys. Tis makes it possible to tease out what combinations of demographic factors (e.g. age, gender, ethnicity) and levels of political knowledge tend to go with specifc political attitudes, and

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

188  Kristoffer Ahlstrom-Vij and Jennifer R. Steele subsequently also estimate what political positions people likely would have taken, had they been informed—a practice typically referred to as counterfactual modeling, a type of causal modeling.9 On an intuitive level, the idea is that we are estimating what would happen if we could somehow administer a “knowledge pill” to all respondents that would bring about a (counterfactual) situation where each of them becomes informed but otherwise remains exactly the way they are. By then looking at any estimated changes in attitudes of interest, we can attempt to isolate the specifc efect of knowledge on attitudes, in the manner that we are afer here.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.1  Dataset, Knowledge Scale, and Model Specifcation For the purposes of estimating what the UK public’s immigration preferences would have been, had they been informed, we used the November 2019 wave of the 2014–2023 British Election Study Internet Panel (Fieldhouse et al., 2020), featuring responses from 34,366 UK respondents. For our dependent variable, we used the question “Some people think that the UK should allow many more immigrants to the come to the UK to live, and others think that the UK should allow many fewer immigrants. Where would you place yourself on this scale?” (0 = many fewer and 10 = many more). Specifcally, a binary variable was constructed to take on a value of 1 for responses of 4 or less, indicating a desire to see immigration levels reduced, and 0 otherwise. We also constructed a measure of how informed respondents were, by using four true/false knowledge items from the dataset to ft an item response theory (IRT) model—an established way to model a continuous trait, such as an ability or level of knowledge, on the basis of a set of response patterns10—on the basis of which each respondent was assigned a numerical value, representing the degree to which they are politically informed.11 Four items might sound like a small set. However, as noted by Delli Carpini and Keeter (1993) (see also Althaus, 2003), provided that scales are made up of items within the broad categories of what government is and does, and of political leaders and parties, they do not need to be long to be diagnostic (for example, Delli Carpini and Keeter’s primary scale consists of fve items). Moreover, since people tend to be generalists in the domain of political knowledge—if someone knows (or does not know) a lot about one area of politics, they will tend (not) to know a lot about other areas—Delli Carpini and Keeter (1996, p. 174) note that “researchers developing national or general political knowledge scales need not be overly concerned with the mix of specifc topics covered by individual items.”

9  Pearl (2000) is the central text here. See also Morgan and Winship (2015) on causal modeling in the social sciences, and Keele (2015) for an overview of causal inference in political science in particular. 10  See DeMars (2010) for an accessible introduction to IRT modelling, and de Ayala (2009) for a more comprehensive treatment. 11  Tese statements were: (1) Polling stations close at 10.00pm on election day; (2) No-­one may stand for parliament unless they pay a deposit; (3) MPs from diferent parties are on parliamentary committees together; and (4) Te number of MPs in Parliament is about 100. See the Appendix for more details on the IRT model used here, including model diagnostics.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Policy, Ignorance, and the Will of the People  189 Moreover, note also that (contra Lupia, 2006) the suggestion is not that competent preference formation requires knowing the answer to the specifc items used in measuring political competency. Te claim is that knowing the answer to such small sets of questions is diagnostic of someone knowing things that are necessary for such competency. Tere are two reasons for thinking this. First, the relevant type of scales tends to have a high degree of internal consistency, and correlate well with both interviewer ratings of the degree to which respondents are informed, and with political behaviors that are reasonably related to the level of a person’s political knowledge, such as degree of political participation (Delli Carpini and Keeter, 1993). Second, the efects of information tend to be consistently higher for groups that we have independent reason to believe will be less politically informed, such as high-­school dropouts, people who are poorer, and people in rural areas; and consistently lower for groups we have reason to believe will be more politically informed, such as those who are college educated, afuent, and living in cities or suburbs (Althaus, 2003).12 Having identifed our dependent variable—tapping into a desire to see reduced levels of immigration—and the independent variable of interest, capturing respondents’ level of political knowledge, we then constructed three counterfactual models, each encoding slightly diferent assumptions about what specifc factors play a causal role in political preference formation, in addition to someone’s level of knowledge.13 In all models, we aimed to control for any confounds that can be expected to have a causal efect on both someone’s degree of knowledge and their political attitudes or preferences, namely gender (vanHeerde-­Hudson, 2020; Plutzer, 2020), level of education (Hebbelstrup and Rasmussen, 2016) and income (Vowles, 2020; Plutzer, 2020). Moreover, to reduce noise in these models, each model also controlled for variables that can be expected to infuence someone’s political preferences, but not necessarily their degree of knowledge, such as ethnicity (e.g. through a “shared faith”; Dawson, 1994), religion (Evans and Northmore-­Ball, 2020), marital status (Denver, 2008), and age (Plutzer, 2020). Te variables mentioned so far—that is, political knowledge, gender, education, income, ethnicity, religion, and marital status—account for all variables included in our frst model, which we will call the purely demographic model. In our second, party identity model, we controlled for everything in the demographic model as well as for partisanship, as measured by party identifcation. Te reason for separating out this variable is that it might actually be either inappropriate or unnecessary to control for it. It is inappropriate if partisanship is afected by political knowledge, for example, knowledge of party and candidate positioning (Brader and Tucker, 2018). Tis would make it a mediator in the language of causal modeling.14 Note that something being a mediator for some variable’s efect on an outcome does not rule out that variable also 12  As discussed in the Appendix, this is consistent with the dataset used here in that those estimated to be above average levels of political knowledge are found in greater proportions among the more educated and those with higher household incomes. 13  For complete information on coding and sample sizes for the diferent levels of factor coded variables, see the Appendix. 14  See Rohrer (2018) for an excellent and accessible discussion of mediators, and more generally the problems that arise for approaches that operate by the assumption that including more control variables is always better than including less.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

190  Kristoffer Ahlstrom-Vij and Jennifer R. Steele acting on the outcome directly. In either scenario, however, controlling for a mediator, in this case a causal node located on a direct or indirect pathway between political knowledge and political preference, will mean misestimating the relevant causal efect, by factoring out all or part of it. Alternatively, if partisanship is not a mediator, controlling for partisanship is likely unnecessary. Partisanship is shaped by socialization early in life (Campbell et al.,  1960) and centered around group-­identity considerations relating to religion, ethnicity, gender, and the like. Consequently, controlling for such group-­level variables (here: gender, ethnicity, and religion) may already account for partisanship. For these reasons, the purely demographic model is slightly better motivated than the party identity model. However, since people might reasonably disagree, and for the purposes of gauging the robustness of estimates across diferent model specifcations, we included the party identity model as our second model. For the same reason, we also included our third model—an EU referendum identity model—which, in addition to the variables controlled for in the party identify model, also controlled for the respondents’ vote (“Leave,” “Remain,” or “Did not vote”) in the 2016 EU referendum. Research suggests that UK voters’ identifcation with “Leave” or “Remain” camps have become political identities in their own rights (Hobolt et al.,  2020). If correct, then much like partisanship, referendum vote choice is likely an unnecessary control because it is a function of socialization variables already controlled for. Alternatively, referendum vote choice is a mediator for knowledge, and as such an inappropriate control variable. But, again, reasonable people might disagree, and by adding the EU referendum identity model as our third model and being able to compare the three, we also get a better sense of the robustness of our estimates.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.2 Results Te models outlined in Section 4.1 were ftted using “doubly robust” estimation, to enable causal inference from observational (non-­ experimental) data (see the Appendix for details). To estimate the efect of knowledge on support for the idea that immigration levels should be reduced, we compared the proportion of support in the UK public with our model estimates of that proportion in an informed population. Te former was estimated by determining the proportion of support in the BES dataset for the idea of reducing immigration levels, afer having weighted each observation using the survey weights included in the dataset, to approximate representativeness. Tis gave a small majority of 53 percent in support of reducing ­immigration, which is in line with the large-­sample surveys mentioned at the outset of this chapter. Te informed proportion of support was calculated by estimating the proportion of support we would have seen, had every participant been “informed,” operationalized as achieving a knowledge score above the (actual) mean.15 We can think of this as the model of the imaginary “knowledge pill” mentioned earlier, that renders each 15  See Appendix for more details.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  191 respondent informed, but otherwise leaves them exactly the way they are (across measured covariates). To make matters concrete, imagine that—prior to the “knowledge pill”—a respondent reported wanting to see levels reduced. Ten, we set the value of their knowledge variable to be above the mean (or simply leave it there, if they already are “informed,” as defned earlier), and ask the models to estimate how likely the respondent would now be to want to see levels reduced. Each model will then assign a probability in the range of 0–1 to their wanting to see levels reduced, if informed. By doing this for each respondent, aggregating all probabilities, and fnally using the same survey weights as above, we can calculate the “informed” proportion of support for reduced levels. Tis proportion was estimated to be 43 percent on the demographic model, 44 percent on the party identity model, and 46 percent on the EU referendum identity model, respectively. In other words, given an informed UK public, aggregate support for the idea that immigration levels should be reduced would likely drop by 7–10 percentage points, with the highest end of that range representing the estimate of the demographic model (a drop from 53 to 43 percent). Our results thereby suggest that attitudes about appropriate levels of immigration can vary with the degree to which one is politically informed as such, when holding constant level of education as well as a number of other variables that likely play a causal role in political attitude formation. Read in this context, the education efect that has been found consistently in studies on immigration attitudes looks less contestable; at the very least, we should increase our credence in the claim that, whether education has an efect on attitudes through inculcating values, it may also have an efect by increasing the degree to which people become informed. More to the present point, however, our model estimates give us reason to believe that becoming politically informed could shif one’s preferences in a more progressive (i.e. less restrictive) direction in relation to immigration, at least in a UK context.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.3  Aversive Racism Revisited At this point, we are also in a position to return to the question of how AQ relates to aversive racism. Remember, on AQ, expressed preferences for reduced numbers of immigrants ultimately refect a desire for controls on the quality of immigrants. In our focus groups, such quality was framed in fscal terms. On the theory of aversive racism, such fscal judgments might in turn also mask racist sentiments framed otherwise for reasons of social desirability. Our results in Section 4.2 do not rule this out, nor are they meant to rule that out. Te results instead suggest simply that, had people been more informed, they would have held less negative views on immigration. But why? One possibility is that people who are more informed tend to hold fewer mistaken beliefs about immigration—either about fscal matters specifcally or about a broader set of relevant topics (see Section  4.4)—and thereby fnd less to object to about immigrants coming into the country. In this case, one’s level of knowledge can operate more or less directly on one’s attitudes. However, turning to aversive racism, it could also be that people who are more informed will no longer have available to them the type of arguments that they might have attempted had they known less,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

192  Kristoffer Ahlstrom-Vij and Jennifer R. Steele and that would have helped them to justify beliefs about immigration restrictions, and thereby couch any racist sentiments in (more) respectable, economic terms. On this possibility, one’s level of knowledge operates on both one’s attitudes but not necessarily on one’s level of racism, which of course does not rule out the possibility that the latter could also have an independent efect on one’s attitudes toward immigration. Since our models do not explicitly incorporate a measure of racism, we cannot speak to which of the two possibilities is most in line with the mechanisms involved. At the same time, since it is possible that partisanship and EU referendum vote might capture aspects of racist attitudes, our results ofer some support for the claim that AQ cannot be explained away with reference to aversive racism. Tat is, the fact that we see a signifcant efect of knowledge on all models suggests that levels of knowledge play a role in the relevant attitudes, whether or not racism does as well. However, getting clearer on the exact relationship between knowledge, racist sentiments, and immigration attitudes would require more empirical investigation, and as such points to a promising avenue for future research.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.4  False Beliefs about What? If AQ holds, our results suggest that becoming more informed but otherwise remaining just the way they are can reduces people’s desire for an immigration system that selects for “good immigrants.” What, then, are the false beliefs that underpin people’s preferences on this score? We do not have a defnitive answer to this question, but there is certainly no shortage of candidates for what people might be mistaken about. People overestimate the proportion of immigrants in their country by a wide ­margin—in the UK, by a factor of two (Blinder, 2015)—and moreover, the more they overestimate, the more they tend to want to see more restrictive immigration policies (Hainmueller and Hopkins,  2014); people are mistaken about the make-­up of the immigrant population, for example by tending to think that refugees and asylum seekers make up a far larger share than they in fact do (Dufy, 2018); they are also mistaken about typical immigrant behavior, including believing that increasing levels of immigration leads to increasing levels of crime, whereas in the UK, the share of the population born in foreign countries is unrelated to the frequency of violent crime, and a growing foreign-­born population has coincided with a reduction in property crime (Bell and Machin, 2013); and people believe that immigrants make up a far larger share of the prison population than they in fact do (in the UK, by a factor of three Dufy and Stannard, 2017). To avoid confusion, it is important to note that, in so far as people hold mistaken beliefs on any of these matters, they have arrived at mistaken statistical conclusions, as in: a (mistaken) conclusion framed in terms of some summary statistic, such as an average. What we are interested in here, however, is the manner through which they have arrived at those conclusions. For that reason, we need to distinguish between statistical conclusions, on the one hand, and anecdotal or statistical evidence, on the one hand—the latter being our main concern. For example, I may conclude that the average x is F on the basis of encountering a single instance that is. In that case,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  193 I  have arrived at a statistical conclusion on the basis of anecdotal evidence. Alternatively, I could arrive at the same conclusion through statistical evidence, for example by establishing through a combination of large-­scale data and statistical inference that the preponderance of xs are indeed F. And what we want to note in relation to the abovementioned candidate misconceptions is that, for each of them, a correct understanding of the relevant facts likely requires taking on board exactly the type of statistical evidence that, at least in our focus groups, tended to be ignored in favor of anecdotal evidence. In light of this, it cannot be ruled out that increased degrees of political knowledge are associated with a greater appreciation for, and handle on, relevant statistical information, which could be a factor in any correction of mistaken beliefs that occurs at a higher degree of knowledge. Moreover, given the substantial shifs we saw on reported preferences about numbers of immigrants in the context of our counterfactual models, there is good reason to believe that preferences for controlling for the quality of immigrants are problematic. Remember, if AQ is correct, then stated concerns about the numbers of immigrants are ultimately about fscal impact, and about controlling immigrants’ access to tax-­funded services and benefts in particular. Tis in turn means that a progressive shif towards a weaker reported preference for a reduction in numbers means a weaker preference for such control—possibly because a correction of misconceptions makes clear that such controls might not be needed. In line with the argument made in Section 3, our combined focus group and modeling results ofer reason to believe that expressed preferences on the matter will in many cases not capture what we, on a deeper level, want. Consequently, whatever else might be said in favor of a restrictive immigration policy, the idea that they refect public preferences is questionable.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

5. Conclusion It is well established that the general population typically lacks in-­depth knowledge about key political and policy matters. What are the implications of such ignorance for policymaking? In this chapter, we considered this question in the context of immigration policy in particular. We reported briefy on a recent focus group study that suggests that reported desires for reduced numbers might ultimately refect a desire for immigrants of (perceived) high quality, not a reduction in overall quantity, where quality is defned in terms of fscal impact. Moreover, we argued that public preferences for such “good immigrants” are problematic. Specifcally, we developed and deployed a number of counterfactual models which suggested that such preferences are based on mistaken beliefs—mistakes that could refect a tendency to favor anecdotal over statistical information—and that, on that account, they fail to refect what the person truly wants. As noted at the very outset, while our focus throughout has been on public preferences on immigration in a UK context, the inference (defended in Section 3) from preferences being grounded in false belief to policies based on those preferences not capturing what the people involved truly desires is a perfectly general one. Indeed,

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

194  Kristoffer Ahlstrom-Vij and Jennifer R. Steele given widespread public ignorance, this serves to highlight an ofen-­overlooked problem with public ignorance for the extent to which policies implemented with reference to popular sentiments truly capture “the will of the people.” Almost two decades ago, Althaus made this point when suggesting that “the public’s low levels and uneven social distribution of political knowledge diminish the quality of political representation provided by opinion surveys” (2003, p. 3). Citizens are well-­ represented when political decisions are made in accordance with their political preferences. Tat is why estimates suggesting that the distribution of opinions we would have seen, had all citizens had the luxury of more information, raise questions about whether policymakers catering to the commitments reported by the public in a survey makes for good political representation. Perhaps this might seem like elitism. However, claiming that citizens’ true political preferences are the ones they would have held had they been fully informed, in the manner we have done, is simply to take the political sentiments of the public seriously enough not to confuse them with mere whims and personal tastes. Political judgments are substantive commitments sensitive to facts, and as such they do not foat freely from factual beliefs. If they did, the manner in which facts are brought to bear in response to challenges of mistaken belief would be “a mere conjuring trick,” as Peter Railton (1986, p. 15) puts it. Instead, appeals to facts are an integral part of how people make sense of their values, to others as well as to themselves. In making such appeals, citizens seek “a toehold [in] facts whose truth-­values do not fuctuate with our particular desires or decisions . . . so that this toehold can support us even when we fnd ourselves in a world with people whose beliefs or ends difer from our own” (Railton, 1986, p. 16). Hence, there is a need to ask whether opinion surveys truly capture what the public want in the face of information efects. Indeed, if people appeal to facts in making sense of their political commitments, there is something slightly cynical about policymakers holding people to the commitments they might be embracing for no other reason than they are working on the basis of bad or incomplete information—especially when expecting citizens to seek out information on all political topics relevant to them is not practically possible. Tis, afer all, is the whole point of representative democracy ofering a cognitive division of labor. Te presence of information efects suggests that this division of labor is not delivering the goods—people are in efect tasked with making judgments on matters without having sufcient information to make an informed decision. As a result, our aggregate voice does not always capture the “will of the people,” and policymakers should not assume otherwise.16

16  Research for this chapter was made possible through a grant from the Leverhulme Trust, “Evidence and Perceptions in the Post-­Brexit Immigration Debate” (RPG-­2017-­127), conducted by the authors together with Heather Rolfe, Nathan Hudson-­Sharp, and Johnny Runge at the National Institute of Economic and Social Research (NIESR) in London. Te authors are grateful to Rolfe, Hudson-­Sharp, and Runge for valuable comments on a previous draf of this chapter, as well as to the participants at the Epistemology, Democracy, and Disagreement workshop at Georgetown University’s Ethics Lab, where an early version of the chapter was presented. Te particular interpretations of the data ofered in this chapter belong to the authors, and do not necessarily refect the views of the larger research team, NIESR, or the Leverhulme Trust.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  195 APPENDIX

1. Dataset Te dataset used for the counterfactual models in Section 4 is from the November 2019 wave of the 2014–2023 British Election Study Internet Panel (Fieldhouse et al., 2020), featuring responses from 34,366 UK respondents. Four percent of values were missing across the variables used (see Table 10.1). Tese were imputed with multiple imputation, using aregImpute in R’s Hmisc package (R Core Team,  2017; Harrell et al., 2019). Coding of individual variables as well as sample sizes for the diferent levels of factor coded variables (afer imputation) are given in Table 10.1.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Knowledge scale Te six knowledge items available in the dataset were used to ft a two-­parameter item response theory (IRT) model in R, using ltm (Rizopoulos, 2006), to estimate the latent level of knowledge of respondents. Four of these (see fn. 11 of the main body of the text) had both good individual discrimination values (1.36, 2.82, 2,28, and 2.61, respectively), and together an acceptable level of internal consistency (std. alpha = 0.7). Te test information function was normally distributed, with a peak of about 5 slightly below an ability value of 0, suggesting that the model discriminated best at around the mean level of knowledge. Te guessing parameters of a corresponding three-­parameter model came out very low, so the simpler, two-­parameter model was used. A valid IRT model should meet three conditions: unidimensionality (the scale taps into only one dimension/trait), local independence (items are uncorrelated afer conditioning on the measured trait), and good model ft. Tests found no evidence of a failure of unidimensionality (via ltm’s unidimTest; p-value = 0.5347). As for local independence, the maximum Q3 value between items was 0.298. Yen (1993) suggests a screening value of +/−0.2. However, as pointed out by de Ayala (2009, p. 137), that screening value is most appropriate for long scales (35 items or more), as shorter scales can be expected to give higher values. In light of this, a maximum value of 0.298 for a four-­item scale seems acceptable. Model ft was evaluated through a plot of observed versus expected values, which suggested a good ft. Finally, the model was used to calculate a knowledge score for each respondent, generating a range from the minimum to the maximum measured level of knowledge of −1.472 to 0.5436, with a mean of 0. For purposes of calculating propensity scores (see next section), the knowledge variable was then recoded as a binary variable, with all observations representing someone with a knowledge score above the mean coded as 1 (16,299 participants), and everyone else as 0 (18,067 participants), thereby separating those who are for our purposes considered “informed” from those who are not. Defning “informed” with reference to the mean level of knowledge serves to pre-­empt the worry that the level we are working with is inappropriately high. Indeed, in the dataset used, only 6 percent of respondents were fully

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

196  Kristoffer Ahlstrom-Vij and Jennifer R. Steele

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Table 10.1  Variable coding and sample sizes Variable

Sample size

Reduce immigration levels 1 (agree) 0 (disagree) Gender Male Female Education No qualifcation Below GCSE GCSE A-­level Undergraduate Postgraduate Household income (divided by UK quintiles) 1st quintile (up to £15,000 annually) 2nd quintile (up to £20,000 annually) 3rd quintile (up to £25,000 annually) 4th quintile (up to £35,000 annually) 5th quintile (from £35,000 annually) Religion Church of England Catholic Other Christian religion Other religion No religion Ethnicity White British Other white ethnicity Mixed Asian Other ethnicity Marital status Civil partnership Divorced Living as married Married Never married (single) Separated Widowed Age 18–25 26–35 36–45 46–55 56–65 Over 65 Party identity Brexit Party Conservatives Green Party

  17776 16590

 

15341 19025

 

2550 1620 7763 7070 11783 3580

 

6359 3426 3695 6560 14326

 

8911 2466 2736 2033 18220

  31259 1275 433 911 488   292 2535 4239 17041 8163 602 1494   2202 4384 5588 6215 6962 9015   1272 10499 1165

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  197 Labor Liberal Democrats Plaid Cymru Scottish National Party UKIP Other party No party EU referendum vote Leave Remain Did not vote

8826 2831 138 921 256 280 8178

 

14986 17661 1719

informed in the strict sense of achieving the maximum knowledge score, as given by the IRT model.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Propensity scores As noted in the main body of the text (see Section 4.2), the counterfactual models were constructed using “doubly robust” estimation for causal inference (e.g. Morgan and Winship,  2015). Te “double robustness” is due to how casual efects are estimated in a context where we have both controlled for (assumed) confounds (see Section 4.1), and improved the balance between the two groups (here: the “informed” and everyone else) to make up for the fact that they have not come about as a result of randomized assignment. In the present case, this second layer of “robustness” was achieved through so-­called “propensity scores.” To illustrate how such scores work, consider the demographic model (see Section 4.1). Te propensity scores for the demographic model measure the probability (or propensity) that an observation will be found in the “informed” category (i.e. taking on a value of 1 on the binary knowledge variable), as a function of someone’s demographic features. Te idea is to then use these scores to remove any correlation between these features and the “informed” category, to justify a causal inference. Why? It is helpful to think of this on the model of a randomized design, where the random allocation of participants to a treatment and a control group means that no feature of the participant is predictive of being found in the treatment as opposed to in the control. Whether female or male, rich or poor (etc.), you are equally likely to end up in one group as opposed to in the other, provided assignment is truly random. In the case of observational data, by contrast, this might not be the  case. In the case at hand, it might (for example) be that some features of the observations—for example their level of education, their income, or what have you— are predictive of ending up in the “informed” category. One way to counteract this is through “matching,” where each “treated” observation is matched with a control observation that is identical along all covariates. While conceptually straightforward, the problem is that, with several covariates and fnite samples, exact matches might not be available for a sufciently large number of observations. Propensity score weighting ofers an alternative approach. Since

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

198  Kristoffer Ahlstrom-Vij and Jennifer R. Steele propensity scores measure the probability of ending up in the “treatment” category, given a set of covariates—in our case, the probability that you would be “informed,” given your age, level of income, level of education, and so forth for all measured covariates—we can use the inverse of those scores as weights (such that an observation with a low propensity is weighted heavily) in ftting the model. Tis improves the balance between the two groups, and—given an appropriately chosen set of covariates when calculating the scores—recreates a situation that would have been expected in a randomized experiment, thereby allowing greater confdence in inferring a causal relationship. Prior to calculating propensity scores, balance was evaluated using balance plots from the cobalt package (Greifer, 2020), which suggested imbalance across all covariates. Particularly noteworthy in this context is that the two groups (i.e. those above the mean knowledge score and everyone else) were imbalanced for gender (men had a greater proportion of informed participants than women), education (informed participants were overrepresented among the educated, and underrepresented among the less educated), income (higher incomes had a greater proportion of informed participants than lower), and age (older age groups had a greater proportion of informed participants than lower). Tis of course serves to highlight the need for propensity score weighting. Against this background, three propensity scores were therefore calculated for each observation—one score for each of the three models, measuring the probability that an observation would be found in the “informed” category on the basis of (i) its demographic features (for the demographic model), (ii) its demographic features and partisanship (for the partisanship model), and (iii) its demographic features, partisanship, and EU referendum vote (on the EU referendum identity model), respectively. Te propensity scores were calculated by way of logistic regression, using glm in R’s stats package. Te scores were evaluated by confrming that applying the scores as weights substantially improved the balance between groups across all covariates. By way of illustration, consider Figure  10.1, where the lef-­hand panel shows the balance (or rather: lack thereof) for the income variable prior to applying the propensity weights—as mentioned above, informed participants (teal bars, designated here as 1 or “treated”) are overrepresented among the wealthy (higher quintiles), and underrepresented among the less wealthy (lower quintiles)—and the right-­hand panel the balance achieved once the weights had been applied:

4.  Counterfactual models Using the propensity scores as weights, three logistic regression models were then ftted using glm, with the model specifcations discussed in the main body of the text (see Section 4.1). Remember, our dependent variable was a binary variable taking on a value of 1 for respondents indicating that they would like to see the number of immigrants coming in to the country reduced, and 0 otherwise. Details and diagnostics for the models are provided in Table 10.2. Model ft was evaluated by way of McFadden values. McFadden (1979) suggests a value in the range of 0.2–0.4 indicates an excellent ft. In light of that, the values in

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  199 Distributional Balance for “income”

Distributional Balance for “income”

Unadjusted Sample

Adjusted Sample 0.4

0.4

0.3 Proportion

Proportion

0.3

0.2

0.1

0.1

0.0

0.2

1st_quintile 2nd_quintile 3rd_quintile 4th_quintile 5th_quintile

0.0

income

income

Treatment 0

1st_quintile 2nd_quintile 3rd_quintile 4th_quintile 5th_quintile

Treatment 1

0

1

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Figure 10.1  Balance plots for income quintiles, before and afer applying propensity weights

Table 10.2 look acceptable. Te “Highest VIF” value gives the highest value across all covariates of a variance infation factor test for multicollinearity. Values substantially higher than 1 suggest potential multicollinearity, meaning there is no evidence of multicollinearity here. In terms of infuential observations, the number of observations with a Cook’s distance value greater than 1 (there were none) or a standardized residual greater than 3 are given above. Te models were re-­ftted without these observations, to see what diference that made for the knowledge coefcient. It did not make a substantial diference: without those observations, the coefcient came out to −0.5488 on the EU Referendum ID model, −0.5626 on the Partisanship model, and −0. 5686 on the Purely Demographic model.

References Achen, C. H. and L. Bartels. 2016. Democracy for Realists: Why Elections Do Not Produce Responsive Government. Princeton, NJ: Princeton University Press. Ahlstrom-Vij, K. 2019. “Te Epistemic Benefts of Democracy: A Critical Assessment.” In M. Fricker, P. Graham, D. Henderson, and N. Pedersen (eds.), Te Routledge Handbook of Social Epistemology. Abingdon: Routledge. Ahlstrom-Vij, K. 2020. “Te Case for Modelled Democracy.” Episteme 1–22; https://doi. org/10.1017/epi.2020.10. Althaus, S. 2003. Collective Preferences in Democratic Politics. Cambridge: Cambridge University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Intercept 1.4148 0.0408 0.0000 Knowledge (reference category = Knowledge score at or below mean) Knowledge score above mean −0.5471 0.0189 0.0000 Gender (reference = Male) Female −0.1724 0.0193 0.0000 Education (reference category = A-­level) Below GCSE 0.2886 0.0494 0.0000 GCSE 0.2932 0.0289 0.0000 No qualifcation 0.5524 0.0435 0.0000 Postgraduate −0.6964 0.0366 0.0000 Undergraduate −0.2546 0.0259 0.0000 Income (reference category = 1st quintile) 2nd quintile 0.0786 0.0373 0.0352 3rd quintile −0.0938 0.0369 0.0111 4rd quintile −0.0673 0.0323 0.0372 5th quintile −0.2352 0.0304 0.0000 Religion (reference category = No religion) Church of England 0.3281 0.0238 0.0000 Catholic 0.2299 0.0371 0.0000 Other Christian religion 0.1317 0.0358 0.0002 Other religion 0.0871 0.0437 0.0461 Ethnicity (reference category = White British) Asian −0.2121 0.0617 0.0006 Other ethnicity −0.6906 0.0868 0.0000 Mixed −0.6387 0.0881 0.0000 Other white ethnicity −0.4819 0.0525 0.0000 Marital status (reference category = Never married) Civil partnership 0.4917 0.1082 0.0000 Divorced 0.2117 0.0421 0.0000

0.0355 0.0175 0.0180 0.0459 0.0268 0.0404 0.0339 0.0241 0.0345 0.0343 0.0300 0.0283 0.0221 0.0345 0.0334 0.0410 0.0589 0.0814 0.0824 0.0502 0.1023 0.0391

0.5043 −0.5614 −0.2203 0.3831 0.3693 0.6380 −0.7915 −0.3659 0.0284 −0.1196 −0.1327 −0.3712 0.3738 0.2646 0.1571 0.0943 −0.3860 −0.7467 −0.6103 −0.6793 0.4621 0.2172

Std. Error

Estimate

p-value

Estimate

Std. Error

Partisanship model

EU referendum ID model

Table 10.2  Model details and diagnostics for causal models

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

0.0000 0.0000

0.0000 0.0000 0.0000 0.0000

0.0000 0.0000 0.0000 0.0216

0.4095 0.0005 0.0000 0.0000

0.0000 0.0000 0.0000 0.0000 0.0000

0.0000

0.0000

0.0000

p-value

0.4086 0.1886

−0.3697 −0.8725 −0.6475 −0.7564

0.5769 0.2826 0.1678 0.0720

0.0000 −0.0844 −0.0940 −0.2617

0.4623 0.4156 0.6759 −0.8936 −0.4175

−0.2636

−0.5710

0.3103

Estimate

0.0936 0.0370

0.0562 0.0779 0.0790 0.0480

0.0207 0.0327 0.0315 0.0389

0.0327 0.0324 0.0283 0.0265

0.0435 0.0254 0.0381 0.0320 0.0227

0.0170

0.0166

0.0312

Std. Error

Purely demographic model

0.0000 0.0000

0.0000 0.0000 0.0000 0.0000

0.0000 0.0000 0.0000 0.0640

0.9989 0.0091 0.0009 0.0000

0.0000 0.0000 0.0000 0.0000 0.0000

0.0000

0.0000

0.0000

p-value

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

McFadden Highest VIF Cook’s > 1 Std. residual > 3

0.2717 1.0711 0 853

Living as married 0.2298 Married 0.1407 Separated 0.1499 Widowed 0.2186 Age (ordinal with six categories) Age (linear) 0.2184 Age (quadratic) −0.2625 Age (cubic) −0.0722 Age^4 −0.0594 Age^5 0.0830 Partisanship (reference category = No party) Brexit Party 1.0295 Conservatives 0.2896 Green Party −0.6817 Labor −0.3961 Liberal Democrats −0.4876 Other party −0.4886 Plaid Cymru −0.8803 SNP −0.5615 UKIP 0.8131 EU referendum vote (reference category = Leave) Did not vote −0.6018 Remain −1.8684

0.0000 0.0000 0.0473 0.0000 0.0000 0.0000 0.0056 0.0139 0.0003 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

0.0347 0.0279 0.0756 0.0538 0.0349 0.0284 0.0261 0.0242 0.0228 0.0717 0.0265 0.0581 0.0266 0.0387 0.1021 0.1738 0.0608 0.1425 0.0433 0.0216

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

0.1856 1.0613 0 830

— —

1.8468 0.6237 −1.1660 −0.7849 −1.0969 −0.3761 −1.2928 −1.1611 1.4559

0.4931 −0.2698 −0.0953 −0.0567 0.0903

0.2523 0.1975 0.0627 0.2321

— —

0.0696 0.0241 0.0543 0.0239 0.0363 0.0919 0.1590 0.0574 0.1385

0.0324 0.0265 0.0243 0.0225 0.0212

0.0323 0.0260 0.0700 0.0497

— —

0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000

0.0000 0.0000 0.0001 0.0119 0.0000

0.0000 0.0000 0.3703 0.0000

0.1122 1.0568 0 890

— —

— — — — — — — — —

0.7382 −0.2395 −0.0593 −0.0473 0.1089

0.1912 0.1767 0.0683 0.2060

— —

— — — — — — — — —

0.0305 0.0252 0.0231 0.0214 0.0201

0.0306 0.0247 0.0663 0.0469

— —

— — — — — — — — —

0.0000 0.0000 0.0103 0.0271 0.0000

0.0000 0.0000 0.3032 0.0000

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

202  Kristoffer Ahlstrom-Vij and Jennifer R. Steele Bell, B. and B. Machin. 2013. “Immigration and Crime: Evidence for the UK and Other Countries,” Migration Observatory Briefng, COMPAS, University of Oxford, UK. Bhatti, Y. 2010. “What Would Happen if We Were Better Informed? Simulating Increased Knowledge in European Parliament (EP) Elections?” Representation 46(4): 391–410. Blais, A., E.  Gidengil, P.  Fournier, and N.  Nevitte. 2008. “Information, Visibility and Elections: Why Electoral Outcomes Difer When Voters are Better Informed.” European Journal of Political Research 48: 256–80. Blinder, S. 2015. “Imagined Immigration: Te Impact of Diferent Meanings of ‘Immigrants’ in Public Opinion and Policy Debates in Britain.” Political Studies 63: 80–100. Blinder, S. and W.  L.  Allen. 2016. “Constructing Immigrants: Portrayals of Migrant Groups in British National Newspapers, 2010–2012.” International Migration Review 50(1): 3–40. Braakman, N., M.  Waqas, and J.  Wildman. 2017. “Are Immigrants in Favour of Immigration? Evidence from England and Wales.” B.E.  Journal of Economic Analysis and Policy 17(1). Brader, T. and J.  A.  Tucker. 2018. “Unrefective Partisans? Policy Information and Evaluation in the Development of Partisanship.” Advances in Political Psychology 39, Suppl. 1: 137–57. Brandt, R. 1979. A Teory of the Right and the Good. Oxford: Clarendon Press. Brennan, J. 2011. “Te Right to a Competent Electorate.” Philosophical Quarterly 61(245): 700–24. Campbell, A, P. E. Converse, W. E. Miller, and D. E. Stokes. 1960. Te American Voter. New York: John Wiley & Sons. Caplan, B. 2007. Te Myth of the Rational Voter. Princeton, NJ: Princeton University Press. Card, D., C. Dustmann, and I. Preston. 2005. “Understanding Attitudes to Immigration: Te Migration and Minority Module of the First European Social Survey,” Discussion Paper Series, CDP No 03/05. Centre for Research and Analysis of Migration. Dawson, M.  C. 1994. Behind the Mule: Race and Class in African American Politics. Princeton, NJ: Princeton University Press. de Ayala, R.  J. 2009. Te Teory and Practice of Item Response Teory. New York: Guilford Press. Deaux, K., V. M. Esses, R. Lalonde, and R. Brown (eds.). 2010. “Immigrants and Hosts: Perceptions, Interactions, and Transformations.” Journal of Social Issues 66(4). Delli Carpini, M. X. and S. Keeter. 1993. “Measuring Political Knowledge: Putting First Tings First.” American Journal of Political Science 37(4): 1179–206. Delli Carpini, M. X. and S. Keeter. 1996. What Americans Know about Politics and Why it Matters. New Haven, CT: Yale University Press. Denver, D. 2008. “Another Reason to Support Marriage? Turnout and the Decline of Marriage in Britain.” British Journal of Politics and International Relations 10(4): 666–80. Dovidio, J. F. and S. L. Gaertner. 2004. “Aversive Racism.” In M. P. Zanna (ed.), Advances in Experimental Social Psychology, Vol. 36. Elsevier Academic Press, pp. 1–52. https:// doi.org/10.1016/S0065-2601(04)36001-6 Dufy, B. 2018. Te Perils of Perception: Why We’re Wrong about Nearly Everything. London: Atlantic Books. Dufy, B. and J. Stannard. 2017. Perils of Perceptions 2017. Ipsos MORI.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Policy, Ignorance, and the Will of the People  203 Dufy, B., K.  Kaur-Ballagan, G.  Gottfried, and A.  Palmqivist Aslaksen. 2017., Shifing Ground: 8 Key Findings from a Longitudinal Study on Attitudes Towards Immigration and Brexit. Ipsos MORI. Dustmann, C. and I.  P.  Preston. 2007. “Racial and Economic Factors in Attitudes to Immigration.” B.E. Journal of Economic Analysis and Policy 7(1). Esses, V. M., P. M. Brochu, and K. R. Dickson. 2011. “Economic Costs, Economic Benefts, and Attitudes Toward Immigrants and Immigration.” Analyses of Social Issues and Public Policy 11: 1–5. Evans, G. and K.  Northmore-Ball. 2020. “Long-Term Factors: Class and Religious Cleavages.” in Fisher et al. (eds.), Te Routledge Handbook of Elections, Voting Behavior and Public Opinion. Abingdon: Routledge, pp. 123–35. Fieldhouse, E., J. Green, G. Evans, J. Mellon, and C. Prosser. 2020. British Election Study Internet Panel Wave 17; available at https://www.britishelectionstudy.com/. Fisher, J., E. Fieldhouse, M. N. Franklin, R. Gibson, M. Cantijoch, and C. Wlezen. 2020. Te Routledge Handbook of Elections, Voting Behavior and Public Opinion. Abingdon: Routledge. Ford, R. and M. Goodwin. 2014. Revolt on the Right: Explaining Support for the Radical Right in Britain. London and New York: Routledge. Ford, R. and A.  Heath. 2014. “Immigration: A Nation Divided?” In British Social Attitudes: the 31st Report. London: Te National Centre for Social Research. Friedman, J. 1998. “Introduction: Public Ignorance and Democratic Teory.” Critical Review 12(4): 397–411. Gaertner, S. L. and J. F. Dovidio. 1986. “Te Aversive Form of Racism.” In J. F. Dovidio and S.  L.  Gaertner (eds.), Prejudice, Discrimination, and Racism. Academic Press, pp. 61–89. Gaertner, S. L. and J. F. Dovidio. 2000. Reducing Intergroup Bias: Te Common Ingroup Identity Model. New York: Psychology Press. Goodin, R.  E. 1995. Utilitarianism as a Public Philosophy. Cambridge: Cambridge University Press. Green, D. P., B. Palmquist, and E. Schickler. 2002. Partisan Hearts and Minds: Political Parties and the Social Identities of Voters. New Haven, CT: Yale University Press. Greifer, N. 2020. “cobalt: Covariate Balance Tables and Plots.” R package version 4.0.0; https://CRAN.R-project.org/package=cobalt. Grigorief, A., C. Roth, and D. Ubfal. 2016. “Does Information Change Attitudes Towards Immigrants? Representative Evidence from Survey Experiments,” Discussion Paper Series, IZA DP No. 10419, Institute of Labour Economics. Hainmueller, J. and D.  Hopkins. 2014. “Public Attitudes Toward Immigration.” Annual Review of Political Science 17: 225–49. Hansen, K. P. 2009. “Changing Patterns in the Impact of Information on Party Choice in a Multiparty System.” International Journal of Public Opinion Research 21(4): 525–46. Harrell, F.  E., C.  Dupont, and collaborators. 2019. “Hmisc: Harrell Miscellaneous.” R package version 4.2-0; https://CRAN.R-project.org/package=Hmisc. Harsanyi, J.  C. 1997. “Utilities, Preferences, and Substantive Goods.” Social Choice and Welfare 14: 129–45. Hebbelstrup, S. and R. Rasmussen. 2016. “Education or Personality Traits and Intelligence as Determinants of Political Knowledge?” Political Studies 64(4): 1036–54.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

204  Kristoffer Ahlstrom-Vij and Jennifer R. Steele Hobolt. S. B., T. J. Leeper, and J. Tilley. 2020. “Divided by the Vote: Afective Polarization in the Wake of the Brexit Referendum.” British Journal of Political Science doi. org/10.1017/S0007123420000125. Hopkins, D.  J., J.  Sides, and J.  Citrin. 2018. “Te Muted Consequences of Correct Information About Immigration; available at: https://ssrn.com/abstract=2798622. Hornikx, J. 2005. “A Review of Experimental Research on the Relative Persuasiveness of Anecdotal, Statistical, Causal, and Expert Evidence.” Studies in Communication Sciences 5(1): 205–16. Hume, D. 2003. A Treatise of Human Nature. Oxford University Press; originally published in 1740. Ipsos MORI. 2018. Attitudes to Immigration. Ipsos MORI. Available at: https://www.ipsos. com/sites/default/fles/ct/news/documents/2018-03/british-attitudes-to-immigrationmarch-2018_1.pdf. Kahneman, D. 2011. Tinking, Fast and Slow. London: Penguin. Katwala, S., S. Ballinger, and M. Rhodes. 2014. How to Talk about Immigration. London: British Future. Keele, L. 2015. “Te Statistics of Causal Inference: A View from Political Methodology.” Political Analysis 23: 313–35. Lawrence, E. D. and J. Sides. 2014. “Te Consequences of Political Innumeracy.” Research & Politics 1(2): 1–8. Lupia, A. 2006. “How Elitism Undermines the Study of Voter Competence.” Critical Review 18(1–3): 217–32. https://doi.org/10.1080/08913810608443658. Migration Advisory Committee 2018. EEA migration in the UK: Final report; available at: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/ attachment_data/fle/741926/Final_EEA_report.PDF. Morgan, S. and C.  Winship. 2015. Counterfactuals and Causal Inference: Methods and Principles for Social Research, Second edition, Cambridge: Cambridge University Press. Oscarsson, H. 2007. “A Matter of Fact? Knowledge Efects on the Vote in Swedish General Elections, 1985–2002.” Scandinavian Political Studies 30(3): 301–22. Oxford Economics 2018. “Te Fiscal Impact of Immigration on the UK: A Report for Te Migration Advisory Committee.” London: Oxford Economics. Pearl, J. 2000. Causality: Models, Reasoning, and Inference. Cambridge: Cambridge University Press. Plutzer, E. 2020. “Demographics and the Social Bases of Voter Turnout.” In Fisher et al. (eds.), Te Routledge Handbook of Elections, Voting Behavior and Public Opinion. Abingdon: Routledge, pp. 69–82. R Core Team 2017. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. Railton, P. 1986. “Facts and Values.” Philosophical Topics 14(2): 5–31. Rizopoulos, D. 2006. “ltm: An R package for Latent Variable Modelling and Item Response Teory Analyses.” Journal of Statistical Sofware 17(5): 1–25. Rohrer, J. 2018. “Tinking Clearly about Correlations and Causation: Graphical Causal Models for Observational Data.” Advances in Methods and Practices in Psychological Science 1(1): 27–42.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Policy, Ignorance, and the Will of the People  205

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Rolfe, H., K. Ahlstrom-Vij, N. Hudson-Sharp, and J. Runge. 2018. Post-Brexit Immigration Policy: Reconciling Public Perceptions with Economic Evidence. London: National Institute of Economic and Social Research. Rutter, J. and R. Carter. 2018. National Conversation on Immigration: An interim report to the Home Afairs Committee. London: British Future and Hope not Hate. Sides, J. and J. Citrin. 2007. “European Opinion about Immigration: Te Role of Identities, Interests and Information.” British Journal of Political Science 37: 477–504. Somin, I. 2016. Democracy and Political Ignorance: Why Smaller Government Is Smarter. Second Edition. Stanford, CA: Stanford University Press. vanHeerde-Hudson, J. 2020. “Political Knowledge: Measurement, Misinformation and Turnout.” In Fisher et al. (eds.), Te Routledge Handbook of Elections, Voting Behavior and Public Opinion. Abingdon: Routledge, pp. 369–82. Vowles, J. 2020. “Te Big Picture: Turnout at the Macro-Level.” in Fisher et al. (eds.), Te Routledge Handbook of Elections, Voting Behavior and Public Opinion. Abingdon: Routledge, pp. 57–68. Yen, W. M. 1993. “Scaling Performance Assessments: Strategies for Managing Local Item Dependence.” Journal of Educational Measurement 30: 187–213.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved. Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

PART III

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

DISAGR E E ME NT A ND POL A R IZ AT ION

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved. Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

11

Problems of Polarization

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Robert B. Talisse

Narratives about the danger posed by polarization saturate contemporary political commentary. What is not clear in the popular discourse, though, is what is meant by polarization beyond the general condition of intransigence among politicians. Of course, stubbornness among politicians is nothing new. And there is a formidable thread running throughout democratic theory according to which logjams and stalemates are a sign of political health. Yet popular commentary is animated by the suspicion that at present deeper and more insidious forces are driving these phenomena. One gets the sense that polarization is used as a blanket term to gesture toward unnamed underlying causes, whatever they may be. For those of us who share the suspicion that our politics is exhibiting new and troubling forms of dysfunction, it would be helpful to develop a more precise account of what polarization is and why it is problematic. Tis is the central task of the present chapter. Afer distinguishing political polarization from belief polarization, I will argue that the intertwining of these two kinds of polarization lies at the root of current political dysfunctions. In addition, the argument will show that belief polarization is best understood as a social-­political-­epistemic dynamic that results not only in the deadlock and distrust that characterizes political polarization, but also a mounting incapacity to enact democratic citizenship. Hence it will be argued that whereas polarization presents several problems for democratic politics, the overarching difculty is that it is politically degenerative—it erodes our democratic capacities. Accordingly, the aims of this chapter are strictly diagnostic; an ameliorative proposal is developed elsewhere.1

1.  Two Concepts of Polarization To begin, two broad species of polarization should be distinguished: political polarization and belief polarization. Political polarization denotes a family of phenomena having to do with what might be called the political distance between political opponents and the consequent dissolution of common ground between them. Tere are several ways in which this distance can be conceived. Here we will consider three. Political polarization is ofen understood as the ideological distance between the platforms of competing political parties. Where platform polarization is acute, 1  See Talisse 2019. Robert B. Talisse, Problems of Polarization In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Robert B. Talisse. DOI: 10.1093/oso/9780192893338.003.0012

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

210  Robert B. Talisse competing political parties diverge sharply on nearly every issue. Hence the political middle ground between them dissolves, leaving no basis for cooperation or compromise. Platform political polarization is diferent from partisan political polarization. Here, polarization is conceptualized as partisan ideological uniformity, what might be aptly characterized as ideological purity among partisans. An especially high level of partisan political polarization is marked by the absence of moderates within partisan groups, and the gradual weeding out of, say, “conservative Democrats” and “liberal Republicans.” As with platform polarization, partisan polarization results in the receding of the middle ground among opposed parties, thus triggering deadlock. On a third view, polarization is an afective distance between political opponents. Afective political polarization is marked by high levels within a partisan group of distrust and antipathy towards the members of opposing groups. Note that afective political polarization may obtain among persons who are not especially at odds over particular policy issues. What’s more, afective polarization might obtain in the absence of signifcant platform or partisan polarization. Nevertheless, afective polarization results in a breakdown of communication and compromise across partisan divisions, and so, like the other forms of political polarization, it leads to political deadlock. Political polarization is to be contrasted with belief polarization.2 Belief polarization besets individuals who talk only or mainly to others who share their fundamental commitments. To be more specifc, when individuals talk with mainly likeminded others about the matters upon which they agree, each discussant tends gradually to embrace a more extreme version of his or her initial opinion. In short, belief polarization is that regularity by which discussion with likeminded others turns us into more extreme versions of ourselves. Te phenomenon gives us a lot to unpack, so we must begin by taking a closer look.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Belief Polarization: Te Basic Phenomenon Belief polarization has been found to be operative within groups of an impressive variety of kinds, from ofcially defned assemblies—juries, panels of judges, Boards of Trust, and investment groups—to informal clusters of ordinary people simply talking about views they share.3 Furthermore, belief polarization does not discriminate between the diferent kinds of belief that group members may hold in common. Likeminded groups polarize regardless of whether they are discussing banal matters of fact such as the elevation of a given city, personal matters of taste (say, assessments of the attractiveness of a face or comfortableness of a chair) or questions about value (Baron et al., 1996; Sunstein, 2009, pp. 18–19). What is more, the phenomenon operates regardless of the explicit point of the group’s discussion. Likeminded groups 2  What I’m calling belief polarization is more commonly known as group polarization. I depart from the standard nomenclature because in the context of the distinction I am employing, the more usual name invites confusion; both political and belief polarization have to do with groups, albeit each in its distinctive way. Nonetheless, as will be noted later, the term belief polarization invites troubles, too. 3  Lamm and Myers (1978, p. 146) write, “Seldom in the history of social psychology has a nonobvious phenomenon been so frmly grounded in data from across a variety of cultures and dependent measures.” See the Appendix in Sunstein, 2009 for summaries of the experimental fndings.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Problems of Polarization  211 polarize when they are talking for the sake of deciding an action that the group will take (e.g. participating in a protest, rendering a verdict, placing a bet); and they polarize also when there is no specifc decision to be reached and no collective action to be taken. Finally, the phenomenon has been studied throughout the world, and is found to be prevalent in likeminded groups regardless of the demographic profles of their members; we are vulnerable to belief polarization regardless of nationality, race, gender, religion, economic status, and level of education. Now turn to the phenomenon itself. One of the earliest experiments regarding group polarization was conducted in France in the late 1960s with a group of teenage men. Te teenagers were asked to discuss both their (variously positive) views about de Gaulle and their (variously negative) views about American foreign policy. Te individuals emerged with more thoroughly positive views of de Gaulle and more intensely negative views of American foreign policy than those they held prior to the conversation (Moscovici and Zavalloni, 1969). In another early study, Michigan high-­schoolers were sorted according to their antecedently expressed level of racial prejudice. Te likeminded groups were then tasked with discussing several issues concerning race in the U.S., including the question of whether racism is the cause of the socioeconomic disadvantages faced by African Americans. Following the conversations with their respective groups of likeminded others, those who antecedently showed a high level of racial prejudice came to embrace more ardently the view that racism is not responsible for the disadvantages faced by African Americans, while those antecedently disposed toward low levels of racial prejudice grew more accepting of the view that racism is the cause of such disadvantages. Once again, discussion among likeminded people amplifed the members’ pre-­discussion tendencies. Accordingly, the ideological distance between the two groups also expanded (Myers and Bishop, 1970, pp. 778–9). A similar experiment involved adults who on the basis of an initial screening were classifed into gender-­mixed groups according to their views concerning the social roles of women. Once sorted into “feminist” and “chauvinist” groups, each discussed amongst themselves the merits of various statements about the role of women in society—statements like “a woman should be as free as a man to propose marriage,” and “women with children should not work outside the home if they don’t have to fnancially.” Te result was that members of the feminist discussion group became more profeminist, while the chauvinist group became more chauvinist—though not to a signifcant degree (Myers, 1975). In 2005, a collection of Coloradoans were sorted according to an initial screening test into “liberal” and “conservative” groups. Each group was then then asked to discuss the following three policy questions: 1. Should sates allow same-­sex couples to enter into civil unions? 2. Should employers engage in “afrmative action” by giving a preference to members of traditionally disadvantaged groups? 3. Should the U.S. sign an international treaty to combat global warming? Te pattern of belief polarization was observed. Afer discussion within likeminded groups, liberal participants, who were antecedently disposed to favor a global

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

212  Robert B. Talisse warming treaty, came to endorse more enthusiastically the proposition that the U.S. should enter into such a treaty. Conservatives who were initially neutral on the idea of such a treaty came to ardently oppose it afer discussion with fellow conservatives. Similarly, attitudes towards same-­sex civil unions and afrmative action belief-­ polarized following group discussion: liberal support intensifed, while opposition among conservatives grew more resolute. Importantly, the shif in group members’ attitudes also resulted in a greater degree of internal homogeneity. Tat is, afer discussion, not only did the members of each group tend to shif to a view more extreme than the one he or she held prior to the discussion, attitudes within the group became more uniform. Each shifed into a more extreme version of his or her prior self and also became more like the others in the group. Tus we see one way in which belief polarization can contribute to partisan political polarization (Hastie et al.,  2007; Sunstein, 2009, pp. 5–8). Tus far, I have described experiments demonstrating that discussion in likeminded groups produces shifs in opinion toward extremity in the direction of the individual’s pre-­discussion inclination. Tese shifs have been described mainly as intensifcations (again, towards extremity in the direction of individuals’ initial inclination) of attitude or view. As one would expect, such shifs also afect behavior. Researchers have appealed to belief polarization in explaining high-­ profle instances of group behavior gone wrong. To cite a few examples, the phenomenon has been invoked to explain the Johnson administration’s decision to escalate the Vietnam War, the risk-­taking at NASA that arguably led to the Challenger explosion, and certain failing trends in fnancial investment (Sia et al., 2002: 71–2). But we need not rely on speculations of this kind; several experiments show that the shifs produced by belief polarization have an efect on individuals’ and groups’ practical deliberations, and thus their behavior.4 Consider two cases where this practical ramifcation of belief polarization is evident. In mock jury experiments involving punitive damage awards, when jury members are initially agreed that the harm is severe and damages should be awarded, their deliberation produces a verdict of a signifcantly larger award than any individual juror’s initial pre-­deliberation assessment. Te same goes for juries whose members are initially inclined to think that the harm in question is not particularly extreme and only a low degree of punishment is in order. Afer deliberation, the verdict is more lenient than individual jurors’ initial inclination (Schkade et al., 2000). Another study fnds that group discussion of an event that participants agree constitutes a serious violation of justice—sex discrimination committed by an elected ofcial, for example—leads to a greater inclination among the discussants to engage in organized protest. Crucially, that’s not all. Among discussants who see the violation as especially egregious, the enhanced readiness to protest is accompanied by a willingness that exceeds their pre-­discussion inclination to protest in ways that are overtly militant and thus riskier (Johnson et al., 1977). 4  In fact, the phenomenon known as “risky shif,” where a group’s post-­discussion likeliness to engage in risky collective behavior exceeds any of its individual member’s pre-­discussion willingness to endorse that level of risk for the group, is frequently understood to be a special case of belief polarization. See Isenberg, 1986, p. 1141.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Problems of Polarization  213 In summary, discussion in likeminded groups produces belief polarization.5 Belief polarization involves a shif towards greater extremity in the opinions of those subjected to the phenomenon. As one would expect, those shifs tend to manifest in behaviors that are more extreme than those that individuals are otherwise inclined to enact. To repeat, discussion with likeminded others turns us into more extreme versions of ourselves. Importantly, this extremity consists in shifs in our attitudes and beliefs as well as in our actions. Te pervasiveness of the belief polarization phenomenon and the vastness of the documentation amplify the need for explanation of the mechanisms that produce belief polarization. In order to begin to address these issues, we must attend to a few philosophical issues that are already within view.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Degree, Content, and Commitment Te claim that some individual has shifed to a more extreme version of her initial opinion is ambiguous on multiple fronts. For starters, it should be acknowledged that the shifs that belief polarization produces ofen occur unconsciously, or at least non-­ intentionally. What is more, belief polarization is not a process by which an individual explicitly reasons her way into a more extreme version of herself. By and large, belief polarization happens to an individual, and ofen unwittingly. One might also wonder what is meant by extreme in this context. It should be clear from what has been said that extremity is understood in a way that makes no reference to an external spectrum of opinion. One becomes more extreme in the course of likeminded discussion simply in virtue of moving further in the direction of the pole suggested by one’s pre-­discussion inclination. Accordingly, the shif from fnding a particular face “somewhat attractive” to fnding it “extremely attractive” is a shif in the direction of extremity as that concept is understood here. Belief polarization consequently need not render one an extremist in the political sense of that term (though it might). For similar reasons it is clear that the concept of extremity being employed invokes no epistemic criteria concerning truth or (objective) evidence. Tere is no presumption that the extremity produced by belief polarization involves the adoption of beliefs that are false or in some way out of step with the evidence that there is. Similarly, it is not assumed that more extreme opinions are ipso facto more likely to be untrue. To be clear, belief polarization typically leaves those subject to it in an overall worse epistemic position in that it produces a shif in one’s belief regardless of one’s (subjective) evidence. Te phenomenon thus ofen leaves those subject to it less well-­justifed in holding their beliefs. But that is a point about the epistemic condition of the believer in the wake of belief polarization, not the beliefs that result from the phenomenon.

5  See Elizabeth Anderson’s contribution to this volume in Chapter 1 for additional details about variations in the degree to which diference groups are susceptible to this phenomenon.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

214  Robert B. Talisse Tese points make some progress in clarifying belief polarization. Still, one must ask: In what sense of extreme does belief polarization render us more extreme? What is involved in “moving further” in the direction of a prior inclination? Te matter is more complicated than it might seem, and the empirical literature is ofen insufciently nuanced when it comes to the specifcs of the shif that belief polarization provokes. For one thing, descriptions of the belief polarization experiments ofen overlook the diference between changes in degree of belief and changes in belief content. Observe that in some of the examples provided above, the shif towards extremity might be taken to consist strictly in a boost in the believer’s confdence in her initial belief. In the case of the feminists and chauvinists, for instance, one might say that discussion with fellow feminists produced a strengthening of participants’ feminist conviction. As the case is described, the participants seem to have the same beliefs afer the group discussion as those they held prior to it; they have shifed towards extremity only in the sense that they have come to hold those beliefs more frmly, or perhaps with greater confdence. Tus the case appears to involve an uptick in degree of belief. And, to be clear, a signifcant increase in one’s degree of belief is plausibly characterized as a shif towards extremity, at least in a certain sense of that term. However, in other belief polarization experiments it looks as if there is a shif in belief content. Take the mock jury experiment. In the course of likeminded discussion, the jurors each came to support a diferent punitive award from the one they advocated prior to talking. Te Colorado experiment appears to involve a similar change; conservatives started with a neutral view of the proposed global warming treaty, but moved in the course of likeminded discussion to a stance of opposition. To be clear, the shif from the belief that the plaintif deserves a $500 award to the belief that the plaintif deserves a $1000 award is plausibly regarded as a shif towards extremity. But a shif of this kind is diferent from an uptick in one’s degree of belief; accordingly the sense in which belief polarization results in increased extremity is correspondingly diferent in the two kinds of cases as well. It is not difcult to see why the distinction between shifs in degree and content is frequently elided in much of the empirical work. Te diference between belief-­ content and belief-­ degree is difcult to discriminate in experimental settings. Moreover, there are contexts in which the two are easily confated, even from the frst-­personal point of view. Tink again of the mock jury experiment. Te belief that the plaintif deserves an award of $100 identifes a more extreme punishment of the defendant than does the belief the plaintif deserves an award of $50. Tis naturally leads one to say, correctly, that when a juror shifs from the latter to the former, her assessment of the appropriate punishment increases in degree. It takes some care, though, not to confuse this point for the fact that the change from the latter to the former belief is a change in the content of the juror’s belief, not a change in her degree of confdence in a belief. In instances like the mock jury case, then, discussants are more properly said to shif to a more extreme belief content, which they might (or might not) hold with a greater degree of confdence than the degree of confdence with which they held their (diferent in content) pre-­discussion belief. Tat’s a mouthful, but it captures what seems to occur in some of the experiments.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Problems of Polarization  215 Perhaps, then, we should just say that there are two kinds of extremity shif that belief polarization might occasion. First, there may be an extremity shif of degree; in this case, in the course of discussion with likeminded others subjects intensify the surety with which they hold a belief with which they began and which persists throughout the conversation. Second, there may be an extremity shif in content; here, subjects replace their initial belief with a more extreme successor. Importantly, these diferent kinds of shif exclude each other. Where there is a shif in belief content, there cannot be an increase in degree-­of-­belief, and vice-­versa. However, this conclusion is too tidy to capture certain features of the examples. Recall that belief polarization involves an increase in uniformity among group members. Tis can occur only if there is a shif in belief content. Yet a shif in belief content precludes an intensifcation of degree of belief. To repeat, where there is a shif in belief content, there is no belief that persists throughout the discussion with respect to which individuals’ degree of belief could increase. Yet surely the mock jury and political protest cases involve both a change in belief content and some kind of overall intensifcation. Accordingly, we must add to the account a third feature with respect to which one might become more extreme. In addition to belief content and degree of belief, we can also speak of a believer’s overall commitment to his or her perspective.6 Tat is, we can say that belief polarization occasions the adoption of a successor belief that is more extreme in content, and it also involves an increase in the believer’s overall commitment to his or her perspective. Belief polarization seems then, to involve an intensifcation of something that is not a belief at all, but more like an overall investment in one’s epistemic condition. Again, as the chauvinist subjects become belief-­ polarized they not only shifed to more extreme chauvinist contents, they became more committed to their chauvinism. One suspects that in all of the experimental cases described above, subjects underwent an extremity shif in belief content in addition to an intensifcation of the overall commitment to their perspective. Here is why we should expect this: It is natural for those who hold more extreme belief contents with regard to some particular issue to also be more ardent or fervent devotees of the overall perspective within which that belief fts. So, for example, those who adhere to a radically uncompromising view about the morality of abortion should also be correspondingly more fervent and confdent pro-­lifers overall. Similarly, those who hold more extreme belief contents with respect to feminism should also be more confdent and ardent overall feminists. To put the point the other way around, there is something incongruous about holding an extreme belief content while also embracing generally lukewarm attitudes with respect to the overall perspective within which that belief resides. And, as I suggest below, the intensifcation of subjects’ attitude concerning their overall perspective is likely a crucial part of the explanation of the extremity-­shif in their belief contents.

6  On this, see Michael Lynch’s discussion of “intellectual arrogance” in Chapter 13 of this volume.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

216  Robert B. Talisse Tis three-­part distinction between belief content, degree of belief, and overall commitment to one’s perspective sheds some light on what it means to say that belief polarization makes us more extreme versions of ourselves. In the course of likeminded discussion, we come to replace antecedent beliefs with successors that have more extreme contents, we come to embrace a more ardent commitment to our overall perspective, and we might also adopt those new beliefs with a higher degree of confdence than the degree with which we held their ancestors. In short, belief polarization is a phenomenon by which we come to believe more extreme things in line with our antecedent inclinations, and also become more fervent advocates of our overall perspective.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.  Te Mechanism of Belief Polarization Let us turn next to the mechanism driving belief polarization. Start by recalling what the experimental cases have in common: discussion among likeminded people drives discussants to shif towards extremity in the direction of their pre-­discussion inclination. What explains this? Tere are two dominant views in the literature, what I will call the informational account and the social comparison account. In this section I will criticize each as incomplete; in Section 5, I will propose a third view that can capture the merits of the other accounts. Te informational account has it that the belief polarization phenomenon admits of the following obvious and refreshingly banal explanation. Tose engaged in discussion with likeminded others are exposed to a high concentration of afrming reasons, ideas, and “persuasive arguments” (Burnstein and Vinokur, 1977). Moreover, it is likely that for any particular participant, some of what is heard in the course of group discussion will be novel and innovative, considerations that he or she had not noticed before. Tis circumstance is accompanied by a corresponding scarcity within the group of articulations of countervailing or disconfrming considerations. Consequently, group members absorb the new information, and revise their own view in light of it. When this revision occurs amidst a general tendency among group members to underestimate or overlook the extent to which the “argument pool” from which they are drawing is skewed decidedly in favor of their position, one should expect shifs in the direction of extremity (Sunstein,  2017, p. 72). In fact, under such conditions, extremity shifs might be fully consistent with rationality; we might in some cases have adequate reason to shif. Although information-­exchange is surely part of the explanatory story of belief polarization, it cannot be the entire story. For one thing, belief polarization has been found to occur even when new and novel information is not presented in group discussion. In fact, it has been found to occur even in contexts where group interactions involve no exchange of information at all; as will be emphasized later, “mere exposure” to the fact that a group shares a general belief tendency is in some circumstances sufcient to produce the polarization (Zajonc,  1968; Myers et al.,  1980). Consider additionally that in likeminded groups, members who already hold an extreme view of the matter under discussion shif to an even more extreme position more drastically and more rapidly than those who begin from a more moderate

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Problems of Polarization  217 stance (Sunstein, 2009, pp. 40–2). Finally, within likeminded groups, the presence of an extremist member does not signifcantly amplify polarization, despite the fact that in discussion the most extreme members talk the most and speak at greater length (Van Swol, 2009, p. 194). Tese fndings are difcult to square with a strictly informational view of belief polarization. Assuming an informational explanation, one should expect polarization to occur only when information is being exchanged within the group; similarly, one should predict that groups with highly vocal extremists will polarize more radically than likeminded groups of moderates, and one should think that members of a likeminded group who are already extreme would tend to be less susceptible to further polarization. Tat said, it is not surprising that in instances where discussion in likeminded groups introduces new information and novel arguments, one fnds enhanced belief polarization. As is well-­documented, there is a general tendency among believers to overestimate the force of novel considerations that favor one’s antecedent view (Vinokur and Burnstein, 1978). But the success of this explanation of why belief polarization is especially pronounced under certain conditions does not establish that the phenomenon is fueled strictly by those processes. Additional explanatory tools are required. Te social comparison account of the mechanism driving belief polarization holds that the phenomenon is driven by social dynamics. Specifcally, the account holds that belief polarization is the result of in-­group comparisons. Here the claim is that discussion within a likeminded group is as much about social membership and cohesion as it is about transmitting information that supports the group’s shared belief. So, members of a likeminded group care about how they are perceived by the other members. In the course of discussion, they get a better feel for the general tendencies within the group, and, wanting to appear to others neither as a half-­hearted outlier nor as an over-­the-­top fanatic, they update their opinions to keep them in pace with what they perceive to be the dominant tendencies. More precisely, group members revise so that their view lies notably above what they perceive to be the mean, but beneath what they regard as unacceptably radical; they seek to be perceived by the group as “basically similar” to the rest of the group and yet “desirably distinctive” (Lamm and Myers,  1978, p. 185). Now, given that group members are engaging simultaneously in this kind of recalibration, and that the most fervent group members are likely to speak more and more ofen in discussion, the tendency to escalating extremity is to be expected. Espousing a view that refects what one perceives to be the non-­fanatical hardline within the group is a reliable way to signal to others one’s authenticity qua group member. Unsurprisingly, then, belief polarization is enhanced in contexts where group identity is made salient to the discussants (Abrams et al., 1990.). Tat is, group members shif further and more rapidly in the direction of extremity when their discussion is accompanied by an acknowledgment that members constitute a group not merely because they agree about the matter under discussion, but that they agree on that issue because they share some deeper social identity. What is more, likeminded groups shif less rapidly and to a lesser extent when discussion of the issue upon which they agree is conducted afer the recognition that the discussants share no deeper identity or are in some signifcant respect unalike (Le, 2007).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

218  Robert B. Talisse Once members of a subgroup within a mixed audience is primed to see themselves as a distinct contingent within the larger whole, they are less likely to attend to statements made to the group by those who are not members of their contingent (Abrams et al., 1990). Individuals who have been belief polarized are also more prone to the “backfre efect”; when a belief they hold on the basis of their group identity is contradicted by someone perceived to be outside the relevant group, their confdence in the challenged belief intensifes—they come to hold the belief even more ardently (Nyhan and Reifer, 2010; Munro and Ditto, 1997). Te informational and social comparison theories are frequently presented in the literature as competitors. However, the two views are not strictly opposed. Afer all, presenting strong arguments in favor of a view shared by a likeminded group is one way of signaling to others one’s authenticity as a group member. And ofering novel arguments that are also independently cogent and forceful can be a way of establishing oneself as particularly valuable to the group (Lamm and Myers,  1978, p. 186). Moreover, sometimes the relevant group to which one belongs fxes its identity in terms of a particular idea, argument, or piece of information. For example, consider identity groups organized around a conspiracy theory; fnding certain kinds of arguments persuasive, or some purported piece of information decisive, is a necessary condition for membership. Tings are much the same among communities of academics or artists who together see themselves as constituting a “school” or “movement”; the group identity has its locus in the acceptance of certain ideas and arguments. So the strict opposition between informational and social comparison views of the mechanism driving belief polarization is artifcial. Still, it should not be overlooked that even if one concedes that belief polarization is the product of social comparison in combination with informational factors, the extremity shif itself is not occasioned by reasons, arguments, and evidence. To be sure, the informational items may prompt or initiate the shif to a more extreme belief content and a heightened level of overall confdence, and, moreover, information might help determine the degree and rate of the shif, but the efect is primarily the product of social dynamics within the group. Tat is, in the absence of the relevant kind of group identifcation, the polarization phenomenon is absent.

5.  Te Corroboration View It might seem, then, that a version of the social comparison view that acknowledges that the relevant comparisons sometimes involve informational items would provide an adequate explanation of the belief polarization phenomenon. Alas, this is not correct; in the end, the social comparison view, like the informational account, is also incomplete. A third model is needed. It turns out that just as belief polarization can occur in the absence of the exchange of information, it can be induced in the absence of in-­group comparisons, too. Indeed, the phenomenon can be activated even in the absence of anything that would count as interaction among the members of the group; it is not face-­to-­face comparisons that drive the phenomenon so much as the subject’s own internal estimations of the dominant tendencies within his or her identity group.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Problems of Polarization  219 Although group discussion and other forms of in-­person interaction among likeminded people provide occasions where belief polarization reliably occurs, neither information-­exchange nor in-­group comparison is strictly necessary for the efect. Rather, the relevant extremity shifs occur simply in light of group-­afliated corroboration of one’s views. Tat is, belief polarization can occur simply when an individual is caused to feel that a group with which she identifes widely shares a view that she espouses. She need not hear any reasons in favor of the view, nor need she be in the presence of other members of the group with whom she can compare herself. Te brute impression that the relevant people afrm roughly the things that she afrms sufces for an extremity shif in belief content (Baron, 1996). In short, the realization that one’s belief is popular among one’s identity group suffces for belief polarization. To explain this, we must begin with a distinction between corroboration and confrmation. As the terms will be used here, confrmation involves evidentiary value, whereas corroboration need not contribute to one’s evidence. So, when one’s belief is confrmed, one obtains some new reason to hold it. Corroboration is diferent. When a belief is corroborated, all that happens is that the belief has been given an additional afrmation. However, as the new afrmation might be based upon the very same evidence as its predecessors, it might not add any evidential support for the belief in question. Simplifying slightly, we might say that whereas confrmation adds evidence, corroboration is simply a matter of popularity.7 From this a third account of the mechanism driving belief polarization emerges, what we can call the corroboration view. According to this account, an extremity shif in belief content and in overall commitment to our perspective can occur simply as a result of corroboration. Importantly, the corroboration can come by way of highly indirect channels. For example, being presented with data showing that liberals widely oppose genetically modifed food can prompt belief polarization within liberals who already incline toward that view. Exposure to a poll showing that conservatives overwhelmingly favor a particular military action can produce an extremity shif in the belief content of a conservative already favorably disposed to that action. It seems that the extremity shif is driven solely by the psychological tendency for corroboration to intensify one’s overall commitment to one’s perspective (Baron et al., 1996, pp. 558–9). Here is a simplistic explanation that captures what is going on. Corroboration from others with whom we identify makes us feel good about our beliefs (Sunstein, 2009, p. 29). When we feel good about what we believe, we experience a boost to our commitment to our perspective, we feel afrmed in our social identity. In turn, when we feel afrmed in this way, we intensify our attitudes and shif to more extreme belief contents.

7  I am setting aside cases where a large number of instances of corroboration might itself constitute a kind of evidence. Tese would be instances where the popularity of a belief among some specifc community is an indication of the strength and quality of the (other) evidence supporting it. In such cases, however, the popularity of a belief is a kind of evidence of there being evidence in its favor; this is arguably not itself evidence for the belief (though it might provide rational permission to adopt the belief in question). Here, the point is that the sheer number of corroborating voices, regardless of any judgment of the underlying evidence, functions to induce the extremity shif.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

220  Robert B. Talisse Tat corroboration can induce belief polarization might help to make sense of the vulnerability of belief-­polarized individuals to the backfre efect. If corroboration based on group identity is indeed the engine driving one’s confdence as a believer, then countervailing evidence profered by a member of an out-­group is bound to be experienced as a threat to one’s social identity. Tis would naturally result in an increased tendency to discount the weight of disconfrming considerations. But, further, in reaction to a posed threat to one’s identity, the “backfre” intensifying of one’s confdence is a way to reafrm one’s group membership, and possibly also a way of summoning additional support and corroboration from the group. For similar reasons, it should come as no surprise that once belief polarization has taken efect, exposure to opposing viewpoints—even in relatively moderate versions—tends to exacerbate belief polarization (Bail et al., 2018). To review, we have been considering a frm experimental result according to which mere corroboration from the relevant group reliably induces belief polarization (Baron et al., 1996, p. 559). Tis indicates that although belief polarization predictably occurs in discussions and other kinds of comparison-­enabling interaction among likeminded people, these settings are not necessary in order to produce the efect. Te question, then, is why corroboration sufces. Te view on ofer has it that there is a reliable psychological tendency by which corroboration of one’s beliefs from a group with which one identifes increases one’s overall commitment to one’s perspective, and this intensifcation of attitude in turn leads to extremity shifs in belief contents. An intriguing implication follows. As the relevant kind of corroboration can be indirect, extremity shifs can be induced simply by features of the social environment that make salient to individuals that some group with which they identify tends to embrace a view that they hold. Tese prompts need not be verbal, explicit, or literal; they can be merely implicit signals to group members that some belief is prevalent among them.8 Note further that as corroboration is really a numbers game, those with the power to present the appearance of widespread acceptance among a particular social group of some particular view thereby have the power to induce extremity shifs among those who identify with that group. I take it that readers are familiar with various warnings about how the Internet and, in particular, social media platforms are polarization machines (Sunstein,  2009, p. 24; Pariser, 2011; Vaidhyanathan, 2018). Te corroboration view suggests a result that is even more troubling. Irrespective of the choices we make to curate our informational exposure, our surroundings—the physical and social environments that we inhabit in our day-­to-­day life—can prompt belief polarization. Tus we have arrived at a conception of belief polarization that is more refned than the one with which we began. Belief polarization is the product not necessarily of discussion or even of comparison-­enabling interaction among likeminded people; it is rather a phenomenon that is activated when we are presented with reason to think that the beliefs that we hold on the basis of our salient social identities are 8  Hence Baron et al., (1996, p. 559) write, “simple attendance at certain events . . . laughter and applause at a joke . . . the wearing of political buttons or other symbolic garb or stigmata . . . may be all that is necessary to create such corroboration” (and thus belief polarization).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Problems of Polarization  221 popular among those who share that identity. In other words, when we are given the impression that our beliefs are highly corroborated among the members of a social group with which we identify, we will shif into more extreme versions of ourselves. Te mechanism by which this occurs seems to be that when something we believe is corroborated from within the relevant social group, we feel emboldened, and this intensifes our commitment to our overall stance; this intensifcation leads us to shif to more extreme belief contents in line with our stance. To repeat, likeminded discussion is a fertile site for belief polarization; but the argument above suggests that this owes less to any features of discussion and more to the fact that group discussion is reliably a context in which shared identity is made salient and dominant group beliefs are corroborated.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

6.  Te Polarization Dynamic Te corroboration view is disconcerting in several respects. It suggests that we are less in control of our doxastic lives—more susceptible to a kind of peer pressure, more conformist, more driven in our thinking by group membership—than we like. It also seems that counteracting these vulnerabilities is more difcult than we might have expected. Features of our environment can induce belief polarization, and they might be impossible to evade. Although it seems natural to locate the political trouble with belief polarization in the individual believers who are subject to it, we must not overlook that the shifs that belief polarization provokes have implications for inter-­group interactions. As belief polarization changes what we think, surely it also changes what we think of others. We should expect that as members of a given identity group transform into more extreme versions of themselves, there will be a corresponding shif in their views regarding those who do not share their group identity, and this will afect inter-­group relations. Indeed, we should expect belief polarization to be accompanied by an intensifcation of negative assessments of opposing groups, their members, and their views. And when it comes specifcally to the case of beliefs invoking our political identities, we should expect the phenomenon to produce afective political polarization. So although belief polarization occurs internally to believers in light of their group identities, it nonetheless produces spillover efects that impact social relations more generally. Tat is, belief polarization initiates a broader social and political dynamic. Corroboration from the relevant identity group increases our overall level of commitment to our perspective and emboldens us to adopt more extreme belief contents in line with our identity commitments. But from the standpoint of that kind of intensifcation, opposing views and countervailing considerations will appear to us as increasingly distorted, feeble, and unfounded. As belief polarization takes efect, we come not only to believe frmly things that are further out of step with our evidence, we also lose sensitivity to the reasons of our opponents. Accordingly, once we are sufciently belief polarized, those who espouse views that difer from ours will strike us as progressively more benighted, incoherent, and

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

222  Robert B. Talisse perhaps even unintelligible.9 Tus the intellectual distance separating us from others will seem to expand momentously; and indeed, those who hold views that run directly contrary to our own will strike us as irrational extremists, people devoted to untenable ideas (Westfall et al., 2015). What is more, as the opposition grows more unintelligible to us, their views will also come to look to us as increasingly monolithic. Consequently, we will be prone to conclude that their ideas, arguments, and criticisms are wholly without merit and thus not worth engaging, and we might even actively avoid contact with them. We will grow more inclined to see those who hold views that difer from ours to be in need of diagnosis and treatment than reasons and explanations; we will thereby also come to see them as political obstacles that need to be surmounted rather than fellow citizens in need of convincing. Of course, this more frmly ensconces us within our own group, which in turn further entrenches our group identity and enhances the belief polarization, driving us to further extremity. I take it that this dynamic will be evident to even casual observers of political discourse in contemporary democracies. Our social environments are increasingly crowded with calls to our political allegiances, which thereby make salient to us our political identities. On the basis of these prompts, we are exposed to belief polarization. And as belief polarization afects our views of others, we can discern a tight link between it and the forms of political polarization that we set out earlier. As we become more extreme versions of ourselves as political believers, we thereby increase uniformity within our political group (partisan political polarization), intensify our distrust and animosity for those who are diferent, and also create progressively deeper afective rifs across partisan divides (afective political polarization); and this in turn incentivizes parties and political elites to punctuate their opposition (platform political polarization). Again, all of this occurs within a self-­reinforcing and downward-­spiraling dynamic: belief polarization generates political polarization, which in turn fosters further belief polarization and sets the polarization dynamic in motion. Tus we actually become more like what our most vehement political opponents say we are; and they grow more closely to ft our images of them.

7.  Conclusion: Polarization as a Treat to Democracy We now can see the fuller range of problems that the various forms of polarization pose for democracy. Te political deadlock and partisan animosity that are commonly lamented are only the most obvious troubles. With a more detailed account of belief polarization, we can see how political polarization ofen has belief polarization at its root. And belief polarization strikes at the most fundamental capacities that we need as democratic citizens. As I suggested at the beginning of this chapter, belief polarization is democratically degenerative. By way of conclusion, it might be helpful to spell this out.

9  For more on this, see Chapters 14 and 13 by Elizabeth Edenberg and Michael Lynch, respectively, in this volume.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Problems of Polarization  223 Democracy of course is many things: a form of constitutional republic, a system of government, a procedure for collective decision, a method for electing ofcials, and so on. But underneath all of these ways of understanding democracy lies a commitment to the moral ideal of collective self-­government among political equals. To be clear, the democrat’s commitment to the political equality of the citizens does not amount to the idea that all citizens are equally admirable, or equal in every respect. Political equality is the commitment to the idea that in politics, no one is another’s subordinate.10 Put diferently, among political equals, all political power is accountable to those over whom it is exercised. In a democracy, even when an outcome is the product of impeccably democratic processes, citizens may still enact various forms of critique, protest, and resistance. Te democratic thought is that where citizens have rights to hold government accountable, they retain their status as political equals even while being subject to the law. Notice, however, that democracy is built on the premise that disputes over how political power should be deployed will be ongoing among citizens. Yet this commitment to enduring political contestation presents a difculty. Democracy depends on  the capacity of citizens to recognize the political equality of even those whom they regard as among their most benighted, vicious, and irrational democratic ­opponents.11 Tat is, in a democracy, citizens must regard each other as political equals, even when they disagree bitterly about things that matter most. Tey must sustain a commitment to recognizing one another as entitled to an equal say in politics, even when they are vehemently opposed on political issues. Maintaining this stance is not easy. Afer all, in a democracy the stakes ofen are high, and everyday policy disputes tend to invoke commitments concerning the most fundamental values, including justice, liberty, autonomy, and dignity. Hence we cannot help but regard at least some of those with whom we disagree as ipso facto committed to derelict conceptions of the values that matter most. Hence we may ask ourselves: In virtue of what are those who are committed to what I am bound to regard as injustice nonetheless entitled to an equal say? It is plausible to think that a central element of our present political dysfunctions lies within the dissolution of our capacities to see our fellow citizens as our political equals, despite our ongoing political disagreements. Te polarization dynamic leads us to regard those with whom we disagree as increasingly alien, irrational, immoral, and untrustworthy. And as we come to see our fellow citizens in these terms, we grow more inclined to see them as mere obstacles and impediments; this in turn encourages us to regard politics as the simple exercise of power.12 In regarding politics in this way, we abandon a core element of the democratic ideal and thereby 10  On this theme, see David Estlund’s Chapter 6 in this volume. 11  Of course, there are limits to what citizens are required to see as being within the bounds of properly democratic politics. Surely there are certain political commitments that are unreasonable in the Rawlsian sense, the holding of which disqualifes a person for democratic citizenship. Unreasonable persons pose special problems that the point here is not meant to address. 12  In his Chapter 12 in this volume, Jeroen de Ridder helpfully distinguishes “cognitive” and “practical” polarization—the former refers to what I have been calling “belief polarization,” whereas the latter captures the related point that as we belief-­polarize, we also grow more inclined to treat those with whom we disagree as outcasts, fools, and subordinates.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

224  Robert B. Talisse further hasten democracy’s dysfunctions. As I said at the beginning, it is not my aim in this chapter to propose a remedy for the problems that polarization raises for democracy. However, as we begin to think about how to address these problems, it helps to see that belief polarization does not merely transform us into more extreme versions of ourselves; given the dynamic it sets in motion, it also turns us into people who are less capable of dealing reasonably with those who are unlike us.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Abrams, D., M. Wetherell, S. Cochrane, M. A. Hogg, and J. C. Turner. 1990. “Knowing What to Tink by Knowing who you Are: Self-categorization and the Nature of Norm Formation, Conformity and Group Polarization.” British Journal of Social Psychology 29(2): 97–119. Anderson, E. 2021. “Epistemic Bubbles and Authoritarian Politics.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Bail, C. A., L. P. Argyle, T. W. Brown, J. P. Bumpus, H. Chen, M. B. F. Hunzaker, J. Lee, M. Mann, F. Merhout, and A. Volfovsky. 2018. “Exposure to Opposing Views on Social Media can Increase Political Polarization.” Proceedings of the National Academy of Sciences 115(37): 9216–922. Baron, R.  S., S.  I.  Hoppe, C.  F.  Kao, B.  Brunsman, B.  Linneweh, and D.  Rogers. 1996. “Social Corroboration and Opinion Extremity.” Journal of Experimental Social Psychology 32(6): 537–60. Burnstein, E. and A. Vinokur. 1977. “Persuasive Argumentation and Social Comparison as Determinants of Attitude Polarization.” Journal of Experimental Social Psychology 13: 315–32. de Ridder, J. 2021. “Deep Disagreements and Political Polarization.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Edenberg, E. 2021. “Political Disagreement on Social Media.” In E.  Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Estlund, D. 2021. “Epistocratic Paternalism.” In E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Hastie, R., D. Schkade, and C. R. Sunstein. 2007. “What Happened on Deliberation Day?.” California Law Review 95(3): 915–40. Isenberg, D. J. 1986. “Group Polarization: A Critical Review and Meta-Analysis.” Journal of Personality and Social Psychology 50(6): 1141–51. Johnson, N.  R., J.  G.  Stemler, and D.  Hunter. 1977. “Crowd Behavior as ‘Risky Shif’: A Laboratory Experiment.” Sociometry 40(2): 183–7. Lamm, H. and D. Myers. 1978. “Group-Induced Polarization of Attitudes and Behavior.” Advances in Experimental Social Psychology 11: 145–87. Le, E.-J. 2007. “Deindividuation Efects on Group Polarization in Computer-Mediated Communication: Te Role of Group Identifcation, Public-Self-Awareness, and Perceived Argument Quality.” Journal of Communication 57(2): 385–403. Lynch, M. 2021. “Political Disagreement, Arrogance, and the Pursuit of Truth.” In E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford: Oxford University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Problems of Polarization  225 Moscovici, S. and M. Zavalloni. 1969. “Te Group as a Polarizer of Attitudes.” Journal of Personality and Social Psychology 12(2): 125–35. Munro, G.  D. and P.  H.  Ditto. 1997. “Biased Assimilation, Attitude Polarization, and Afect in Reactions to Stereotype-Relevant Scientifc Information.” Personality and Social Psychology Bulletin 23(6): 636–53. Myers, D. G. 1975. “Discussion-Induced Attitude Polarization.” Human Relations 28(8): 699–714. Myers, D.  G. and G.  D.  Bishop. 1970. “Discussion Efects on Racial Attitudes.” Science 169(3947): 778–9. Myers, D. G., J. B. Bruggink, R. C. Kersting, and B. A. Schlosser. 1980. “Does Learning Others’ Opinions Change One’s Opinion?” Personality and Social Psychology Bulletin 6: 253–60. Nyhan, B. and J.  Reifer. 2010. “When Corrections Fail: Te Persistence of Political Misperceptions.” Political Behavior 32(2): 303–30. Pariser, E. 2011. Te Filter Bubble: What the Internet is Hiding from You. New York: Penguin. Schkade, D., C. R. Sunstein, and D. Kahneman. 2000. “Deliberating about Dollars: Te Severity Shif.” Columbia Law Review 100(4): 1139–75. Sia, C.-L., B. C. Y. Tan, and K.-K. Wei. 2002. “Group Polarization and Computer-Mediated Communication: Efects of Communication: Cues, Social Presence, and Anonymity.” Information Systems Research 13(1): 70–90. Sunstein, C. R. 2009. Going to Extremes: How Like Minds Unite and Divide. New York: Oxford University Press. Sunstein, C. R. 2017. Republic: Divided Democracy in the Age of Social Media. Princeton, NJ: Princeton University Press. Talisse, R.  B. 2019. Overdoing Democracy: Why We Must Put Politics in its Place. New York: Oxford University Press. Vaidhyanathan, S. 2018. Anti-Social Media: How Facebook Disconnects Us and Undermines Democracy. New York: Oxford University Press. Van Swol, L.  M. 2009. “Extreme Members and Group Polarization.” Social Infuence 4(3): 185–99. Vinokur, A. and E.  Burnstein. 1978. “Novel Argumentation and Attitude Change: Te Case of Polarization Following Group Discussion.” European Journal of Social Psychology 8(3): 335–48. Westfall, J., L. Van Boven, J. R. Chambers, C. M. Judd. 2015. “Perceiving Political Polarization in the United States: Party Identity Strength and Attitude Extremity Exacerbate the Perceived Partisan Divide.” Perspectives on Psychological Science 10(2): 145–58. Zajonc, R. 1968. “Attitudinal Efects of Mere Exposure.” Journal of Personality and Social Psychology Monograph Supplement 9(2): 1–27.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

12

Deep Disagreements and Political Polarization Jeroen de Ridder

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.  Introduction: Happy Toughts about Disagreement Is disagreement good news? Yes, say a choir of voices in the sciences and ­humanities.1 Disagreement presents opportunities for epistemic self-­improvement. It presents us with opportunities to weed out false beliefs, to acquire true beliefs, to better proportion our beliefs to the evidence, and to recalibrate the reasons we have for our beliefs. Under certain conditions, diverse groups are better at solving problems and reaching true beliefs than even the most expert individuals (Hong and Page’s (2004) “diversity trumps ability” theorem). For the context of liberal democratic politics, in particular, some philosophers have sung the praises of disagreement as well.2 In fostering freedom of speech and inquiry, inclusiveness, equality, and reasonable deliberation, liberal democracy is supposed to be particularly good at harnessing the epistemic power of disagreement among citizens. Tere are dissonants, however. Te idea that deliberation among disagreeing citizens leads to epistemically superior decisionmaking—or at least guards against false or unjustifed outcomes—has been criticized by a host of empirical work in psychology and political science.3 It is also unclear that other means of harvesting the “wisdom of crowds,” such as belief aggregation, Condorcet’s jury theorem, or the “diversity trumps ability” theorem, have much application in political reality.4 In this chapter, I want to add a distinctly epistemological objection to the epistemic ideal of liberal democracy. It stems from the occurrence of deep disagreements in liberal democracies. Such disagreements undermine a crucial presupposition of epistemic democracy, to wit the availability of common ground for reasonable debate and deliberation. Moreover, they lead citizens to see each other as less than fully

1  Surowiecki 2004; Hong and Page 2004; Sunstein 2006; Christensen 2007; Page 2007; Landemore and Elster 2012. 2  Popper  1945; Mill 1977; Cohen  1986; Habermas  1990; Anderson  2006; Estlund  2008; Ober  2010; Ahlstrom-Vij 2012; Landemore 2013, 2017; Goodin and Spiekermann 2018. 3  See Mendelberg 2002; Ryfe 2005; Achen and Bartels 2016 for elaborate reviews. Brennan 2016, ch. 3 gives a particularly pessimistic reading of the fndings. I will review some relevant literature in Section 4 below. 4  Brennan 2014; Houlou-Garcia 2017; Ahlstrom-Vij 2019. Jeroen de Ridder, Deep Disagreements and Political Polarization. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Jeroen de Ridder. DOI: 10.1093/oso/9780192893338.003.0013

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Deep Disagreements and Political Polarization  227 rational, as morally subpar, or worse. Tis, in turn, feeds into polarization, which makes reasonable debate harder still. Tis objection is not supposed to be independent of the empirical worries mentioned above. In fact, some of the empirical fndings readily lend themselves to an interpretation along epistemological lines and my purpose is to bring this to the fore. In doing so, I hope to draw attention to the unnoticed normative-­epistemological dimensions of polarization. Tis can help us understand the empirical fndings better and can contribute to thinking about solutions. Te plan is as follows. Section 2 explains what deep disagreements are. Section 3 outlines conceptual relations between deep disagreement and our evaluation of each other’s rationality and moral standing. Tis shows how deep disagreements threaten the epistemic benefts of democracy. In Section 4, I connect deep disagreement with empirical literature from psychology and political science and argue that deep disagreements are implicated in political polarization. Deep disagreement is not just a theoretical problem, but creates real trouble in society. Section 5 addresses some worries and adds further texture and qualifcations to my proposed account of the relation between deep disagreement and polarization. Section 6 wraps things up.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2.  Disagreement: Ordinary and Deep People disagree on lots of things: from the mundane and insignifcant to the profound and life-­ changing. Many of these disagreements are what I will call ordinary. Tey concern things like everyone’s share of a restaurant check, whether your friend was at last week’s meeting or not, or what the most efective measure is to do something about your department’s defcit. Ordinary disagreements have a number of characteristics. First of all, they are localized. Parties disagree about a single proposition or a few closely related ones, but they agree on lots of issues surrounding the disputed question. Second, they tend to be rationally resolvable. Typically, there are mutually agreed upon courses of action the parties can undertake to settle who is right: redo the calculations, check the minutes of the meeting, investigate further, etc. Doing so typically leads to a resolution that is acceptable to both parties. Tis is not to say that ordinary disagreements are always easily resolvable. Resolutions can be cumbersome and time-­consuming, for instance when a complex scientifc issue is at stake. But even in such cases, there is a reasonable expectation that the disagreement will be rationally resolved sooner or later. Tird, ordinary disagreements take place within a shared normative framework. Both parties agree on core moral or political values, on what counts as evidence or as a good argument, how evidence ought to be weighed or arguments evaluated, etc. Tis is why they both recognize and accept that checking calculations or consulting minutes is the way forward. Te methods for resolving their dispute are uncontroversial. Sometimes, however, disagreements are deep. Deep disagreements are typically long-­standing, intractable, and entrenched. Examples include traditional political divides between lef and right, progressive and conservative, or secular and liberal.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

228  Jeroen de Ridder Ofen, these divides play out around more specifc issues: the role of government, economic inequality, taxation, abortion, physician-­assisted suicide, immigration, racism, etc. Diferent schools in scientifc disciplines can also exemplify deep disagreements: classical, Marxian, Keynesian economics; methodological individualism versus holism in social science; empiricism versus rationalism in philosophy. John Rawls (1996) famously assumed that reasonable pluralism and the deep disagreements accompanying it are a fact of life in liberal societies. As a frst approximation, deep disagreements are disagreements that lack the three characteristics of ordinary disagreements listed above. First, they are not just local, but involve clusters of related issues. People who disagree about immigration policies tend to also disagree on the causes and efects of racism, on welfare programs, and economic inequality. Second, deep disagreements are very difcult if not impossible to resolve, because the parties involved do not agree on how to resolve them. If they need to cooperate anyway, they might “agree to disagree” and work out a practical compromise without conciliating on the substance of their disagreement. Tird, the reason for this is that parties do not share a common normative framework. Tey have diferent underlying views and commitments about what good evidence is, how diferent sources ought to be weighed, who the experts are, etc. As a result, they cannot agree on how to resolve their dispute: one party might take the issue to be clearly settled by, say, the available scientifc evidence, whereas the other might question the force of this evidence, pointing to counterevidence provided by other scientists or (perceived or genuine) experts. Some deep disagreements involve conficting underlying epistemic norms and principles, but deep disagreements—especially in politics and morality—can also involve conficting underlying moral commitments.5 Disagreeing parties might hold diferent fundamental moral values or they might rank the same values diferently. Religious believers will value obedience to a deity, while this makes no sense to nonbelievers. Part of what is at stake between libertarians and communitarians is that they give diferent weight to the value of individual freedom. Similarly, the disagreement between the pro-­choice and the pro-­life positions may involve a diferent view on the value of human life.6 To summarize this in a succinct formulation, we can characterize deep disagreements as follows: (DD) Deep disagreements are those disagreements in which parties disagree about (or are committed to disagreeing about) relatively fundamental epistemic or moral values and principles even afer full disclosure.

5  Depending on one’s views about the nature of morality, one might worry that moral disagreements are too diferent from factual disagreements to be treated in the same way. I will address this concern briefy in the following discussion by indicating what assumptions are involved in broadening the characterization of deep disagreement in this way. 6  To be clear, I’m not suggesting that this is all that is at stake in these disputes. Fundamental political and moral disputes typically also involve signifcant factual disagreement about, say, human psychology, the nature of personhood, etc. (cf. Fogelin, 1985).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Deep Disagreements and Political Polarization  229 A few clarifcations are in order. Deep disagreements can concern factual issues, but must always also involve relatively fundamental epistemic or moral normative principles. Tis is why DD contains the parenthetical clause. A physician and an antivaxxer might disagree about the factual question whether vaccines can cause autism, but their disagreement is deep because they also disagree about underlying epistemic principles concerning the weight of diferent sources of evidence.7 Te depth of disagreements is a gradual phenomenon on DD. Epistemic and moral norms and principles come in diferent levels of abstractness and generality. Some principles are fundamental and broad: under relevantly normal conditions, sense perception is reliable; pain is bad. Others are derived and specifc: randomized clinical trials give stronger evidence than explorative studies with small sample size; don’t break promises. Tis means that not all disagreements which involve disagreements about underlying epistemic or normative principles are deep—or at least that they are not all equally deep. When parties disagree about which specifc experts to trust, they disagree about an epistemic principle or an application thereof. But as long as they agree about more basic principles stipulating the qualities of genuinely trustworthy experts, they may still be able to resolve their dispute relatively easily by talking through the correct way of applying these more basic principles to the situation at hand. Deep disagreements thus come in degrees. Tey are deeper to the extent that they involve more fundamental normative commitments. Tis also means that there will be borderline cases between ordinary and deep disagreements. Te deepest disagreements are those in which one party outright rejects a fundamental epistemic or moral principle that the other party accepts. Te epistemological literature on disagreement has given prominence to peer disagreement: situations in which the two disagreeing parties are roughly equal in terms of cognitive virtues, freedom from distorting infuences, and familiarity with the relevant evidence.8 Te reason DD does not include peerhood as a condition is that the very nature of deep disagreements makes it difcult to say whether or not disputants are epistemic peers. In the absence of a shared normative framework of underlying epistemic or moral principles, each party will easily—and with some justifcation, at least by their own lights—think the other less cognitively virtuous, fundamentally misguided, or badly informed. Deep disagreements can even prevent one party from recognizing the other as an epistemic superior when this is in fact the case according to objectively correct epistemic or moral standards. A creationist might write of an evolutionary biologist and (wrongly) disregard her opinions altogether. A racist might lend no credence to empirical fndings establishing unequal treatment of people of color, because he takes most social scientists to be biased. Tis raises hard questions: If parties in a deep disagreement have a hard time judging each other’s epistemic credentials, are they rationally of the hook in ignoring each other’s opinions? Are deep disagreements inevitably deadlocks where neither 7  Another way of describing the situation would be to say that the factual disagreement as such is not deep, but the one about the underlying epistemic principles is. I have no serious objections to this, except that it strikes me as artifcial to divide what appears to be one disagreement into two separate ones. 8 Te literature has grown vast, but some landmark contributions include Kelly (2005), Feldman (2006), Christensen (2007), Feldman and Warfeld (2010), and Christensen and Lackey (2013).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

230  Jeroen de Ridder party is in a position to learn from the other? I will not attempt to address these questions here.9 My concern is diferent: I want to look at how deep disagreements are implicated in polarization in society. DD stipulates that the disagreement must remain even afer full disclosure. Tis is to prevent those disagreements from counting as deep in which one party has simply never considered some epistemic or moral principles held by the other party, but, were she to do so, would readily accept them. Someone who was raised a Kantian and has never heard of utilitarianism might immediately be converted once she learns of it. Someone with a poor grasp of scientifc methods might readily come to accept scientifc expertise once she acquires a better understanding of science. Whether a given disagreement is deep depends not just on the disputed issue, but on the underlying normative commitments of the disagreeing parties. Not any disagreement about, say, open borders is automatically deep, although it certainly can be. Disputants may well share a moral and epistemic framework, but have conficting beliefs about the economic efects of immigration, how easily immigrants blend into a new culture, crime rates among immigrants and non-­immigrants, etc. For others, a dispute about the same issue can be deep because it arises from diferent underlying moral principles about the importance of nation-­states, national cultures, or even ethnic purity. Another consequence is that disagreements may become more or less deep over time as a result of disputants changing their moral or epistemic commitments. My characterization of deep disagreements difers from similar proposals by Michael Lynch (2010, and Chapter  10 this volume) and Klemens Kappel (2012, 2018).10 Both limit their account to disagreements involving fundamental epistemic principles, whereas I have included disagreements about relatively fundamental moral principles. My reason for doing so is that many real-­life deep disagreements do not involve (only) factual issues but (also) moral values and principles. Broadening the characterization of deep disagreements in this way does introduce complications, since whether or not disagreements about normative moral principles can be treated in the same way as disagreements about normative epistemic principles depends partly on meta-­ethics. For error-­theoretical, relativist, or non-­cognitivist views in meta-­ethics, this is not the case because moral principles are not objectively true or false, or not even the sort of thing that can be true or false, or correct or incorrect.11 For simplicity’s sake, I will assume that there are objective normative facts in both the moral and epistemic domain.12 Another diference with Lynch and Kappel is that both of them narrow the class of deep disagreements to those involving fundamental epistemic principles, which they understand as epistemic principles that can only be defended by epistemically 9  See Lynch (2010), Kappel (2012, 2018), Ranalli (2018a) for discussion and proposed answers. 10  It also difers from the Wittgensteinian theory of deep disagreements, according to which deep disagreements involve commitments to diferent hinge propositions. For a comparison of these views, see Ranalli (2018b). 11  Unless one subscribes to fully general versions of these views and also takes them to apply to the epistemic domain. Tese views are relatively rare in mainstream analytic epistemology, however. 12  It remains to be seen whether what I say below about deep disagreement and polarization holds without this assumption, but investigating this in detail would take us too far afeld.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Deep Disagreements and Political Polarization  231 circular arguments.13 I agree that disagreements involving such principles are deep, but this characterization rules out many realistic cases of seemingly deep disagreements. Nobody (save the imaginary radical skeptic) really rejects fundamental principles specifying our justifed reliance on basic sources of knowledge such as sense perception, memory, induction, deduction, testimony, etc. Real-­life deep disagreements concern derived but still relatively fundamental epistemic principles. Take the example of the creationist and the evolutionary theorist, which both Lynch and Kappel use to illustrate deep disagreement. Tis does not involve wholesale ac­cept­ ance or rejection of sense perception, inductive reasoning, or testimony, but derived principles about sense perception and reasoning that are applied to certain topics, to textual interpretation of historical sources, and to testimony (or the relative weights of these principles). Te exclusive focus on fundamental epistemic principles in Lynch’s and Kappel’s proposals detracts from an important issue. Relatively fundamental derived epistemic or moral principles sometimes function as fundamental for individuals, in the sense that they are unwilling to give them up, yet cannot defend them by independent reasons. Tis need not be a matter of irrational stubbornness. Applying fundamental principles to the real world ofen involves judgment calls and sometimes there may be more than one way of making these calls reasonably.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Epistemic and Moral Frameworks Epistemic rationality (or reasonableness) is a normative notion with positive valence. Saying that someone’s belief is epistemically rational, is to say that this belief is good, acceptable, responsible, or legitimate; that it conforms to standards of epistemic goodness for beliefs.14 In order to avoid having to delve into the extensive literature on epistemic rationality, I will rely on an abstract characterization of the notion, which should be widely acceptable regardless of people’s more specifc views. Following Alvin Goldman (2010), we can think of epistemic rationality as characterizable by a set of norms and principles about how to think; about how to form, maintain, or abandon beliefs.15 Such norms link an agent’s evidential situation and broader cognitive position and context to what doxastic attitudes are epistemically appropriate for her. For instance, for a subject in typical circumstances, hearing the sound of an approaching car makes it appropriate for her to believe a car is approaching. For a radiologist with relevant experience and background knowledge to see certain patterns in a patient’s X-­ray photograph makes it appropriate to draw conclusions about the patient’s condition. We implicitly rely on such norms when we 13  Cf. Alston (1993) for discussion of the inevitable epistemic circularity of arguments defending sense perception and other basic sources of knowledge. 14  As Richard Feldman (2006, p. 221) points out, “reasonable” is sometimes used in a minimal sense as a synonym for “not crazy.” On this use, saying that someone is reasonable just means that she is not doing very badly from an epistemic point of view. Te notion of rationality that I am interested in here is more demanding. Being rational is not just steering clear of the worst; it requires doing well. 15  Depending on one’s views about the nature of rationality, these norms might be norms that express or codify what epistemic justifcation, epistemic responsibility, proper cognitive functioning, etc. amounts to.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

232  Jeroen de Ridder evaluate our own or someone else’s cognitive conduct. We can call a complete set of such norms and principles an epistemic framework, or e-­framework for short.16 We can say something analogous about morality. To say that someone behaves morally is to pay her a compliment; it is to say that she is doing well from the moral point of view. Morality can also be thought of as describable by a system of norms and principles linking the morally relevant features of situations and agents to what the appropriate (morally permissible, right) actions are for the agent.17 Together, these norms and principles form a moral framework, or m-­framework. Tis characterization, too, is intentionally neutral between diferent substantive moral views and theories. A question for both the epistemic and moral domain is whether there is one objectively correct framework. For my purposes, however, it is enough to observe that people in fact adhere to diferent e/m-­frameworks and that they can be rational in doing so. Given people’s upbringing, education, and social environment, it makes perfect sense that diferent people will trust diferent experts, attach diferent weight to scientifc evidence, or trust common sense to difering degrees; and that they hold diferent basic moral values, rank them diferently, derive diferent moral norms from them, or apply norms in diferent ways. Even if some of these frameworks are in fact objectively wrong, people do employ them and can be fully rational in doing so.18 Tere is a straightforward connection between deep disagreements and e/mframeworks. As DD stipulated, deep disagreements involve disagreement about ­relatively basic epistemic or moral values and principles. In other words, they involve disagreements about (parts of) e-­ frameworks or m-­ frameworks. When parties employ frameworks that difer in relatively basic principles and hold conficting beliefs as a result, their disagreement will be deep. If you rely on systematic scientifc evidence about the safety of vaccination and I trust anecdotal evidence from my friends and put more stock in natural medicine in general, our disagreement is deep, because our factual disagreement about the safety of vaccines is the results of our reliance on diferent relatively basic epistemic principles, that is, our reliance on diferent e-­frameworks. People with diferent e/m-­ frameworks do not necessarily disagree about everything, nor will every disagreement between them be deep. It all depends on how their frameworks bear on the issue at hand. Climate change sceptics and IPCC-­ members might agree on basic statistical data about temperatures, even though they 16  Tis characterization intentionally leaves a lot unspecifed so as to accommodate diferent views of rationality: whether or not rationality is permissive, whether or not it can be specifed through a formal framework such as Bayesianism, whether (and how) contextual factors or practical stakes make a diference to what is rational, etc. 17  One might worry that this is false for virtue ethics and particularism, since both of these views hold that morality cannot be captured in universal and context-­independent norms and principles. In reply, note that we can think of moral norms and principles as being very narrow in scope, applying only to one or a few particular circumstances, and dependent on lots of contextual particularities. 18  See Goldman (2010) for an argument to this efect about e-­frameworks. His argument is easily adaptable to apply to m-­frameworks, too. John Rawls ofers a similar explanation for pluralism, which appeals to the “burdens of judgement,” i.e., the intractable difculties that even reasonable people of good will face in answering the deepest questions about how to live (Rawls, 1996, pp. 54–8).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Deep Disagreements and Political Polarization  233 disagree deeply on the explanation of the data. Tis is because diferent e/mframeworks typically have considerable overlap. Where a particular belief relies only on principles that are common to diferent frameworks, we should not expect disagreement, or only ordinary, easily resolvable disagreement. Because deep disagreements involve difering e/m-­ frameworks, rationally resolving them is difcult or impossible. When disputants rely on diferent epistemic or moral principles in giving reasons for their beliefs, they will have trouble recognizing each other’s reasons as rational contributions to the debate. When a young earth creationist appeals the authority of the bible or principles of biblical exegesis in arguing for her belief about the age of the earth, this will carry no weight whatsoever for the scientifcally minded geologist. Te latter will not even see the former’s argument as a rationally acceptable contribution to the discussion. Similarly, the pro-­lifer’s insistence on the sanctity of the life of an early-­stage fetus might strike the pro-­choicer as a clear misapplication of a moral value. In so far as deep disagreements involve incompatible derived e/m-­principles, parties can try to resolve their disagreement rationally by appealing to more fundamental principles. But when the disagreement is entrenched and involves relatively fundamental e/m-­principles, this is likely to fail. Disputants will be as committed to what they take to be correctly derived principles as they are to fundamental principles themselves. Tat is to say, they will treat relatively fundamental e/m-­principles as just as fxed and resistant to rational revision as absolutely fundamental ones. When a geologist proposes that the fundamental principle of trust in testimony leads to a derived principle saying that scientifc experts ought to be trusted more than ancient scriptures, the creationist will disagree with that specifcation and maintain that divinely inspired infallible biblical reports take precedence. Let us consider the efect of deep disagreements on how disagreeing parties evaluate each other. Obviously, people who disagree see each other as mistaken. So do people who disagree deeply. However, the connection between deep disagreement and e/m-­frameworks adds a further layer to how the disputants evaluate each other. Not only will they see each other as having made a mistake about an issue, they will also see each other’s way of thinking or reasoning about the issue as mistaken. Tey will consider each other as less than fully rational or moral. Depending on factors such as how diferent their e/m-­frameworks are, how important or central to their identity they take their non-­shared e/m-­principles to be, and how confdent they are about their own e/m-­framework, this mutual evaluation can go from less than fully rational to irrational to stupid, and from morally subpar to immoral to downright evil. Te upshot is that deep disagreements challenge any optimistic take on the efects of disagreement. Te broadly Millian ideal in which disagreement leads to productive intellectual engagement, correction of error, and mutual learning presupposes that  disagreeing parties recognize each other’s utterances as reasonable, valid contributions to the debate. Tis may be fne for ordinary disagreement, but deep disagreements undermine this presupposition. When parties don’t share an e/mframework, they will fail to recognize some of each other’s reasons as epistemically reasonable or morally acceptable ones. Depending on how important they deem their non-­ shared e/m-­ principles and how confdent they are about their own frameworks, they will come to see each other as irrational, immoral, or even stupid

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

234  Jeroen de Ridder or evil. Tis, in turn, will make them even less inclined to take each other seriously and learn from each other. Rather than produce mutual learning, then, deep disagreements can easily lead to mutual rejection and animosity.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4. Polarization My argument about the trouble with deep disagreement so far has been a priori. Investigating the nature of deep disagreements showed that they pose an obstacle to reaping the benefts of disagreement. In this section, however, I want to make the case that deep disagreement is implicated in real troubles in society by connecting what I have said so far with empirical literature from psychology and political science on political polarization. For ease of exposition, I will frst sketch a simplifed picture of how deep disagreement can contribute to diferent kinds of polarization. In Section 5, I will qualify this picture and explain more carefully how it relates to the empirical fndings. As Robert Talisse’s Chapter 11 in this volume makes clear, there are diferent kinds of polarization. Here, I want to draw a further distinction between what I will call cognitive polarization and practical polarization. Te former has to do with how individuals think about each other, the latter with how they treat each other. Te more cognitively polarized people are, the less they think of each other’s rationality, intelligence, moral decency, trustworthiness, etc. Te more practically polarized they are, the worse they tend to treat each other.19 As people act on the basis of their beliefs, these two forms of polarization are related. Consider the relation between deep disagreement and cognitive polarization frst. People in a deep disagreement will easily come to see each other as irrational, immoral, or worse. Naturally, you don’t give much weight to the opinions of someone you regard as such. You will discredit their opinions or, in the case of protracted disagreement, decide it is not worth listening to them at all anymore. Not only is this a natural response; it is arguably an epistemically justifable and rational response, too.20 Just as we take ourselves to have excellent reason for not taking horoscopes seriously, we will take ourselves to have good reasons to ignore those with whom we disagree deeply. Moreover, this efect spreads out. Deep disagreements tend to be clustered because e/m-­principles typically afect belief formation on a number of more or less related issues. As a result, we will disregard our opponents’ opinions on not just a single issue, but a cluster of issues. Tis situation sets us up for a host of well-­documented psychological efects. Tis frst is myside bias (or confrmation bias, cf. Nickerson (1998) and Baron (2008, pp. 203f) for details and references). Tis stands for a set of tendencies that humans 19  Note that my distinction cross-­cuts Talisse’s distinction between political polarization and belief polarization to some extent: my cognitive polarization involves both what he calls afective political polarization and belief polarization, whereas he doesn’t talk explicitly about practical polarization. 20  See Kelly (2008), Rini (2017), and Fantl (2018) for arguments to the efect that such “partisan epistemology” can be justifed and rational. Whether a partisan response is justifed and rational will depend on other factors, such as the number and kinds of people with whom you agree and disagree; see Lackey (2013) for discussion.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Deep Disagreements and Political Polarization  235 have when they think about issues and process evidence. People tend to pay more attention to evidence that confrms what they already think, even in cases where their beliefs have very little support to begin with. Tey treat new evidence asymmetrically: evidence that sits well with their prior beliefs is accepted uncritically, whereas disconfrming evidence is subjected to more critical scrutiny. In so far as available evidence is ambiguous or open-­ended, people interpret it in such a way that it confrms, rather than calls into question, their prior beliefs. Tey also actively look for evidence that confrms what they already believe and avoid disconfrming evidence. Tese tendencies are even stronger when the issue at stake is emotionally charged: something that people feel strongly about. Needless to say, this ofen happens with moral or political issues, which tend to form part of people’s social identities. An experiment by Dan Kahan and collaborators illustrates these tendencies very strikingly (Kahan et al.,  2017). People were presented with numbers about the efectiveness of a skin cream for treating a rash. Arriving at a correct answer about the cream’s efectiveness required some mathematical ability: the required computations were really a version of an exercise ofen used in social science to test people’s aptitude for slow, refective (system 2) thinking. Predictably, people who were better at math tended to get the right answer more ofen. So far so good. Next, however, people were presented with a politicized version of what was mathematically the same problem. Te numbers were now presented as being about whether gun control laws are efective in decreasing crime. Startlingly, people’s general math aptitude was no longer the best predictor of whether they would get the answer right. Instead, it was people’s political identity that predicted best whether they would solve the puzzle correctly. Liberals tended to solve the problem correctly when the numbers showed gun laws were efective and did not when they showed the opposite. Conservatives’ performance was the mirror image: they did well when the numbers supported their prior beliefs and badly when they did not. Most disturbingly, perhaps, the better people were at math, the worse they did when the numbers did not support their prior convictions. Apparently, people “successfully” use their sophisticated reasoning skills to wriggle their way out of evidence that disconfrms their political views. Similar results were found in other experiments (Kahan et al., 2011; Kahan, 2013). Te connection with deep disagreement is straightforward. If people evaluate those with whom they disagree deeply as irrational or immoral, this plays directly into myside bias. It is already tempting to discard information that does not ft with your prior belief anyway, let alone if it comes from sources you take to be irrational or immoral. Tings become even worse when group dynamics are brought into the picture. As Robert Talisse’s chapter details, it is a widely confrmed fnding in social psychology that various sorts of groups tend to polarize.21 Groups of likeminded people move towards more extreme versions of their views, to higher degrees of conviction, and to greater confdence in their overall perspective on the issues. Tis happens not only 21  Sunstein (2017, ch. 3) provides an even more elaborate treatment of the phenomenon with reference to the psychology literature.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

236  Jeroen de Ridder when they deliberate and process new evidence, but already when they merely talk about old evidence. If Talisse’s proposed account of the mechanisms driving polarization is right, it can even happen in the absence of any talk or deliberation. It is enough if group members’ sense of being right is corroborated one way or another through explicit or implicit signals from the physical or social environment. Te connection between group polarization and deep disagreement is again fairly straightforward. When we disagree deeply with others, we think their e/m-framework is misguided and thus we easily see them as less than fully rational or not morally upstanding. As a result, we pay less attention to their views. Since this efect is ­asymmetrical—we do not downgrade the opinions of those who agree with us and share our e/m-­frameworks—we are lef with more confrmation for our beliefs and our e/m-­framework.22 In other words, we move to a situation where the textbook conditions for group polarization are satisfed. As a result, your prior beliefs are bolstered, you might accept more extreme versions of your beliefs, and your confdence in your own framework increases. Tis, in turn, makes it seem even more justifable to write of the opinions of others who do not share your framework and to see them as irrational or immoral, leaving you with even more self-­corroboration. Vicious circle completed. Let us turn to practical polarization next. An optimist might still feel things are not so bad. As long as people only think poorly of each other, but continue to treat each other well, no real harm is done. But this would be rather remarkable. Even though thinking less of our fellow human beings does not necessarily lead to treating them worse, it is very easy to treat “those people” a bit worse, at least in some respects and some of the time. Tis is familiar enough from everyday experience. We would rather not invite that uncle who makes inappropriate racist jokes to our birthday party. If at all possible, we avoid giving the foor to the department’s retired Wittgensteinian who inevitably scorns speakers for failing to see that they were speaking about a pseudo-­problem. Isn’t it fun to laugh at those rednecks with their conspiracy beliefs or, in Europe, small-­ town nationalistic populists with their narrow-­ minded beliefs about immigrants, traditional values, or the EU? And vice versa, Republicans were looking forward to “making liberals cry again” in 2020. Besides anecdotal evidence, there is scientifc research which establishes connections between thinking poorly of each other’s epistemic or moral standing and treating each other worse.23 I will limit myself to the literature on intergroup bias and confict here (Hewstone et al.,  2002). As we saw above, cognitive polarization divides people into “us” and “them” and increases contrasts between the two. Tis then leads to stereotyping or caricaturing outgroup members, ingroup favoritism, prejudice against the outgroup, and discrimination or other forms of unfair and poor treatment of outgroup members, especially—but not exclusively—in situations 22  Tis is not to say there cannot be sustained and polarization-­inducing disagreement among people with the same e/m-­framework. But this is less likely than disagreement with people who have diferent e/m-­frameworks and, importantly, there is no similarly strong basis for discarding those with whom you disagree as irrational or immoral. 23  Section 2 of Talisse’s Chapter 11 gives further references.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Deep Disagreements and Political Polarization  237 where groups perceive each other (correctly or incorrectly) as being in competition for goals or limited resources.24 Various mechanisms have been proposed for when, why, and how exactly all this happens, but for my purposes it is enough that intergroup bias breeds adversarial intergroup behavior. A widely-­cited study by Shanto Iyengar and Sean Westwood (2015) provides striking evidence in a politically polarized context. In the experiment, over 1000 participants with diferent self-­ ascribed political identities (Democrat, Lean Democrat, Independent, Lean Republican, or Republican) were asked to evaluate résumés of recent male high-­school graduates with an eye to awarding a $30,000 scholarship for apolitical purposes. Tey created diferent versions of the résumés which varied on two variables: candidates’ academic achievement, expressed by their GPA (either 3.5 or 4.0), and their political identity, signaled by extracurricular activities (president of either the Young Republicans or the Young Democrats). Te results clearly show ingroup favoritism at work. Democrats ended up selecting Democratic candidates in about 80 per cent of the cases; Republicans selected the Republican candidate in about 70 per cent of the cases. But most tellingly, academic achievement mattered less than co-­partisanship. While, arguably, GPA should be the crucial determinant in awarding a scholarship, it turned out that political identity mattered more. Even when the Republican candidate had a higher GPA than the Democrat, the probability of a Democrat selecting the Republican candidate was only 0.3. When the Democratic candidate had a higher GPA, the odds of Republicans choosing the Democratic candidate were only 0.15.25 Te same study also showed that people tend to give less money to members of the opposing party in trust and dictator games. Although the literature on intergroup confict does not talk about deep disagreement explicitly, the connections are easy to spot. Deep disagreements tend to occur between groups and they lead people to see those with whom they disagree as irrational or immoral (or worse). Tis creates a sense of us versus them and this afects intergroup behavior. Deep disagreement, then, is not merely a conceptual-­ theoretical problem for a lofy philosophical ideal, but is implicated in actual social processes and conficts. Tere is an infuential view in political science that fts very well with what we have found about polarization here. It is called political realism and Jason Brennan’s Chapter 8 in this volume contains a good introduction to it. According to political realism, politics is chiefy about group alliances, political identities, and partisanship, rather than informed and thoughtful deliberation, preference-­formation, and voting. Although the view is not without critics, it makes sense of a wealth of empirical evidence about voter ignorance, voter behavior, and political participation. As Christopher Achen and Larry Bartels write:

24  See Hogg (2013) and Brown and Gaertner (2003) for overviews and pointers to the original studies. 25  Te study employed a similar design to test for racial bias and favoritism. Te results were broadly similar, but less pronounced for race. Tis suggests that political identity forges stronger ties than racial identity. Tis was confrmed by a more recent transnational study on the relative strengths of political partisanship versus other social identities (Westwood et al., 2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

238  Jeroen de Ridder voters choose political parties, frst and foremost, in order to align themselves with the appropriate coalition of social groups. Most citizens support a party not because they have carefully calculated that its policy positions are closest to their own, but rather because “their kind” of person belongs to that party. (Achen and Bartels, 2016, p. 307)

If politics is all about us versus them, polarization is to be expected. Commenting on the same large body of evidence that Achen and Bartels draw on, Ilya Somin likens voter behavior to that of sports fans: “political fans” derive enjoyment from rooting for their preferred parties, candidates, ideologies, and interest groups, while deriding the opposition. Tey may also derive satisfaction from having their preexisting views validated, and from a sense of afliation with a group of like-­minded people. (Somin, 2013, pp. 78–9)26

On this view, then, politically relevant deep disagreements are bound up with socio-­ political identities and partisanship, which in turn feed both cognitive and practical polarization.27

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

5.  Qualifcations and Clarifcations So far, I have presented the connections between deep disagreement and polarization as if they form a simple linear and mono-­causal story. People disagree deeply, which leads them to see each other as irrational and/or immoral, which, in turn, causes cognitive and practical polarization. Tis is a simplifcation. Complex social phenomena such as political polarization do not have a single cause, but are always part of a nexus of infuences and feedback loops. Te basic picture sketched above must be qualifed and complicated. First, the relations between deep disagreement, cognitive polarization, and practical polarization go both ways and can be mutually reinforcing. We saw how cognitive polarization reinforces itself by making people more confdent about their perspectives. Tis entrenches and deepens their disagreement. Practical polarization can also strengthen cognitive polarization. People tend to justify and rationalize their actions, so as to preserve a positive self-­image. When we treat other people badly, we come up with reasons why this was justifed. If we did not hire someone, she must have been unqualifed or otherwise undeserving. Second, deep disagreement is not the sole or even most important driver of polarization. Polarization is a complex social phenomenon and, as such, is afected and 26  Jason Brennan even compares voters to hooligans: Hooligans are the rabid sports fans of politics. . . . Tey have strong and largely fxed worldviews. . . . are overconfdent in themselves and what they know. Teir political opinions form part of their identity . . . Tey tend to despise people who disagree with them, holding that people with alternative worldviews are stupid, evil, selfsh, or, at best, deeply misguided. (Brennan, 2016, p. 5) 27 Mason (2018) and Klein (2020) also detail how contemporary American politics has become extremely polarized and identity-­driven.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Deep Disagreements and Political Polarization  239 moderated by a variety of individual, social, and historical factors and processes. If deep disagreement were the key factor behind polarization, the increased polarization over the past decades should refect a corresponding increase in the number or depth of disagreements. Tere is no clear evidence to support this. It is not as if people have found new topics to disagree deeply over or adopted more starkly diferent epistemic or moral principles. Tird, deep disagreement is neither sufcient nor necessary for polarization. As to the former, there are deep disagreements in science and the humanities, but we typically do not see anything like the degree of polarization there that we fnd around political divides. Philosophers are in deep disagreement about consequentialism and deontologism in ethics or about whether philosophy requires naturalized methodology, but they do not (typically) think less of each other or treat each other worse as a result. Only certain kinds of deep disagreements lead to polarization, such as those concerning emotionally charged issues that are central to people’s personal or social identity. Deep disagreement is not necessary for polarization either. People with the same e/m-­frameworks might still polarize, based on other factors that make up their social identities. Lifestyle preferences or allegiance to sports teams come to mind. Note that saying that deep disagreement is neither sufcient nor necessary for polarization is not to deny that deep disagreement is causally relevant to polarization. Rather, my proposal is that deep disagreement is one element of a complex whole of factors that together form an important (if perhaps not the only possible) cause of polarization.28 Fourth, some might take issue with the causal language I have been using. Robust support for causal claims, one might insist, can only come from empirical work that targets the connection between deep disagreement and polarization directly. Such work has not been done and I, as a philosopher, am not qualifed to do it. In response, I would say my causal claims nonetheless have considerable plausibility, because much of the empirical work on polarization that has been done lends itself very naturally to an interpretation in which deep disagreements are implicated. To repeat the key connections: disagreeing deeply with others means that you take their e/mframeworks to be (partially) wrong. Tis means you will evaluate them as not fully rational, not quite morally upstanding, or worse. Tis, in turn, puts you in a situation where the conditions for cognitive and practical polarization are satisfed. Taking these qualifcations into account, let me paint a more realistic picture of how deep disagreement is causally relevant to polarization. Most people do not spend lots of their time thinking long and hard about politics and morality, let alone epistemology. Instead, they go to work or school, socialize with neighbors, colleagues, and friends, watch TV, talk about the news, play sports, go to church, do volunteer work, etc. All of these things shape, and are shaped by, people’s social identities— their sense of belonging to a certain group, of being a certain kind of person. Much of the work in social psychology and political science discussed above suggests that people’s political beliefs, attitudes, and behavior are not supported by extensive 28  For those cognizant of the literature on causality, the idea here is supposed to be reminiscent of Mackie’s (1965) analysis of causes as, minimally, INUS conditions. See Strevens (2008, ch. 3) for an up-­to-­ date analysis of causal relevance along these lines.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

240  Jeroen de Ridder thinking and reasoning, but by their social identities. As Achen and Bartels write: “Citizens tend to adopt the views of the parties and groups they favor” (2016, p. 310). But if people’s social identities—and not deep disagreements—cause polarization, does that not speak against what I have been arguing? No, social identities and deep disagreements should not be opposed like this. Tey are not in competition. Rather, deep disagreements are a part of social identities, because social identities involve commitments to e/m-­principles. Not in the sense that they always come with fully developed e/m-­frameworks, but they typically involve prioritizations of moral values or epistemic principles determining how to think about issues and which experts or branches of science to trust.29 In many cases, then, having sharply diferent social identities entails disagreeing deeply over some issues. When people identify with a social group, they implicitly commit to certain e/mprinciples. Deep disagreements can thus emerge indirectly: not because people work out their own e/m-­frameworks and carefully think through the issues on their own accord, but because they adopt the e/m-­principles that are characteristic of their social identities, which lead them to disagree deeply with others. By way of example: the average Democrat will trust the broad consensus among climate scientists, because that is what Democrats do. Te average European populist voter distrusts what “elite politicians” claim about the European Union, because that is what nationalists do. As a result, they will see those who disagree with them not just as mistaken, but as relying on misguided “experts” or as prioritizing the wrong moral values or, in other words, as employing incorrect e/m-­principles. Teir opponents are not quite rational and do not have their moral priorities straight. Te average Democrat comes to see the average Republican as irrational or dumb, as morally confused or malignant. Te European populist sees social-­democrats or liberals in a similarly negative light. No need to take what they say very seriously or to engage respectfully. Te result is increasing cognitive and practical polarization, which leads to even stronger and more extreme social identities. So we end up with a feedback loop in which social identities imply deep disagreements, deep disagreements lead to increased polarization, and increased polarization to even more entrenched and radically opposing social identities.

6.  Conclusion: Sad News about Deep Disagreement I began this chapter by pointing out that there are extensive literatures in science and the humanities singing the praises of disagreement. By now, it should be clear that an exception must be made for deep disagreements. I have argued that, rather than promote self-­ correction, mutual learning, and other epistemic goods, deep disagreements are more likely bad news. When people are in a deep disagreement because they have diferent and conficting epistemic or moral frameworks, they will have trouble recognizing (some of) each other’s reasons as valid or good ones. Tis undermines the possibilities for rational, truth-­ seeking deliberation. Deep 29  Of course, social identities are not themselves active agents who come up with principles and frameworks. Tey ultimately come from individuals and groups who shape the characteristic attitudes and commitments that make up social identities.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Deep Disagreements and Political Polarization  241 disagreement is not merely a theoretical problem. Te empirical evidence shows that deep disagreements are implicated in a polarization feedback loop. Social identities ofen come with attachments to conficting e/m-­principles. As a result, people with diferent social identities will end up disagreeing deeply over clusters of issues. Tis leads to cognitive and practical polarization, because people in a deep disagreement about issues that are central to their identity will tend to see each other as less than fully rational, morally subpar, or worse. Tis, in turn, reinforces their social identities and entrenches their disagreements further.30

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Achen, C. and Bartels., L. 2016. Democracy for Realists. Princeton, NJ: Princeton University Press. Ahlstrom-Vij, K. 2012. “Why Deliberative Democracy Is (Still) Untenable.” Public Afairs Quarterly 26(3): 199–220. Ahlstrom-Vij, K. 2019. “Te Epistemic Benefts of Democracy: A Critical Assessment.” In M. Fricker, P. Graham, D. Henderson, N. Pedersen, and J. Wyatt. (eds.), Te Routledge Handbook of Social Epistemology. London: Routledge, pp. 406–14. Alston, W. P. 1993. Te Reliability of Sense Perception. Ithaca, NY: Cornell University Press. Anderson, E. 2006. “Te Epistemology of Democracy.” Episteme 3(1–2): 8–22. Baron, J. 2008. Tinking and Deciding. 4th ed. Cambridge: Cambridge University Press. Brennan, J. 2014. “How Smart Is Democracy? You Can’t Answer that Question a Priori.” Critical Review 26(1–2): 33–58. Brennan, J. 2016. Against Democracy. Princeton, NJ: Princeton University Press. Brennan, J. 2021. “Does Public Reason Liberalism Rest on a Mistake? Democracy’s Doxastic and Epistemic Problems.” In E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Brown, R. and S.  Gaertner (eds.) 2003. Blackwell Handbook of Social Psychology: Intergroup Processes. Malden, MA: Blackwell. Christensen, D. 2007. “Epistemology of Disagreement: Te Good News.” Philosophical Review 116(2): 187–217. Christensen, D. and J. Lackey (eds.). 2013. Te Epistemology of Disagreement: New Essays. New York: Oxford University Press. Cohen, J. 1986. “An Epistemic Conception of Democracy.” Ethics 91(1): 26–38. Estlund, D. 2008. Democratic Authority. Princeton, NJ: Princeton University Press. Fantl, J. 2018. Te Limitations of the Open Mind. New York: Oxford University Press. Feldman, R. 2006. “Epistemological Puzzles about Disagreement.” In S.  Hetherington (ed.), Epistemology Futures. Oxford: Oxford University Press, pp. 216–36. Feldman, R. and T. Warfeld (eds.) 2010. Disagreement. New York: Oxford University Press. 30  Tanks to Catarina Dutilh Novaes, Elizabeth Edenberg, Michael Hannon, Tirza Lagewaard, Rik Peels, Chris Ranalli, Emanuel Rutten, and René van Woudenberg for comments on an earlier version of this chapter. Tanks also to audiences at the Delf Philosophy Colloquium in February 2018, the European Epistemology Network Conference in June 2018 in Amsterdam, the Georgetown Political Epistemology workshop in October 2018, and the Political Epistemology Seminar at the Institute of Philosophy in London in December 2018 for helpful questions and discussion. Research for this chapter was made possible by a Vidi grant (276-­20-­024) from the Netherlands Research Council (NWO).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

242  Jeroen de Ridder Fogelin, R. 1985. “Te Logic of Deep Disagreements.” Informal Logic 7(1): 1–8. Goldman, A. 2010. “Epistemic Relativism and Reasonable Disagreement.” In R. Feldman and T. Warfeld (eds.), Disagreement. New York: Oxford University Press, pp. 187–215. Goodin, R. and C. List. 2001. “Epistemic Democracy: Generalizing the Condorcet Jury Teorem.” Journal of Political Philosophy 9(3): 277–306. Goodin, R. and K.  Spiekermann. 2018. An Epistemic Teory of Democracy. Oxford: Oxford University Press. Habermas, J. 1990 [1983]. Moral Consciousness and Communicative Action, translated by C. Lenhardt and S. W. Nicholsen. Cambridge, MA: MIT Press. Hewstone, M., M.  Rubin, and H.  Willis. 2002. “Intergroup Bias.” Annual Review of Psychology 53: 575–604. Hogg, M. 2013. “Intergroup Relations.” In J. DeLamater and A. Ward (eds.), Handbook of Social Psychology, Second edition. New York: Springer, pp. 533–61. Hong, L. and S. Page. 2004. “Groups of Diverse Problem Solvers Can Outperform Groups of High-Ability Problem Solvers.” Proceedings of the National Academy of Sciences 101(46): 16385–9. Houlou-Garcia, A. 2017. “Collective Wisdom, Diversity, and the Misuse of Mathematics.” Revue Française de Science Politique 67(5): 1–20. Iyengar, S. and S. Westwood. 2015. “Fear and Loathing Across Party Lines: New Evidence on Group Polarization.” American Journal of Political Science 59(3): 690–707. Kahan, D. 2013. “Ideology, Motivated Reasoning, and Cognitive Refection: An Experimental Study.” Judgment and Decision Making 8(4): 407–24. Kahan, D., H.  Jenkins-Smith, and D.  Braman. 2011. “Cultural Cognition of Scientifc Consensus.” Journal of Risk Research 14: 147–74. Kahan, D., E. Peters, E. Cantrell Dawson, and P. Slovic. 2017. “Motivated Numeracy and Enlightened Self-Government.” Behavioural Public Policy 1(1): 54–86. Kappel, K. 2012. “Te Problem of Deep Disagreement.” Discipline Filosofche 22(7): 7–25. Kappel, K. 2018. “Higher Order Evidence and Deep Disagreement.” Topoi online frst, https://doi.org/10.1007/s11245-018-9587-8. Kelly, T. 2005. “Te Epistemic Signifcance of Disagreement.” In J.  Hawthorne, and T.  Gendler. (eds.), Oxford Studies in Epistemology, Volume 1. New York: Oxford University Press, pp. 167–96. Kelly, T. 2008. “Disagreement, Dogmatism, and Belief Polarization.” Journal of Philosophy 105(10): 611–33. Klein, E. 2020. Why We’re Polarized. New York: Simon & Schuster. Lackey, J. 2013. “Disagreement and Belief Dependence: Why Numbers Matter.” In D.  Christensen and J.  Lackey, (eds.), Te Epistemology of Disagreement: New Essays. New York: Oxford University Press, pp. 243–68. Landemore, H. 2013. Democratic Reason. Princeton, NJ: Princeton University Press. Landemore, H. 2017. “Beyond the Fact of Disagreement? Te Epistemic Turn in Deliberative Democracy.” Social Epistemology 31(3): 277–95. Landemore, H. and J. Elster (eds.). 2012. Collective Wisdom: Mechanisms and Principles. Cambridge: Cambridge University Press. Lynch, M. J. 2010. “Epistemic Circularity and Epistemic Disagreement.” In A. Haddock, A.  Millar, and D.  Pritchard. (eds.), Social Epistemology. Oxford: Oxford University Press, pp. 262–77.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Deep Disagreements and Political Polarization  243 Lynch, M.  J. 2021. “Political Disagreement, Arrogance, and the Pursuit of Truth.” In E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Mackie, J. L. 1965. “Causes and Conditions.” American Philosophical Quarterly 12: 245–65. Mason, L. 2018. Uncivil Agreement: How Politics Became Our Identity. Chicago: University of Chicago Press. Mendelberg, T. 2002. “Te Deliberative Citizen: Teory and Evidence.” In M.  Delli Carpini, L. Huddy, and R. Y. Shapiro (eds.). Research in Micropolitics, Volume 6: Political Decision Making, Deliberation and Participation. Amsterdam: Elsevier, pp. 151–93. Mill, J.  S. 1977 (1859). Essays on Politics and Society. Collected Works Vol. 18. Ed. J. M. Robson. London: Kegan & Paul. Nickerson, R. 1998. “Confrmation Bias: A Ubiquitous Phenomenon in Many Guises.” Review of General Psychology 2(2): 175–220. Ober, J. 2010. Democracy and Knowledge. Princeton, NJ: Princeton University Press. Page, S. 2007. Te Diference. Princeton, NJ: Princeton University Press. Popper, K. 1945. Te Open Society and Its Enemies, 2 Volumes. London: Routledge. Ranalli, C. B. 2018a. “Deep Disagreement and Hinge Epistemology.” Synthese online frst, https://doi.org/10.1007/s11229-018-01956-2. Ranalli, C.  B. 2018b. “What is Deep Disagreement?” Topoi online frst, https://doi. org/10.1007/s11245-018-9600-2. Rawls, J. 1996. Political Liberalism. Rev. edn. New York: Columbia University Press. Rini, R. 2017. “Fake News and Partisan Epistemology.” Kennedy Institute of Ethics Journal 27(2S): 43–64. Ryfe, D. 2005. “Does Deliberative Democracy Work?” Annual Review of Political Science 8: 49–71. Somin, I., 2013. Democracy and Political Ignorance. Stanford, CA: Stanford University Press. Somin, I., 2016. Democracy and Political Ignorance: Why Smaller Government Is Smarter. Stanford, CA: Stanford University Press. Strevens, M. 2008. Depth: An Account of Scientifc Explanation. Cambridge, MA: Harvard University Press. Sunstein, C. 2006. Infotopia: How Many Minds Produce Knowledge. Oxford: Oxford University Press. Sunstein, C. 2017. #Republic: Divided Democracy in the Age of Social Media. Princeton, NJ: Princeton University Press. Surowiecki, J. 2004. Te Wisdom of Crowds. New York: Doubleday. Talbott, W. 2016. “Bayesian Epistemology.” In E. N. Zalta, (ed.), Te Stanford Encyclopedia of Philosophy, Winter 2016 edn., https://plato.stanford.edu/archives/win2016/entries/ epistemology-bayesian/. Talisse, R.  B. 2021. “Problems of Polarization.” In E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Van Fraassen, B. C. 2002. Te Empirical Stance. New Haven, CT: Yale University Press. Westwood, S., S. Iyengar, S. Walgrave, R. Leonisio, L. Miller, and O. Strijbis. 2018. “Te Tie that Divides: Cross-National Evidence of the Primacy of Partyism.” European Journal of Political Research 57(2): 333–54.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

13

Political Disagreement, Arrogance, and the Pursuit of Truth Michael P. Lynch

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.  Introduction In March 2020, during the height of the 2020 coronavirus pandemic, it was reported that Dr. Anthony Fauci, the respected director of the National Institute of Allergy and Infectious Diseases, was being targeted by far-­Right conspiracy theories alleging that his briefngs were intended to undermine the president. As one Twitter user put it, “Sorry liberals, but we don’t trust Dr. Anthony Fauci.”1 Te pursuit of truth and accurate information is always important in a democracy. For one thing, the public and government both need accurate information in order to make informed decisions—whether in the voting booth or on the foor of the Senate. But in times of crisis—in war, or during a global pandemic—the need for what Hannah Arendt called “factual truth” becomes urgent. So what happens when we disagree over who the experts are—when we disagree over who really knows? In extreme circumstances, when the “fog of war” is thick, some such disagreements are naturally expected. Te available information is ofen incomplete or even inconsistent, making it more difcult to judge who or what can be trusted—even when one is simply motivated to fnd the facts. But it can also happen that our political diferences can afect our epistemic diferences, and in particular our political attitudes and convictions can alter our epistemic attitudes in ways that can undermine a society’s ability to respond to the crisis at hand. Americans’ reactions to the COVID-­19 pandemic serve as a painful illustration of this fact. During the early days of the epidemic in the U.S. and even as the rate of infection was spiking across the country, polls were suggesting that a person’s political views were predictive of how seriously they perceived the public health risk, with followers of President Trump, at least initially, more likely to think the problem was overblown.2 As the epidemic spread across the country, these diferences, unsurprisingly, became less pronounced among most citizens. But as the aforementioned distrust of Dr. Fauci by some in the far-­Right suggests, Americans continue to be polarized in a diferent way. Even as most people (on both the Right and the Lef) came to regard the pandemic as real, diferences in political attitudes continued to

1  As quoted in Te New York Times (Alba and Frenkel, 2020). 2  See the discussion on Gallup (Newport, 2020). Michael P. Lynch, Political Disagreement, Arrogance, and the Pursuit of Truth. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Michael P. Lynch. DOI: 10.1093/oso/9780192893338.003.0014

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Political Disagreement, Arrogance, and the Pursuit of Truth  245 impact our epistemic commitments—in particular, they continued to shape which experts and information sources we trust. As recent work on polarization and the above example both suggest, the phenomenon comes in diferent forms.3 Tus political groups can be polarized over issues and/or attitudes. To use Robert Talisse’s way of putting it in Chapter 11 in this volume, groups are platform polarized, to the degree that members of each group disagree with each other about the issue—for example over whether COVID is a hoax or not. Some research casts doubt on whether platform polarization really has worsened as dramatically as ofen assumed.4 While clearly there is signifcant disagreement between the Right and the Lef on any of number things, it also appears that ordinary Americans continue to agree on many concrete policies. Yet that same research indicates that we are polarized in a very diferent way—in our attitudes towards those with diferent political viewpoints. And as the COVID crisis drives home, this kind of polarization—what is sometimes called afective polarization— does seem to be increasing: there is an increasing distance between our perceptions of, and attitudes toward, the other party. Note that afective polarization itself can have at least two diferent aspects. One aspect—and the aspect that grants it the label “afective”—is that groups are polarized in their emotional states. Te have strong emotional reactions to each other, feelings of contempt and disgust for example. But another aspect is cognitive. Even as we may share more beliefs than we think, members of distinct political groups can regard each other with deep suspicion, as less rational, intelligent, and trustworthy. Cognitions of this sort will typically be accompanied by afective states, but what they are attitudes about are the capacities and reliability of other epistemic agents. As a result, using Jeroen De Ridder’s language, we can call this (aspect of afective polarization) cognitive polarization—with Americans’ polarized attitudes towards Fauci acting as a prime example.5 In this chapter, I explore two contributing factors to cognitive polarization. Te frst is what I will call epistemic disagreement—or disagreement over what is known, who knows it, or how we know. Crucially, I will argue, even the perception that such disagreement is widespread—whether or not it actually is—can be dangerous. Te second factor is intellectual arrogance. Tis is arrogance about what we know or think we know; it is the kind of arrogance that tells whites they have nothing to learn about racism from people of color and that reassures those who believe they know more about infectious diseases than those who spend their lives studying them. As I will try to argue here, these two factors can be mutually reinforcing. Tat makes them doubly dangerous, because by increasing cognitive polarization, they in turn undermine the democratic value of the pursuit of truth. Democracies are rightly thought to have a special interest in the pursuit of truth. As I have argued elsewhere, a practical way of understanding that is to see it as a concern with reliable social epistemic practices (Lynch,  2021). Social-­epistemic practices are those procedures, methods and sources that help us acquire true beliefs 3  Two recent examples are Talisse’s helpful distinctions in his contribution to this volume); see also Mutz’s important work, e.g. Mutz 2018. 4  See, in particular, Pew Research Center, 2016, 2017. 5  See Chapter 12 in this volume.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

246  Michael P. Lynch

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

via interaction with other epistemic agents. Tese are the kinds of practices we aspire to employ, for example, in scientifc, historical, educational journalistic and legal institutions. We aim to use such practices when teaching students, engaging in archival research, investigating a crime, replicating an experiment, employing blind-­review, and independently confrming a source. It is a central value of democracy that we encourage, protect, and fairly distribute access to such practices and the institutions that employ them. We do so, for example, by providing primary education to all citizens, protecting free assembly, speech, a free press, and the norms of academic freedom. In democracies, we might say, the means by which we pursue truth via reliable social-­epistemic practices are a primary social good; protecting and providing access to that good is therefore a central democratic aim. A commonly cited reason for why democracies value the pursuit of truth in this way is that the widespread use of such practices leads to a more informed public. Tis broadly Millian argument has its merits; but it clearly has its limits as well, since it speaks to the efcacy of the practices and not to their fair distribution and promotion—the more central democratic value. Tus a better explanation is that promoting and providing fair access to reliable social-­ epistemic practices is demanded by the democratic ideals of reason-­giving and deliberation. Democracies, at least in theory, should be spaces of reasons where decisions are made without recourse to violence and where the exchange of reasons is encouraged. In this chapter, I will argue that the democratic value of the pursuit of truth is undermined by the cognitive polarization encouraged by epistemic disagreement and group intellectual arrogance. Tat is because the ideal is undermined when we take ourselves to no longer agree on how we should pursue the truth; and it is undermined when we think we no longer need to pursue it because we know it already.

2.  Epistemic Disagreement and Its Perception Typically, we think of political disagreement as being disagreement over either principles or policies. Of course, real political disagreements—such as disagreements about vaccination or immigration—are combinations of both. Tey concern not only what is right but how best to implement what is right. Moreover, the rightness here is typically moral and political rightness. In the last several decades, however, there has been increasing attention paid to the pervasiveness of disagreements over basic facts in contemporary democracies. Many current disputes between the Lef and the Right in the U.S., for example, seem to involve disagreements over what is true—that is, whether climate change is “a hoax” or whether a former president was even a citizen. Disagreements over facts ofen go up the epistemic ladder. In other words, those involved may begin to question not only whether the other side has their facts straight but do so partly on the basis that the other side’s views are based on unreliable social-­epistemic practices. When debates get to this point, they are rightly called epistemic disagreements—because they concern overtly epistemic matters.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Political Disagreement, Arrogance, and the Pursuit of Truth  247

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

It is not difcult to see why epistemic disagreements are problematic for democracies, especially in times of crisis. You cannot determine what to do in the face of the facts if you disagree over what the facts are. And you cannot agree on the facts if you do not agree on which epistemic and social-­epistemic practices are reliable. Moreover, epistemic disagreements, I believe, play a crucial and ofen overlooked part in cognitive polarization—even if, as De Ridder points out, cognitive polarization itself is a multi-­faceted phenomenon that involves both epistemic, political, and moral factors (De Ridder, Chapter  12 this volume). Tat is because while many people expect our moral views to be wrapped up in our political views, we ofen fool ourselves into thinking that things are diferent when it comes to our epistemic views. One way to formally characterize an epistemic disagreement is to describe it as a disagreement over epistemic principles. By an epistemic principle, I mean a normative principle to the efect that some epistemic practice has some valuable epistemic status, such as reliability.6 I will say that A disagrees with B over some epistemic principle EP just when A does not believe EP and B does. So one disagrees with someone over an EP in this sense when one either disbelieves the EP or withholds belief in it (e.g. when one doubts that it is true). A overtly disagrees with B over some EP just when A explicitly withholds assent from an EP B asserts. Te paradigm case of overt epistemic disagreement is where assent is withheld because the relevant EP is simply denied. An overt epistemic disagreement is mutual just when both sides to the dispute deny an epistemic principle the other asserts. Epistemic disagreements, like disagreements of all types, can be more or less deep. Elsewhere (Lynch, 2010) I have argued that the hallmarks of deep disagreement are as follows: 1. Competition: If the parties afrm distinct principles with regard to a given domain, those principles (a) pronounce diferent practices to be the most reliable in a given domain; and (b) these practices produce incompatible beliefs about that domain.

Any epistemic disagreement involves competitive principles—although the degree of their competition might vary—for example, diferent practices in a chemical laboratory might produce incompatible results in only a few cases, where some religious and scientifc practices might produce very diferent results across a range of cases. 2. Nonarbitration: Tere is no further epistemic principle, accepted by both parties, which would settle the disagreement.

If there is a third shared principle that would settle the matter, the dispute is intuitively not very deep—since both parties have at their disposal the means by which they could resolve it. 6  I have defended the value of true belief elsewhere; see Lynch (2004, 2020, 2021).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

248  Michael P. Lynch 3. Commonality: Te parties to the disagreement share common epistemic goal(s).

Te idea here is that both parties aim, or at least see themselves as aiming, at the same goal—in most cases, truth. Should one party be aiming at truth and another at believing whatever is the most politically expedient, it is not clear they would be disagreeing about epistemic principles—although there may be an implicit, meta-­ disagreement over the proper aim of inquiry.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4. Circularity: Te epistemic principle(s) in question can be justifed only by means of an epistemically circular argument—that is, by an argument whose premises can be shown to be justifed only if its conclusion is justifed. 7

Te Circularity condition is the hallmark of the deepest epistemic disagreements. Its obtaining it is the sign that the disagreement has reached bedrock—it concerns principles so fundamental that they cannot be justifed except by appeal to themselves—or a set of principles that cannot be justifed except by appeal to one another. Deep epistemic disagreements have long been regarded as a serious philosophical challenge. As the Pyrrhonean skeptics were arguably the frst to note, the challenge becomes more difcult the more basic the practices in question happen to be. When it comes to very basic practices, like induction, then as Hume argued, or perception (as Alston,  1986 famously argued) it is unclear how I can rationally defend those practices without at some point relying on them and assuming them to be reliable. When we try to defend our most basic practices for determining what to believe, it seems like, in Wittgenstein’s words, our spade is turned on bedrock. Reasons run out. One can accept this philosophical point, however, and still wonder how serious a problem epistemic disagreement really is. Te question might seem to depend on how deep our disagreements actually are. And by the above criteria, one might think, most epistemic disagreement in our political life will not be super deep. In most cases where my sources of belief formation are questioned, Nonarbitration, and hence Circularity fail—as might happen when one journalist questions another about the reliability of a given source of information, say a website, and the dispute is settled by appeal to other more basic sources, such as scientifc experts. In other cases, subsequent investigation may reveal that the sources in question do not really compete—perhaps because one is more restricted in application than another. Afer all, you might think, most people still implicitly seem to accept standard medical and engineering practices—they go to the doctor, trust engineers to build safe bridges, rely on computer science to construct their digital devices, and so on. Perhaps, one might hope, most real epistemic disagreements are not that deep. Of course, the fact that truly deep epistemic disagreements are rare does not mean they cannot or do not exist in political culture. Tey arguably do, as I have argued elsewhere (Lynch, 2012). But more importantly, that question is entirely independent of the real political concern with epistemic disagreement. Tat is because theoretically 7  Tis is only a crude summary of epistemic circularity; see Lynch and Silva (2016) and Alston (1986) for a fuller discussion.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Political Disagreement, Arrogance, and the Pursuit of Truth  249 shallow disagreements can nonetheless function as deep disagreements relative to certain contexts of information. Or to put it more plainly, opposing political groups ofen perceive each other to be in more or less deep epistemic disagreement—even when those disagreements might not be as deep as they believe. To see this, let us think about what it means to perceive yourself to be in deep epistemic disagreement. According to the above analysis of the depth of disagreement, you perceive yourself as being in a deep epistemic disagreement to the degree that you accept the following: (a) that you are disagreeing with someone over the reliability of some epistemic practice; (b) that the other party believes their practices to be producing truth; (c) that you do not share anything in common that could resolve the debate; and (d) that the practices in question are basic and the principles involved are fundamental. Note that the acceptance of (a)–(d) is a matter of degree (both because one might accept some rather than others but also because one’s acceptance itself is a matter of degree). And acceptance can also be implicit and dispositional and not overt and conscious. In any event, in the sense intended here, one perceives oneself to be in deep epistemic disagreement with another to the extent that one (implicitly or otherwise) accepts (a)–(d). Te point I wish to stress here—and one I wish I had stressed in earlier work on the topic—is that you could accept all of these things without their being true. As we just noted, many epistemic disagreements, like our journalist example above, will be relatively shallow—but it does not follow from that fact that either party will recognize that fact. Far from it. When human beings disagree—especially over matters connected to politics—passions tend to run high. And when passions run high, we can easily ignore facts—including the fact that we might share something in common with those with whom we are disagreeing—that we would have accepted in a cooler moment of refection. And it seems plausible to think that many of our disputes over whether to trust various sources and practices for producing beliefs are like this. In the heat of political battle, we see adherents of Fox News and Breitbart as using a closed system of sources none of which can be justifed but by appeal to one another. And that may well be true—or it may not, depending on the details of the case. But that just illustrates the point I am making: that it is the perception (accurate or not) that matters here, for the simple reason that the perception of deep epistemic disagreement is by itself sufcient to contribute to cognitive polarization. And that means that from the standpoint of democratic politics, the perception of epistemic disagreement can be almost as bad as real disagreement. First, the fact that political parties perceive themselves to epistemically disagree is consistent with the recent research on cognitive polarization mentioned earlier. Data based on surveys collects information about what people report about a situation. What people report is (in the typical case) taken by the researchers as evidence of how they perceive that situation. So research on afective polarization—which ofen concerns people’s perceptions of their political opponents—is in part based on evidence which researchers take to indicate how people perceive their disagreements with others, whether or not those perceptions track reality.8 8  It is of course possible that reports do not report perceptions and that if they do, the perceptions are not accurate; both points are consistent with the claims made here.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

250  Michael P. Lynch If that is right, then it holds for epistemic disagreement as well. Many Americans on the Right perceive democrats as pushing fawed or politically biased data in service of a “a climate change agenda.” Many on the Lef perceive those on the Right as relying on non-­scientifc, religious, and simply unreliable practices. In short, whether or not members of political parties actually disagree over epistemic principles, they perceive each other to do so. And since groups are cognitively polarized when they perceive each other as unreliable and uninformed, the perception of epistemic disagreement leads to, or even could be described as a kind of cognitive polarization. Crucially, however, these polarizing perceptions are not formed in a political vacuum. Epistemic disagreement, like moral disagreement, generally arises in a political context. Our political convictions, as I called them at the outset, can feed the perception of epistemic disagreement. A striking illustration of this fact are “epistemic spillovers.” An epistemic spillover occurs when a “shared political conviction infuences people’s desire to consult and to use people’s views on a task that is entirely unrelated to politics”(Marks et al., 2019, p. 83). In one study, for example, participants were able to learn both about the political orientation of other participants, and their competency at a given unrelated non-­political task (categorizing shapes). Ten they were asked who they would consult to aid them in categorizing the shapes themselves. Te results were striking: people “consult and are infuenced by the judgments of those with shared political convictions even when they had observed evidence suggesting that those with diferent convictions are far more likely to ofer the right answer.”9 Tus, people are more likely to trust those of the same political tribe even on non-­ political matters. Moreover, they will continue to do so even when they have evidence that undermines their faith in political cohorts and even when they have fnancial incentives to follow that evidence. Put another way, democrats are more likely to trust the advice and testimony of democratic doctors, democratic plumbers, and democratic accountants than republican ones—even when they have evidence to think that will lead to worse results. Te upshot here is that cognitive polarization—particularly that related to ­epistemic disagreement—brings mistrust in its wake. Naturally, that mistrust will extend to those the other side takes as experts. If you think that democrats are brainwashed slaves of false ideologies, you are not going to trust their supposed experts when they tell you that there is a pandemic, or that the earth is warming or that the seas are rising as a result. And that, in turn, can place a strain on democratic decision procedures. Tis is especially true if the distrust is rampant on both sides. For it means that there will be little chance of agreement on the facts, not to mention rational discussion about them—even in times of crisis. Tis points to the wider impact of cognitive polarization and epistemic disagreement. Tat wider impact is on the core democratic value of the pursuit of truth itself. Tat value, I have suggested, consists in democracies having a special interest in reliable social-­epistemic practices and the institutions that employ them. But mutual 9  Marks et al., 2019, p. 83.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Political Disagreement, Arrogance, and the Pursuit of Truth  251 mistrust in the other party’s experts ofen encourages cynicism about expertise itself. Amongst some, it may engender the idea that there really are no experts on say, the climate, or viruses (because it is perceived, incorrectly, that the experts “all disagree”) and hence that maybe there are no reliable practices for pursuing the truth about it. Tat in turn can threaten a society’s commitment to protecting and fairly distributing access to reliable social epistemic practices. When people come to perceive (even mistakenly) that no one really knows, for example, how viruses are transmitted, or what scientifc practices help us to determine whether it is safe or not to return to work, then the idea that we should value the institution of science itself will become less compelling.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Arrogance My argument so far has been that part of the story of cognitive polarization is that widespread political disagreement tends to go hand in hand with disagreement—real or perceived—over which social-­epistemic practices are reliable. And that in turn has a number of bad efects on democratic politics and core values. But the story so far has a missing element. To see it, refect for a minute on the ancient pyrrhonists. In their view, becoming aware of epistemic disagreement was supposed to make us less dogmatic. Tey hoped it would cause us to suspend our beliefs and so achieve a more peaceful life. But for most people, that does not seem to be what happens at all. Political opponents who regard each other as employing unreliable sources or practices for forming beliefs are not prone to become more refective about their own practices. Tat is even the case when people become cynical about expertise in general. Such folks do not give up their opinions about vaccines or climate change and so on. On the contrary, they form social media chat groups to tell each other how much smarter they are then everyone else. So one might think the pyrrhonists were good philosophers but lousy psychologists. But that raises a question. Why do people not respond in the way that the pyrrhonists hoped we would? Why are we not more prone to suspend our beliefs? Any explanation for this sort of phenomenon must be multi-­faceted. But I suggest that part of the story must concern our own psychological attitudes towards our own and others’ beliefs and the political/cultural factors which give rise to those attitudes. Attitudes, in the sense of that term I have in mind, are positively/negatively valanced motivational states. Such states typically involve commitments of various sorts. Epistemic attitudes involve distinctively epistemic commitments. By an epistemic commitment, I mean a commitment either to an explicitly epistemic principle, or—as is the case with most epistemic attitudes—a commitment to going about believing in a certain way. One example, arguably, is open-­mindedness: to the extent that I have this attitude, to that extent I am willing and committed to considering relevant intellectual alternatives.10

10  For discussion of open and closed-­mindedness, see Battaly, 2018a and 2018b.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

252  Michael P. Lynch Open-­mindedness is typically a positive epistemic attitude in a certain obvious sense. Not all epistemic attitudes are typically positive, however. One of the more unsettling attitudes, and unsettlingly common, is what is sometimes called intellectual arrogance. Intellectual arrogance is the psycho-­social attitude that you have nothing to learn from anyone else about some subject or subjects because you know it all already. Tis is the arrogance of the know-­it-­all. And it is this attitude, I suggest, that acts as psychological grease on the rails of epistemic disagreement— which motivates human beings to see such disagreement as a warning to be more cautious in their claims to knowledge, but as a reason to dogmatically dig further in to their own position. Intellectual arrogance isn’t simply about misplaced overconfdence; it involves a self-­delusion about that confdence’s basis. While the intellectually arrogant think their felt superiority is due to their knowledge—which they may well have, since the know-­it-­all may indeed know a-­lot—that is not the basis of their arrogance. Teir arrogance is more likely compensation for insecurity due to a perceived threat.11 Psychologically speaking, arrogance is a combination of insecurity and a felt sense of superiority. At the limit, the intellectually arrogant person will confuse ego with truth: he may even implicitly commit to the irrational idea that whatever he thinks is true because it is he who thinks it. Intellectual arrogance is clearly an individual problem; but it poses a distinct political problem when it becomes partisan—or indexed to a group.12 Tis is what happens when a group becomes convinced that “we” know and “they” do not. Such group or partisan arrogance is typically directed at specifc groups and the sources of information that are associated with those groups. As a result, someone who is arrogant toward African Americans and Latino immigrants will dismiss sources perceived by them to be friendly to those groups—for example, CNN and the New York Times. When convictions that are central to the group’s identity become part of its shared cultural narrative, it is more likely for the members of the group to be intellectually arrogant about them, simply because any threat to them threatens the group identity. It is perceived as a threat to “who we are.” As a result, such convictions become immune from revision by members of the tribe and protected at all costs from counterevidence. Tis unwillingness, when expressed as a form of widely shared group arrogance, results in what is sometimes called active ignorance—a systematic and coordinated refusal to acknowledge evidence.13 Moreover, as Elizabeth Anderson argues in her own contribution to this volume, this unwillingness to update beliefs in the light of evidence is a hallmark of epistemic bubbles.14 Te most politically damaging, and widespread form of group intellectual arrogance is white arrogance. White arrogance has fueled whites’ resistance to any attempt to explain the meaning behind #Black lives Matter, or the meaning of Colin Kaepernick’s kneeling to protest police brutality and anti-­black violence in our society. White arrogance is what fueled the widespread conspiracy, led by Donald Trump, 11  Tis point was frst made by Alessandra Tanesini’s important 2016a. 12  For a fuller discussion, see Lynch, 2019. 13  See for example, Medina (2013). For an earlier account see Gordon (1995). 14  See Anderson, Chapter 1, this volume.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Political Disagreement, Arrogance, and the Pursuit of Truth  253 that President Barack Obama was not an American citizen. In all of these cases, evidence presented by people of color and their allies was ignored or dismissed or undermined. Experiences with anti-­ black violence were written of as either exaggerated or rare. In all cases, white arrogance involves the idea that white experience is representative, that white experience should be treated with default trust, while experiences that run contrary to it should be marginalized and undermined.15 It is not accidental that historically, white arrogance has been both reinforced and used by certain kinds of political ideologies. It has always been in the interest of white authoritarian leaders to encourage a toxic mix of two feelings in their followers. Te frst is fear—to convince them that there is a threat that “I alone (the leader) can fx.” Tis threat is typically racialized and it is portrayed as urgent (“Caravans are coming!”). Te second feeling, ofen bolstered by a stirring narrative, is one of felt superiority—our race, our way of doing things—is both epistemically and morally superior. Put these two feelings together and you have the psychological base for arrogance—the mixture of feelings of threat and superiority that supports the attitude that “we” know, “they” do not, and therefore we deserve and they do not. Te far Right has practically made white arrogance an article of faith; but increasingly it has extracted from that an over-­all ideology of arrogance. It puts that ideology to use not only in explicit discussions of race as mentioned above, but also in disagreements concerning climate change and the coronavirus. Tink back to the examples we started with: the conspiracy theories concerning the COVID-­ 19 epidemic. Tese theories range from claims that it is a hoax, that Bill Gates created it, that the Chinese pulse it into bodies through 5G networks, or that it can be treated by various vitamins or by injecting disinfectant. As ofen happens where group arrogance is concerned, such claims are ofen made from a position of defensiveness— accompanied by accusations that it is the scientifc experts that are arrogant, that “they” are the ones who are not listening to reason. Tis is all weird enough—afer all, this is an actual global disaster and one would hope our responses to it would be data-­driven. But the advocacy of similar wild theories by political leaders—including at points the President himself—indicates they are not one-­of throwaways. Such conspiracies are driven by ideologies of arrogance.16

4.  A Dangerous Feedback Loop What I take the above considerations to suggest is that authoritarian ideologies coupled with the psychological attitudes like white arrogance only deepen perceived disagreements and encourage dogmatic responses to then. As Andrew Aberdein remarks, that may not be surprising, since

15  Readers may notice a parallel between these remarks and Toole’s argument (Chapter 5, this volume) that white supremacy is a harmful epistemic system. See also Dotson (2011). 16  I draw here from Cassam’s (2019) study of conspiracy theories.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

254  Michael P. Lynch any disagreement with an arrogant individual is more than likely to feel like deep disagreement, even in cases where there is an easily accessible resolution. Te phenomenology of disagreeing with an arrogant person and of being in deep disagreement may be similarly frustrating, even though their ultimate cause is quite diferent.  (Aberdein, 2020, pp. 47–8)

Aberdein’s point here is that the frst-­person experience of disagreeing with an arrogant individual is akin to the experience of deep disagreement—the feeling that one has simply hit rock bottom and reasons have given out. In both cases, there will be a feeling that rational persuasion is of the table, even if “the origin of the arrogant arguer’s unwillingness to back down and indiference to the assertions of others lies in his character” as opposed to the nature of the epistemic principles in question (2020, pp. 47–8). Importantly, however, Aberdein recognizes that arrogance can, in an intuitive sense, make disagreement worse, in the sense of making it even harder to rationally resolve. As he put it:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

the arrogant individual is a risk factor for some of the most promising strategies for resolution of deep disagreements. Specifcally, it is precisely the behaviour of the arrogant arguer that gives rise to Wittgenstein’s worries about the missionaries and the natives  (Aberdein, 2020, p. 48).

Wittgenstein’s worries, to which Aberdein here alludes, are refected in his remark that “where two principles really do meet which cannot be reconciled with one another, then each man declares the other a fool and a heretic. . . . At the end of reasons comes persuasion. (Tink of what happens when missionaries convert natives).” (Wittgenstein,  1969, §§611–612). Te idea is that intellectual arrogance makes people less likely to use rational persuasion to resolve disagreements and instead try to simply convert others—perhaps (like many missionaries of old), by way of force. Tis is an important point. And it holds at the group level as well, and arguably with even more force. When political groups or parties disagree, and individuals are tribally, not just individually intellectually arrogant, the chances that the disagreement will be perceived as deeper than it is, and the chances that rational resolution will be pursued, are lessened. One reason for this is that when confronted with someone in the grip of an arrogant ideology, it can seem pointless to try and engage in rational discussion. Consider, for example, disagreements that sprang up in the U.S. in the spring of 2020 over the reliability of scientifc models that predicted that without social distancing and lockdowns, deaths from COVID-­19 would increase. People who dismissed the scientifc evidence, who “didn’t trust Fauci” and believed him to be in the grip of a political ideology, were themselves most likely displaying group arrogance. Tey were the ones in the grip of ideology. Te prospect of arguing with someone like this, who might even think that COVID-­19 is a hoax, struck many as exhausting if not pointless. Tey would not listen, so the only way forward might seem to be to simply punish those who violated lockdown orders. Tere was neither the time, nor one might think, the possibility of rational persuasion. Likewise, people of color

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Political Disagreement, Arrogance, and the Pursuit of Truth  255

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

confronting white arrogance are not apt to see further “discussion” as particularly helpful. It is already clear that whites, as a group, turn a blind eye to, if not willingly engage in, anti-­black violence. Again, when confronting the arrogant, rational discussion seems moot. But of course, the point works from the other way around as well. Te arrogant will likewise see their political opponents as similarly not worth the time. More precisely, those who have fallen into thinking that they already know what is true, that they have it all fgured out and that their view on some matter cannot improve, are not prone to pursue further truth. Why pursue further truth if you know it already? At the very least, intellectual arrogance, by its very nature, directs us away from the sincere give and take of reasons. Tat is because arrogance blinds us to evidence that does not already confrm our views and inclines us to confuse ego with truth. Tere is a dangerous kind of feedback loop here. As just argued, arrogance on either side of a disagreement can lead people to think that disagreement is deeper than it may actually be. But crucially, the perception (accurate or not) that we deeply disagree with one another can also cause us to be perceived as arrogant. Tere are several reasons for this. One is that if the A’s see the B’s as hopelessly reliant on faulty sources and practices, then the A’s will be more inclined to be dismissive of their reasons. But that very dismissiveness will cause them to seem arrogant to the B’s—even if the A’s are correct about the B’s reasons. Moreover, if a member of the B’s then challenges some member of the A’s to defend a basic social-­ epistemic practice that they (the A’s) adhere to, it may be difcult to (legitimately) ofer a reason for the reliability of that practice that does not presuppose its reliability already. (Tis can happen, for example, when one is pushed to defend the reliability of scientifc modes of reasoning—it is difcult to do so without relying on them). At that point, B will likely perceive A as dogmatic, and possibly even arrogant. And so it goes: arrogance deepens perceived disagreement and that perceived disagreement only amplifes arrogance.

5. Conclusion It is an eternal truth that in politics, perception counts almost as much as reality. Our discussions have provided yet another illustration of this truth. Te perception of arrogance, just like the perception of epistemic disagreement can be damaging to democracy. How damaging may depend on whether factional arrogance is increasing in our society. Afer all, one might think, human beings are naturally biased tribal, arrogant creatures, and that has not changed. You will not get an argument from me on that point. But it is also important to recognize that like any psycho-­social attitude, factional arrogance can be infectious. Like fear or hatred, it can spread throughout populations when conditions are right. And while it is a question I have lef to one side here, I do not think it is difcult to see that certain conditions currently operative in our culture are conducive to the spread of factional arrogance. As I have argued elsewhere (Lynch, 2019) two factors

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

256  Michael P. Lynch

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

seem particularly salient. One is the rise of racist, authoritarian, and populist political ideologies across the globe.17 Another is the tendency for social media to harden our political convictions and increase cognitive polarization.18 Tere is much to be understood about how both of these conditions contribute to cognitive polarization and how they impact democratic politics generally. But I think it is clear enough already that the damage from these forces can be considerable. I have noted that democracies have a vested interest in protecting and promoting reliable social-­epistemic practices. Tey have that interest because, among other things, they have an interest not only in an informed public but a public that engages with each other in ways that convey mutual respect. Such respect need not be respect for the rationality of each other’s views. Nothing about democratic politics requires us to see our political opponents as knowledgeable or wise. But, democracy does demand that we treat them as not just moral but as epistemic agents—as possible knowers and possible participants in the give and take of reasons.19 As a result, democracies have a vested interest in protecting and promoting practices that allow people to acquire and exchange reasons and knowledge. Yet when a democracy fnds itself awash with both partisan arrogance and epistemic disagreement—real or perceived—its citizens will fnd it more difcult to show such respect for each other. And that in turn, can lead the fraying of the very institutions by which we pursue the truth. Tat is because when a society no longer seems to agree on how best to pursue the truth, and yet many people remain convinced they have it in their grasp, then it may seem not tragic, but useful, that reasons have run out. It gives them—and us—an excuse to give up on the hard work of reason-­giving. Tat is the lasting poison of intellectual arrogance. Tose who are arrogant lack the motivation to participate in the give and take of reasons; and their opponents rightly perceive them as being immune from learning anything new. Meaningful public discourse grounds to a halt; and we take a step closer to dealing with our political opponents in the way that fools and heretics have always been dealt with, not by way of reason, but by way of the sword.20

References Aberdein, A. 2020. “Arrogance and Deep Disagreement.” In M. P. Lynch and A. Tanesini (eds.), Polarisation, Arrogance, and Dogmatism: Philosophical Perspectives. London: Routledge, pp. 39–52. Alba, D. and S. Frenkel. 2020. “Medical Expert Who Corrects Trump is Now a Target of the Far Right.” Te New York Times, March 28, 2020. 17  For commentary, see Anderson (Chapter 1, this volume) and Stanley (2018). 18  See Rini, 2017, Sunstein, 2017, and Lynch, 2019 for discussion. 19 For more on the relevance of mutual respect, see Elizabeth Edenberg’s excellent Chapter  14 in this volume. 20  I wish to thank Elizabeth Edenberg, Michael Hannon, and my fellow participants at the Georgetown conference on political epistemology, as well as audiences at Yale University, University of Connecticut, and Princeton University for comments and criticism.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Political Disagreement, Arrogance, and the Pursuit of Truth  257 Alston, W. 1986. “Epistemic Circularity.” Philosophy and Phenomenological Research. 47: 1–30. Ballantyne, Nathan. 2019. “Epistemic Trespassing.” Mind 128(510): 367–95. Battaly, Heather. 2018a. “Can Closed-Mindedness Be an Intellectual Vice?” Royal Institute of Philosophy Supplement 84: 23–45. Battaly, Heather. 2018b. “Closed-Mindedness and Dogmatism.” Episteme 15: 261–82. Cassam, Quassim. 2019. Conspiracy Teories. London: Polity. Dotson, Kristie. 2011. “Tracking Epistemic Violence, Tracking Practices of Silencing.” Hypatia 26: 236–57. Gordon, L.  R. 1995. Bad Faith and Antiblack Racism. Atlantic Highlands, NJ: Humanity Books. Lynch, M. P. 2004. True to Life: Why Truth Matters. Cambridge, MA: MIT Press. Lynch, M.  P. 2010. “Epistemic Circularity and Epistemic Incommensurability.” In A.  Haddock, A.  Millar, and D.  Pritchard (eds.), Social Epistemology. Oxford: Oxford University Press, pp. 262–77. Lynch, M.  P. 2012. In Praise of Reason: Why Rationality Matters for Democracy. Cambridge, MA: MIT Press. Lynch, M. P. 2013. “Epistemic Commitments, Epistemic Agency and Practical Reasons.” Philosophical Issues 23/1: 343–62. Lynch, M.  P. 2019. Know-it-All Society: Truth and Arrogance in Political Culture. New York: Norton. Lynch, M. P. 2020. “Polarisation and the Problem of Spreading Arrogance.” In M. P. Lynch and A.  Tanesini (eds.), Polarisation, Arrogance, and Dogmatism: Philosophical Perspectives. London: Routledge. Lynch, M. P. 2021. “Truth as a Democratic Value.” Forthcoming. NOMOS. Lynch, M. and P.  Silva. 2016. “Why Worry about Epistemic Circularity?” Journal of Philosophical Research 41: 33–52. Marks, J., E.  Copland, E.  Loh, C.  Sunstein, and T.  Sharot. 2019. “Epistemic Spillovers: Learning Others’ Political Views Reduces the Ability to Assess and Use Teir Expertise in Nonpolitical Domains.” Cognition 188: 74–84. Medina, J. 2012. Te Epistemology of Resistance: Gender and Racial Oppression, Epistemic Injustice, and the Social Imagination. New York: Oxford University Press. Mutz, D. C. 2018. “Status Treat, Not Economic Hardship, Explains the 2016 Presidential Vote.” Proceedings of the National Academy of Sciences of the USA 115(19): E4330–9. Newport, F. 2020. “Te Partisan Gap in Views of the Coronavirus.” Gallup, May 15, 2020. Pew Research Center. 2009. “Public Praises Science; Scientists Fault Public, Media.” July 9. www.people-press.org/2009/07/09/public-praises-science-scientists-fault-publicmedia. Pew Research Center. 2016. “Partisan and Political Animosity in 2016.” June 22. www. people-press.org/2016/06/22/partisanship-and-political-animosity-in-2016. Pew Research Center. 2017. “Te Partisan Divide on Political Values Grows Even Wider.” October 5. www.people-press.org/2017/10/05/the-partisan-divide-on-political-valuesgrows-even-wider. Plato. 1992. Te Republic. Translated by G. M. A. Grube and C. D. C. Reeve. Indianapolis: Hackett.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

258  Michael P. Lynch

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Rini, R. 2017. “Fake News and Partisan Epistemology.” Kennedy Institute of Ethics Journal 27: 43–64. Stanley, J. 2018. How Fascism Works: Te Politics of Us and Tem. New York: Random House. Sunstein, C. 2017. #Republic: Divided America in the Age of Social Media. Princeton, NJ: Princeton University Press. Talisse, R. 2019. Overdoing Democracy. Oxford: New York. Tanesini, A. 2016a. “‘Calm Down, Dear’: Intellectual Arrogance, Silencing and Ignorance.” Aristotelian Society Supplementary volume 90(1): 71–92. Tanesini, A. 2016b. “Intellectual Humility as an Attitude.” Philosophy and Phenomenological Research 96(2): 399–420. Wemple, E. 2019. “Here’s Why Fox News Is Number 1.” Te Washington Post, May 15. www.washingtonpost.com/opinions/2019/05/15/heres-why-fox-news-is-no/?utm_ term=.ef5865fcb0bb Whitcomb, D., H. Battaly, J. Baher, and D. Howard-Snyder. 2017. “Intellectual Humility: Owning Our Limitations.” Philosophy and Phenomenological Research 94: 509–39. Wittgenstein, L. 1969. Philosophical Investigations. Cambridge: Basil Blackwell.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

14

Te Problem with Disagreement on Social Media Moral not Epistemic

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Elizabeth Edenberg

Contemporary political discourse ofen feels like a battleground between diametrically opposed worldviews. Today our intractable disagreements have surfaced in nasty ways, persistently threatening to fracture the very ground upon which we build political community. Complicating this broader disagreement on questions of morality are new challenges to agreement on basic facts about our world (Kappel, 2017, 2018; Sinnott-­Armstrong, 2018). Much of this division is fueled by the new ways we access information, namely through digital means increasingly tailored to show us what we want to see. Social media allows us to sort ourselves into increasingly likeminded groups, who consume information from diferent sources, and end up in polarized and insular echo chambers of our own making. In response, many have called for new social media literacy campaigns, fagging questionable news sources, increased fact checking, and other ways to correct inaccuracies that spread online (Crowell, 2017; Mosseri, 2017; Breakstone et al., 2019; Tugend,  2020). Tese approaches attempt to bridge political divides by improving our epistemic capacity to responsibly assess information online. Philosophers have also begun to weigh in on this issue, diagnosing the deeper epistemic failures that underlie intractable political disagreements and suggesting ways to correct them (Lynch, 2016, 2019; Rini, 2017; Kappel, 2017; Cassam, 2019). What these approaches have in common is the idea that both the problem and the solution are grounded in epistemology: political disagreement online is an epistemic failure in need of an epistemic solution. If only we could shore up individuals’ epistemic capacities and our collective epistemic resources, we could return to agreeing on the facts, if not the values, relevant to politics. I will argue that this epistemic approach does not show us the complete picture and thus misses a more basic form of moral respect required for reestablishing a common ground for cooperation across disagreements in a divided society. Eforts rooted in improving epistemic responsibility cannot solve our polarization problem1 because they misdiagnose it. To mend our divides and reestablish a common sense of political community, we must fnd some basis for cooperating with others with 1  See Rob Talisse’s Chapter 11 and Jeroen de Ridder’s Chapter 12 in this volume for discussion of many distinct problems of polarization. Diferent aspects of the problem are likely to call for diferent remedies. Elizabeth Edenberg, The Problem with Disagreement on Social Media: Moral not Epistemic. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Elizabeth Edenberg. DOI: 10.1093/oso/9780192893338.003.0015

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

260  Elizabeth Edenberg whom we disagree. However, cooperation must be grounded in something other than epistemology. Solving our epistemic woes will do little to bridge the divides in our society without a more basic form of moral respect for our fellow citizens. I argue that the breakdown of discourse online provides renewed reasons to draw out not an epistemic but a moral basis for political cooperation among diverse citizens—one inspired by Rawlsian political liberalism. Rather than viewing politics as a battleground between factions, we must fnd ways to cultivate mutual respect for our fellow citizens in order to reestablish common moral ground for political debate. Te argument will proceed in fve parts. I frst outline the key challenges social media poses for political disagreements (Section 1). To solve these challenges, many argue for various ways of cultivating more responsible epistemic agency (Section 2). Yet these epistemic proposals face challenges of their own. I outline three initial problems with proposed solutions grounded in epistemic responsibility (Section 3) before moving on to show why epistemic solutions are actually secondary to moral concerns (Section  4). Approaching the problems of political disagreement in primarily epistemic terms takes us of course and ofers solutions likely to backfre in our current polarized environment unless they are grounded in a more fundamental moral respect for our fellow citizens. In closing (Section 5), I argue that we should fnd ways to restructure social media to facilitate respectful rather than divisive interactions.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.  Te Challenge Te intractable disagreements that plague contemporary political discourse seem to be driven by a breakdown in shared agreement on basic facts. Social media is a major contributing force to this new epistemic landscape. Consider the warped informational landscape that is our social media newsfeeds. Newsfeeds blend information from a remarkable number of sources: personal updates, journalistic pieces, self-­published articles, paid posts presented as news “articles,” intentional disinformation, and propaganda campaigns, etc. Once we consider the immense variety of inputs—coupled with no editorial oversight to ensure that information shared is objective, accurate, and well-­researched—it is no surprise that individuals are coming to many diferent conclusions on a host of factual matters. In a healthy epistemic environment, people can count on others to alert them to new information that brings to light their factual errors. But social media is not a healthy epistemic environment; it produces epistemic bubbles2 that repeatedly champion certain views while deriding or simply excluding others. Accordingly, social media users are far less likely to confront dissenting challenges to their views. Moreover, the repeated sharing of views itself amplifes a person’s likelihood to 2  According to Nguyen, epistemic bubbles are social epistemic structures in which other relevant voices have been lef out. He argues that the term echo chamber should be reserved for cases in which relevant voices have been actively excluded and discredited (Nguyen, 2020). Anderson (Chapter 1, this volume) uses the term epistemic bubble for a self-­segregated network for the circulation of ideas that is resistant to correcting false beliefs.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Problem with Disagreement on Social Media  261 believe them. Even if we are unsure of the truth of an initial post, once it has spread and gained sufcient likes, comments, and engagement by others who treat it as true, we are much more likely to believe the content (Rini,  2017, pp. 48–9). Tis latter phenomenon plays on our natural tendencies to rely on others for vetting the information we come across. Cass Sunstein explains, “if one person sees that fve, ten, a hundred, or a thousand people are inclined to say or do something, there is a tendency to think that each has made an independent decision to say or do it” (Sunstein,  2017, p. 99). And if those posting on our social media newsfeeds are trusted friends or colleagues, we tend to accord more credibility to their independent stamps of content approval. Finally, conformity pressures make us even likelier to believe (or at least express) certain views more than others. Whether or not someone believes a particular story shared on social media, they tend to go along with the crowd signaling their endorsement of the content, suppress their own reservations, and refrain from sharing information that contradicts the dominant position (Kuran,  1997; Muchnik et al.,  2013; Sunstein,  2017, pp. 100–1). Together, these efects can pollute social media’s epistemic environment and undercut rational independent evaluation of the information we encounter. Even those not on social media are not entirely insulated from these efects. Traditional news sources report on “trending” stories, and public opinion polls incorporate analysis of social media trends and popularity as one proxy for the overall popularity of and opinions about certain policies or political fgures. Furthermore, in recent elections, a strong social media presence has fueled the popularity and support of political fgures across the spectrum, as seen in the Obama, Trump, and Sanders campaigns. Ironically, then, the platforms touted as tools to bring the world closer together, build community, and give people the power to share information,3 are also driving us farther apart. As Eli Pariser explains, “democracy requires a reliance on shared facts; instead, we’re being ofered parallel but separate universes” (Pariser, 2011, p. 5). Divisions within our democracy run deep and without common ground, we lack a foundation for bridging these divides.

2.  Te Epistemic Response What is a responsible democratic citizenry to do? Underlying the push to improve our ability to spot fake news and vet sources for their accuracy is the common belief that the root of the problem is epistemic. Quassim Cassam, for example, argues that political harms are a downstream “consequence” of the epistemic harms rooted in epistemic vice (Cassam, 2019, p. viii), which he views as personal intellectual failings that “systematically obstruct the gaining, keeping, or sharing of knowledge” (Cassam, 2019, p. 23). Relatedly, Regina Rini argues that fake news spreads because of the bent testimonial norms on social media (Rini,  2017). And Michael Lynch argues that epistemological explanations are the foundational cause of our current 3  To rif on the mission statements of Facebook and Twitter.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

262  Elizabeth Edenberg political controversies. He explains, “how we act is the result of how we think—what we believe and therefore think we know” (Lynch, 2019, p. 5). An epistemic diagnosis seems to suggest an epistemic solution: improve individuals’ epistemic capacities, and the epistemic crisis will improve. Various proposals have been ofered to improve our ability to discern the truth and form responsible beliefs on the basis of good evidence, including media literacy campaigns, fxing the best testimonial norms on social media by improving accountability, and cultivating intellectual humility to combat tribal arrogance. I briefy elaborate on each below. Te frst type of proposal to remedy our epistemic crisis seeks to improve individuals’ capacity to spot “fake news” and encourage more refective scrutiny of the information we encounter. Social media platforms, political pundits, and educational institutions have all adopted some version of this epistemic response to the current challenges. Facebook’s evolving response to the 2016 U.S. election relies heavily on individual Facebook users and third-­party organizations to report false news stories, which may then show up lower on newsfeeds. Facebook hesitates to play the role of arbitrator of truth, a position they have doubled down on recently in controversies over false and misleading claims in political advertisements (Zuckerberg,  2019).4 Instead, they maintain that it should be up to individuals to evaluate sources to determine their accuracy. To aid in this efort, Facebook has helped sponsor third-­party eforts, including the New Integrity Initiative aimed at “helping people make informed judgments about the news they read and share online” (Mosseri, 2017).5 Likewise, Twitter’s response to the 2016 election touted the importance of their original design for allowing individuals to challenge misleading claims. “Twitter’s open and real-­time nature,” is itself “a powerful antidote to the spreading of all types of false information” (Crowell, 2017).6 Like Facebook, Twitter emphasized the importance of engaged citizens correcting and challenging claims, holding that the company “should not be the arbiter of truth” (Crowell,  2017). Individuals should each evaluate the epistemic quality of information they encounter through social media platforms to decide for themselves what they should believe. Educational institutions have also approached the issue of polarization through an epistemic lens focused on improving students’ epistemic capacities through media literacy training. While media literacy education is not new, it has expanded its reach since the 2016 election (Breakstone et al., 2019; Tugend, 2020).7 Te core idea behind media literacy education is for students to learn how to critically evaluate online content. Lessons involve learning to question information sources, the motivation 4  As this chapter was going to press, the covid-­19 crisis hit. In response Facebook took an unusual (for them) stance on the truth and have diverged from their hands-­of approach by prioritizing information shared by the World Health Organization and Center for Disease Control. However, they continue to afrm their hands-­of stance towards political information and campaigns, allowing false and misleading information to spread. 5  At the time of this post, Adam Mosseri was the Vice President in charge of Facebook’s News Feed. 6  Colin Crowell was Twitter’s VP of Public Policy, Government, and Philanthropy when he posted this clarifcation of Twitter’s response to misinformation on Twitter’s blog. 7 For more information on current educational eforts, Stony Brook University’s Center for News Literacy maintains a list of peer organizations who are involved in media literacy projects in addition to providing their own resources for educators (Stony Brook, 2020).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Problem with Disagreement on Social Media  263 behind sharing stories, the broader context of the story shared, and whether it is possible for two apparently contradictory claims to be true. Tese eforts are epistemic in focus, aiming to improve individuals’ critical capacities to evaluate information. Following good epistemic hygiene, students learn to take a more refective and critical stance towards forming their beliefs on the basis of good evidence and learn to seek out sufcient justifcation before claiming knowledge. Te range of proposals targeting individual epistemic failings may fnd some philosophical support in Cassam’s work on epistemic vice. Cassam argues that individual epistemic vices have important downstream political consequences and “the onus is therefore on us as individuals to minimize the harms done by our epistemic vices” (Cassam,  2019, p. 187). Te types of remedies proposed by social media companies and educational institutions that seek to improve individuals’ epistemic habits to cultivate more responsible epistemic behavior when seeking information are ultimately encouraging us to try to correct our epistemic vices and cultivate responsible epistemic agency. While these eforts are admirable and can certainly help improve individuals’ epistemic capacities, I will argue that they will not do much to bring our divided democracy together. Te matter is more complicated than simply learning to spot and avoid spreading false claims. Partisan politics plays a big role in which false stories individuals believe. People are particularly susceptible to believing false claims presented as true by members of their political party. Philosophers are split on whether this partisanship in testimonial belief is epistemically virtuous (Rini, 2017) or vicious (Cassam, 2019; Lynch,  2019), and thus are also divided on how best to correct the problems partisanship causes. A second type of proposal for correcting the epistemic problems created by sharing political information on social media, particularly fake news, seeks to improve individual accountability for sharing information in order to correct the bent testimonial norms on social media. Regina Rini convincingly argues that the root of the epistemic problems generated by social media lie in the bent testimonial norms of social media rather than individual epistemic vices. Norms of communication on social media are unstable and disputed, which challenge traditional mechanisms for accessing the veracity of testimony acquired on these platforms (Rini, 2017, pp. 46–9). Typically, individuals assess both the content of the statement and the credibility of the speaker to ascertain whether or not to rely on her testimony (Coady, 1992; Lackey, 2008). Yet social media sharing warps these norms because it is ofen unclear whether or not someone intends to assert a claim is true when they share, like, or retweet content. In this warped landscape, Rini argues that it can actually be reasonable for individuals to assign greater credibility to their co-­partisans— even when this leads to sharing or believing fake news (Rini 2017, pp. 50–4). When forming beliefs on the basis of testimony, we should look for which testifers we can trust. We rationally assign greater credence to co-­partisans because we judge them to have a good track record of getting normative questions right—we see them as our epistemic peers in normative domains (Rini, 2017, pp. 51–2). Terefore, to try to fx the epistemic problems generated by social media, we need to ensure that the information shared tracks the truth so that we can build common ground on politically relevant facts.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

264  Elizabeth Edenberg Clarifying the testimonial norms of social media requires more than individuals gaining media literacy (as many social media companies and educational institutions suggest). Instead, we need to correct the bent testimonial norms by establishing broader social accountability mechanisms for sharing information that we refectively endorse. Rini proposes a solution that pairs individual accountability with institutional mechanisms for tracking the testimonial reputation of social media users in order to foster a broader norm of accountability for epistemically responsible social media testimony (Rini, 2017, p. 57). Structural changes are needed to support our eforts to hold one another accountable for responsible epistemic practices because it is nearly impossible for individuals to track the sources of information they encounter through social media. Afer all, an individual’s social media “friends” consist of an expansive list of people linked in networks with vastly many other people. However, social media platforms can readily track individuals’ testimonial reputations and Rini argues that doing so could provide the necessary infrastructure that would allow people to evaluate the trustworthiness of their fellow citizens. Tis would promote more individual epistemic responsibility for evaluating the truth of stories before sharing them with others. A third type of proposal argues that the root of our problems is “tribal arrogance” that is fueled by social media and therefore in response we should cultivate the epistemic virtue of intellectual humility. Michael Lynch identifes the cause of our uncivil political disagreements and polarization as “tribal arrogance,” which is not only an individual but a social epistemic vice (Lynch, 2019 and Chapter  13, this volume). Tribal arrogance relies on group divisions pitting “us” against “them” in hierarchical epistemic relations in which we become convinced of our group’s superior knowledge (2019, pp. 6, 25–6). Our partisan identities become bound up with our convictions, setting the background against which we interpret both factual disputes and questions of right and wrong.8 Unfortunately, once a factual dispute is transformed into a partisan conviction, we are far more likely to engage in identity-­ protective reasoning when faced with outside challenges to our beliefs because they also challenge our core values and personal identity (Lynch, 2019, pp. 66–71). As we become resistant to challenges, we become increasingly arrogant and confdent in our group’s superior knowledge. Social media fuels tribal arrogance by reinforcing shared convictions through highly charged emotional content. All of this make us vulnerable to exploitation of our tribal identities to fuel hate and divisions within our democracy, rather than building common ground. To correct these problems, Lynch argues we must cultivate the epistemic virtue of intellectual humility, which requires an openness to improving one’s worldview by incorporating new evidence and the experience of others (Lynch, 2019, p. 149). We should learn to accept the limits of our own knowledge and fnd ways to secure the types of institutional changes that can support our common pursuit of the truth. We also need to protect the institutional structures that support intellectual humility by providing evidence and the means to exchange our reasons with one another, for 8  Our convictions express who we are, what we value, and how we want to be seen by others—impacting not just what we do but also what we believe (Lynch, 2019, pp. 55–7, 60–2).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Problem with Disagreement on Social Media  265 example a free press, science education, and peaceful dissent (Lynch, 2019, pp. 161–3). However, at bottom, in order to be able to learn from one another, we need to respect one another as epistemic agents. Lynch argues that “when we own what we don’t know and remain open to what others do”—in short, when we are intellectually humble—“we exemplify a basic respect for our fellow citizens that is demanded by democracy” (Lynch, 2019, p. 170, see also pp. 155–6). Ultimately, each of these proposals highlights an important facet of the epistemic challenges that impact politics. Ofen individuals are lax in their evaluation of the information they encounter, social media sharing warps testimonial norms in ways that make it difcult to assess the information we encounter, and the political groups with whom we identify have a signifcant impact on what and whom we believe. Yet while looking at diferent aspects of the epistemic landscape, all three responses share a common tactic: they identify the root problem as epistemic and argue for epistemological solutions. It is this tactic, I will argue, that ultimately renders each proposal insufcient to the task at hand.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Te Inadequacy of Strictly Epistemic Approaches Media literacy campaigns, improved testimonial norms on social media, and cultivating intellectual humility are all important for mending internet age democracy. But eforts rooted in epistemic responsibility (even if supported by institutional change) approach the problem with too narrow a scope that leads them to miss the broader picture. I will argue that epistemic approaches face three distinct, but related challenges. First, our divisions are not only a result of individually irresponsible epistemic behaviors. Second, solutions grounded in individual epistemic responsibility fail to account for the social and institutional structures in which we form our beliefs. Tird, focusing on epistemic elements of our political divisions fails to account for important structural features embedded in the design of social media that contribute to society’s pernicious divisions. First, what ofen seems like a failure of individual epistemic responsibility (the target of many popular solutions to our current crisis) may not be. For instance, some polarization in society is not a result of epistemic capacities misfring,9 but rather a product of individual epistemic capacities functioning well.10 Te group dynamics of deliberation amongst likeminded people make it likely that when individuals’ epistemic capacities are functioning well, they end up adopting more extreme views or claim more confdence in their initial position. But this can be a sign of epistemic responsibility. Afer all, confdence in our beliefs should increase as we encounter additional supporting evidence for our views. And in a group of 9  Sometimes polarization is the result of our epistemic capacities functioning poorly, whether due to epistemic vices, cognitive biases (both implicit and explicit), or simply social pressures to conform. 10 Sunstein argues that reasonable argumentative practices partially explain polarization (Sunstein, 2002, pp. 179–80, 2017, pp. 71–2). Recently, some philosophers have begun defending polarization as epistemically reasonable (see, e.g., Singer et al., 2019) while others have outlined features of group-­ level rationality (Hedden, 2019) that could also be useful in explaining reasons why polarization is actually reasonable.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

266  Elizabeth Edenberg likeminded individuals, we are likely to encounter new evidence that supports our views and may therefore justifably come away with more confdence in our initial position. On its own, becoming more confdent in one’s views is not a pernicious epistemic tendency. It only becomes so when the arguments and evidence we encounter are not representative of the range of reasonable positions about the issues at stake or we increase our confdence in our views independent of their evidentiary support. To put it diferently, one of the main problems lies in the insularity of our ep­i­ste­ mic bubbles, not individual failures to sort truth from falsity. To fx this problem, it matters less that we improve our epistemic capacities themselves and more that we pay attention to the groups within which we deliberate. For instance, knowledge about the impact of public policies is distributed asymmetrically across society (Anderson,  2010, p. 98). To attain a broader view of the information relevant to public policies, we need to explicitly consult people who occupy diferent social positions and seek out their knowledge and expertise. We can learn from those who have a diferent view of the social world in order to correct for our blind spots, fll in gaps in our knowledge, and destabilize the tendency to view one’s own perspective as authoritative for all.11 Doing so is, in some ways, a call for the kind of intellectual humility Lynch seeks to address our partisan woes (Lynch, 2019 and Chapter  13, this volume). Tis points to the second challenge to individualistic models of epistemic responsibility: the problems with our political disagreements have as much to do with our social structures as they do with individual epistemic vices or failures of responsible testimony. Unlike solutions focused primarily on individual epistemic agents, Rini and Lynch’s proposals are both attentive to the social dynamics of belief formation on social media.12 Expanding our focus from internal belief forming processes to the broader social networks within which we form beliefs is important for understanding the more complex picture of epistemic agency.13 Claims are not only evaluated on their merits. As social creatures, we look to cues from those within our community about the plausibility of adopting (or expressing) certain beliefs. Accordingly, proposals that seek to bolster an individual’s epistemic abilities only address one 11  Feminists have called attention to the way in which one’s social location provides a unique perspective that infuences both what is known and what is recognized as knowledge (see, e.g. Collins,  1989; Anderson,  1995; Harding,  2004; Fricker,  2007; Mills,  2007; Medina,  2013; and Dotson,  2014). Tese insights are politically signifcant and add epistemic arguments in support of forms of government that leverage the insights of people diferently situated in structural relations of power (see, e.g., Young, 2000; Anderson, 2010). 12  I do think both proposals ultimately come back to encouraging individual epistemic responsibility even when they take into account the social process of belief formation. 13  To understand the changes in the epistemology of our political culture, Elizabeth Anderson argues that we should expand our focus from individual epistemic agents towards a group cognition model (Anderson, Chapter 1, this volume). Notably group cognition involves a shared commitment to expressing the belief together and agreeing to use it as a premise in the group’s reasoning. Individuals within the group need not believe the assertion, provided they commit to using it in their group’s reasoning. Te expression of the group’s beliefs can function as identity-­expressive speech acts, which can signal party loyalty and identity rather than reporting on individuals’ personal beliefs. Similarly, Jason Brennan and Michael Hannon both champion an in-­group cheerleading model of political discourse, treating political claims as expressive instances of team sports (Brennan, 2016; Hannon, Chapter 16, this volume).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Problem with Disagreement on Social Media  267 piece of the puzzle14—equally important is attending to the way information spreads across a network of individuals. As Cailin O’Connor and James Weatherall argue, “we cannot understand changes in our political situation by focusing only on individuals. We also need to understand how our networks of social interaction have changed, and why those changes have afected our ability, as a group, to form reliable beliefs” (O’Connor and Weatherall, 2019, p. 16). In what ways have our networks of social interaction changed such that focusing on individual epistemic powers falls short? One major change is the introduction of a new medium of interaction—social media. Whether using social media to follow journalists, public fgures, and breaking news stories or to engage in debate on the platform with one’s friends, followers, or the general public—for its users, the platforms helps facilitate public debate. In addition, social media is a major locus of political advertisement and a medium through which politicians connect with their constituents.15 Tis brings us to the third challenge facing epistemic responsibility models: solutions grounded in epistemic responsibility fail to address the non-­epistemic factors embedded in the design of social media that contribute in pernicious ways to society’s divisions. So many of our political activities are now mediated through algorithms that structure online engagement. Regardless of an individual’s purpose for using social media, they will be impacted by the algorithmic design. How these algorithms are optimized impacts what we see on our newsfeeds, which posts go viral, and who has the most infuence on the platforms. Te design of the algorithms powering social media impacts individuals’ beliefs; however, the problems created by social media cannot be corrected by encouraging individuals (or groups of people) to be more epistemically responsible when engaging on these platforms. Te design of the systems will need to change in ways that target the root causes of the divisions we see, which will require looking beyond a strictly epistemic approach. Social media is not a neutral medium. Te algorithms that power social media sites are optimized for specifc tasks: typically, some combination of engagement (measured by views, clicks, reactions, shares, comments, etc.), growth, and time spent on the platform.16 Te best way to keep people engaged is to ensure a personalized experience that shows users content they like. Tis is done by machine learning 14  Bolstering individuals’ epistemic capacities is the solution that is most commonly cited in popular discussions of the problem, including the social media companies and educational institutions (see Section 2). 15  Facebook, for example, draws on the vast information it has about its users to help political campaigns efectively target their message (Vaidhyanathan, 2018, pp. 161–3). 16  Tese metrics are important because the business models underlying social media rely on selling targeted access to the people who use the platform to their customers (e.g. companies, political candidates) who seek to advertise on the platform. Te more all of us stay engaged with the platforms, the more ads we see and, more importantly, the more information they can collect about us to refne their models. Te better the models know us, the more precise the targeting of personalized ads, which then leads to increased revenue for the social media companies. Te data are complex and varied, so machine learning builds mathematical models based on analyzing historical data of past use to predict future engagement, and this model is dynamic, continually updating itself based on how we and those who are like us interact with the platform. For an accessible picture of how algorithms and machine learning works, see Broussard, 2018 and Kearns and Roth, 2020. For a discussion of their detrimental efects on society, see O’Neill 2016 and Eubanks 2017.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

268  Elizabeth Edenberg models optimized to do just that. As these models learn what we interact with most, they strive to give us more of that content, which results in a splintered information landscape. To make matters worse, within that already splintered landscape, algorithms continue to select posts that tend towards divisive content. As noted by Facebook’s own internal review of its role in polarization, “our algorithms exploit the human brain’s attraction to divisiveness,” promoting divisive content “in an efort to gain user attention and increase time on the platform” (Horwitz and Seetharaman, 2020). In their various eforts to combat fake news, polarization, foreign infuence, clickbait content, and inauthentic users, Facebook’s teams found that “bad behavior came disproportionately from a small pool of hyperpartisan users” (Horwitz and Seetharaman,  2020). Yet the engagement-­based metrics promote content from its most active users, who tend to be more partisan and more likely to behave suspiciously.17 Te problem is not unique to Facebook; all social media platforms are powered by algorithmic recommendations that prioritize highly viewed, engaging content. Tese algorithms can be exploited by those who seek to garner more attention for their political positions.18 A simple change to cease promoting the content of the most active users would indirectly reduce divisiveness without impeding freedom of speech, while also signifcantly reducing the reach of those who seek to manipulate and exploit the system. Key to understanding our polarization problem is not only the breakdown of shared sources of information, bent testimony, or arrogance—as stressed by the epistemic responsibility model—but a targeted fueling of antipathy and extremist content. In short, no matter what we do, the major platforms mediating our inquiry are not neutral. Understanding how algorithms work to drive engagement on online platforms thus gives us another lens for understanding why there has been so much fuel added to the fre of our social divisions. Understood at the level of algorithmic design, it becomes obvious that the issue extends beyond what individuals can do (e.g. better curated newsfeeds) or what groups can do (e.g. more responsible sharing practices or combatting tribal arrogance). Nor will the problem be solved if individuals delete or avoid social media, as even ofine news relies heavily on social media, trends, and stories.19 We need structural change to address these problems. Any proposal to redesign the systems through which we communicate should ensure that the changes will address the root of the problems. However, our problems 17  For example, posting for 20 hours a day or engaging in spam-­like behavior, suggesting either manipulation by people actively working a single account in shifs or bots Horwitz and Seetharaman, 2020). For more on the weaponization of social media, see Regina Rini’s Chapter 2 in this volume. 18  Te targeted advertisement mode of political campaigning tends to highlight the issues that most divide the public along party lines—ignoring the many areas of agreement, consensus, and shared sentiments that bring people together into a community with one another. Politicians frequently emphasize the issues that divide them from their political opponent to strengthen their base. As many have noted, partisan identity is ofen strengthened by demonizing other groups and encouraging “us”/“them” mentalities that lean into divisions between groups (see, e.g., Mason, 2018). 19  I am not calling for a return to a time without social media—rather, we should understand the ways the systems work in order fgure out the appropriate systemic solutions. Tese platforms have helped us to connect with each other from afar, and their promise for this use is especially salient as the world fghts the coronavirus and is forced to move most of their social interactions online.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Problem with Disagreement on Social Media  269 are not strictly epistemic, thus solutions that focus only on features related to epistemic responsibility will be inefective. Solutions grounded in epistemic responsibility misfre in two ways: they are overly individualist and they view the foundational problem to be epistemic. A number of popular solutions proposed by social media companies, popular media, and educational institutions target individual epistemic responsibility. As we have seen, the individualistic focus is too narrow. An efective solution needs to grapple with the broader social infuences on belief formation and the ways the structural design of social media platforms impacts individuals’ beliefs. Proposals, like Rini’s and Lynch’s, which take a broader view of the social-­epistemic problems and the design of social media still err in their recommendations, which ultimately bottom out at individual epistemic responsibility. For example, Rini proposes institutional changes to track the testimonial reputation of individual users in order to encourage a norm of individual accountability for responsible sharing practices. Likewise, Lynch diagnoses the problem as one of tribal arrogance fueled by social media; however, his proposal to cultivate intellectual humility is ultimately a proposal for what each person can do to become more responsible and virtuous epistemic agents. Finally, even if we were to make our ep­i­ ste­mic recommendations even more social, they would still be insufcient since the root of our problems is not strictly epistemic. So long as social media maintains its current algorithmic design, no amount of improved epistemic responsibility (for either individuals or groups of people) will solve our current problems. In what follows, I will argue that the solution is moral, not epistemic. Tat is, we need to cultivate moral respect for one another and fnd ways to restructure social media to facilitate morally respectful rather than divisive interactions.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

4.  Why Epistemic Responsibility Is Secondary to Moral Concerns I have raised several reasons to be skeptical about the epistemic diagnosis of political divisions and the proposed epistemic solutions to repair our divided democracy. In this section, I will show that the epistemic responsibility model will only work if built on a strong foundation of moral respect across diferences. Epistemic proposals capture one piece of the puzzle, but they are secondary to the moral concerns that must be addressed frst in order for any epistemic solution to be efective. First, without addressing the deeper moral concerns, epistemic solutions may ultimately backfre because epistemic respect is subject to assessments of credibility that is also subject to partisan divisions. Second, epistemic approaches tend to approach the question of disagreement as a problem to be resolved, rather than a persistent feature of our society that we need to learn to manage. Tird, focusing on the epistemology of disagreement keeps the focus on what divides us, rather than what unites us. First, in order to work, epistemic responsibility models require cultivating epistemic respect for those who hold diferent views. Both Rini’s proposal to track the testimonial reputation of individual social media users and Lynch’s proposal to cultivate intellectual humility rely on assessing the epistemic standing of ourselves and our fellow citizens. According to these views, we need to regard our fellow citizens as reliable testifers and to enter the discussion with a basic respect for our

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

270  Elizabeth Edenberg interlocutors’ standing as epistemic agents. However, since epistemic respect is subject to assessments of the credentials and track record of our fellow citizens, it is not stable enough to serve as the foundational respect needed for democracy.20 Appraising an interlocutor’s epistemic credentials is rightly subject to ongoing assessments of their track record and performance, which is likely to backfre in a polarized society. Even if two citizens initially enter a discussion with a mutually respectful epistemic stance, the discussion itself ofen provides new evidence about whether one’s interlocutor is trustworthy, competent, and reliable. If it becomes clear that one interlocutor is untrustworthy, fails to see the force of the evidence, denies logical conclusions, or is otherwise prone to faulty reasoning, the other interlocutor now has good epistemic reasons to doubt their testimony. Unfortunately in our polarized society, the tendency to doubt our interlocutors’ testimony is all too common—and all too commonly divided along partisan lines. We trust those who share our beliefs and view others with great skepticism. In contemporary political debates, we ofen see people claiming the “other side” is a dupe of media—they follow the wrong sources of evidence, are liable to deception and manipulation, and fail to be responsive to reasons. Even if we hold one another accountable for responsible epistemic practices, we are currently in a situation where the standards for facts, evidence, and expertise are frequently as contested as the particular claims we are trying to evaluate.21 If we lack a common standard for adjudicating disagreements, engagement with the other side can backfre and lead to further polarization. In this landscape, the more we learn about their views the more inclined we are “to see the opposition as pathological” (Aikin and Talisse, 2019, p. 196). In short, trying to assess citizens’ epistemic credibility risks further dividing an already fragmented political society. Second, strictly epistemic approaches are unlikely to bridge our political divisions because they tend to highlight social divisions as opposed to social unity. Since epistemic challenges to our views are prompted by encountering disagreement, it seems the solution should involve adjudicating our disagreements in pursuit of the truth. For this task, it makes sense to focus on why we disagree. Yet focusing on areas of disagreement highlights those issues that most divide us rather than the common values that unite us. Focusing on our diferences reinforces an us/them mentality that promotes social divisions (Mason, 2018). Rather than viewing our fellow citizens as members of a common community, bound together by a common fate and 20  As I will argue below: appraisal respect is contingent and changing whereas recognition respect for a person’s moral status can serve as a stable foundation for building common ground. For more on appraisal respect versus recognition respect see Darwall 1977, 2006. 21  According to a recent Pew Study (2020) trusted sources and standards for epistemically responsible news vary according to political preferences. Pointing to these diferent standards, some epistemologists have argued that political disagreements are “deep disagreements,” because the disagreement is not only about the content of political statements, but also the standards for adjudicating these disagreements (see de Ridder and Lynch, Chapters  12 and  13, respectively, in this volume; for a contrary perspective see Hannon, Chapter 16 in this volume). Even if political disagreements are not truly deep disagreements, they need only be perceived as deep disagreements to drive cognitive polarization (Lynch, this volume). People who think their political opponents are engaged in motivated reasoning in support of a political agenda are likely to distrust both their fellow citizens and the experts on whose testimony their political opponents rely.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Problem with Disagreement on Social Media  271 political project, we encounter those whose views difer from ours as obstacles to achieving our policy goals. We see them as “civic enemies”22 rather than respected civic peers who can be relied on to pursue society’s cooperative goals. To build bridges across the intractable political divisions in society, we need to strengthen the foundation of moral respect for one another in order for any subsequent disagreements to be productively managed in society. But where should we turn for moral respect? Here, our society is less divided than it may seem. Despite the deepening polarization in society, many people nevertheless embrace shared values of equality, freedom, dignity, and respect. Our diferences are rooted in how we apply these values to specifc policy proposals, like whether or not government regulation of businesses is benefcial or harmful for society. More focus on the core values that unite us can serve as an important leverage point for bringing us closer together. Even the studies that were foundational to establishing the polarization trends in the U.S.  also showed that when groups were brought together by their commonalities rather than their diferences, polarization actually decreased (Mutz, 2006, pp. 76–85; Sunstein, 2017, pp. 91–2). Finally, focusing on the epistemic roots of our political divisions encourages a stance towards political disagreement that seeks rational resolution.23 If our disagreements are based on what we do or do not know, we should adjudicate competing claims to fgure out what to believe. Which view is correct? What do you know that I don’t? What do I know that you don’t? If our knowledge divides us, we should trade and dispute claims until we reach an epistemic alliance. Treating political disagreement as rooted in difering knowledge claims suggests that we can bridge partisan divides by correcting misconceptions about an issue, say, by trying to educate each another about the facts at stake.24 In this way, a knowledge-­based approach to our disputes tends to seek rational resolution. However, approaching political disagreements in this way is unlikely to resolve disagreements and bring our divided society together. While it is true that political disagreements can sometimes be solved by correcting misinformation on behalf of one or more parties who believe difering facts, political disagreements are rarely so simple. So ofen in political disagreements, the disagreement remains intact even afer discussion has taken place. As Rawls argues, “citizens’ total experiences are disparate enough for their judgments to diverge, at least to some degree, on many if 22  I borrow this from Jason Brennan (2016) who argues that democracy itself pits citizens against each other as civic enemies. I do not mean to take a stance on democracy itself in this chapter. Rather, I think this situation aptly describes the way highlighting our disagreements without a foundational moral respect for our fellow citizens risks fueling deeper divisions. 23  I say rational resolution, but that need not mean we come to the same stance. Whether one or more stances towards a particular proposition is rationally justifable amongst epistemic peers is the subject of much debate. Tere are key strands of social epistemology (e.g. permissivism or epistemic relativism) that hold that it can be rational to continue to disagree. My point here is that when epistemologists tackle questions of disagreement, the aim is to prove that some position (whether that is remaining steadfast in one’s views or conciliating by revising our confdence in light of peer disagreement; or whether there is one uniquely rational position given a body of evidence or not) is rationally appropriate. For a range of diferent epistemological positions on disagreement, see, e.g. Feldman and Warfeld, 2010 and Christensen and Lackey, 2013. 24  In Chapter 10 in this volume, Ahlstrom-­Vij and Steele are optimistic that correcting factual errors will help to resolve political disagreements.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

272  Elizabeth Edenberg not most cases of signifcant complexity” (Rawls,  2005, p. 57).25 Each person’s life experiences, history, family, community, work, standing in the social order, upbringing, geographic location, and many other factors infuence their beliefs, knowledge, and expertise. People are familiar with the portions of the social world they occupy and their social position and life experiences impact what they believe, what new experiences and information they encounter, and shape the ways that they assess new information. Tese factors lead to predictable diferences of opinion that are likely to persist over time. Unlike epistemologists who seek rational resolution to disagreements, political philosophers seek to manage persistent disagreements without expecting resolution or agreement. For them, disagreement is “the inevitable outcome of free human reason” (Rawls, 2005, p. 37).26 Given the intractable sources of many political disagreements, “we are destined to disagree” but that does not mean we cannot fnd mutually acceptable terms of cooperation to guide us in politics (Gutmann and Tompson, 1996, p. 361). We need to learn to live with persistent disagreements in a way that fosters mutual respect between citizens and ensures just and fair modes of government—even when citizens disagree with the outcome of political decisions and believe the decisions to be incorrect, irresponsible, immoral, or unjust. Political cooperation does not require resolution to our diferences—it requires cultivating mutual respect in spite of our diferences to ensure the legitimacy of political institutions. Drawing together the lessons from this section, the proposed epistemic solutions require an epistemic landscape very far from our current epistemic reality. As things stand, epistemic solutions are likely to exacerbate the very problems they attempt to solve. When disagreement persists in a discussion trying to bridge our divides through common grounding in facts, the parties may leave with the disagreement intact, but their respect for their fellow citizens’ epistemic standing battered. In so doing, an epistemic approach threatens to chip away at the foundation of mutual respect in each interaction. With enough persistent disagreements, the entire edifce may fall unless there is another support system grounded in something other than epistemic respect.

5.  How to Respectfully Disagree At the root of the approaches considered so far is the idea that our divisions are epistemic and that our best defenses will likewise involve cultivating more ep­i­ste­ mically responsible and virtuous behavior. Yet, our ability to trust the testimony of our fellow citizens and to take the evidence they ofer as knowledge is predicated on a more basic level of moral respect. While truth continues to be politically important, approaching disagreement through the lens of epistemology leads us of track. 25  Rawlsians will recognize the diferent factors that cause disagreement as the set of factors he refers to as the “burdens of judgment.” 26  Rawls is particularly interested in “reasonable” disagreement, which he uses as a technical qualifcation to carve out a set of views that are morally upright and committed to fair cooperation.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Problem with Disagreement on Social Media  273 Te breakdown of discourse online provides renewed reasons to draw out a moral basis for political cooperation among diverse citizens—one inspired by Rawlsian political liberalism. Te key lies in cultivating moral respect for citizens that can undergird any epistemic respect built on that basis. We must learn to see one another as well motivated citizens seeking a fair basis for social cooperation. We must see our political opponents’ motives and proposals as morally decent, even if we deeply disagree or think they are misguided. Finding a moral basis for mutual respect and understanding amongst otherwise divided citizens must ground any approach to fxing our current political crisis. For this, I turn to Rawls. As he argues, political disagreements are inevitable in any society that protects basic freedoms. Conscientious citizens who reason responsibly can nevertheless come to difering views about morality, religion, philosophy, and how to lead a good life. However, this is not a problem to overcome, but rather a persistent feature of political life that should prompt us to fnd ways to cooperate on just and fair terms with fellow citizens who disagree. For Rawls (2005), provided they are between people who respect one another as free and equal moral persons, persistent disagreements should be tolerated. Moreover, many of our persistent disagreements are morally reasonable—that is, arising between well-­ motivated citizens who all aim to pursue a good life and cooperate with others on fair terms. Rawls’ understanding of reasonableness involves both a moral and an epistemic component. Te moral component is foundational: reasonable people seek fair terms of cooperation among free and equal citizens on the basis of mutual respect (2005, pp. 49–54). Te epistemic component is tied to the moral one; it demands we respect our fellow citizens as well-­motivated epistemic agents. Tis means we recognize that conscientious citizens may weigh evidence diferently and come to opposing, but equally reasonable, conclusions even afer a full and free discussion has taken place (2005, p. 58).27 Well-­motivated citizens can disagree without thereby being moral monsters or incompetent fools. Fair cooperation requires “recognition respect” for fellow citizens—respect owed in virtue of our basic humanity and moral status as persons.28 Recognition respect is independent of any appraisal of our fellow citizens’ epistemic merits or faws. Te broad lesson to draw from Rawls is that when trying to correct the deep and persistent divisions fracturing society, we should frst seek to establish a foundation of moral respect for one another that can support any further epistemic work to fght misinformation, intellectual arrogance, or deepening antipathy for our political opponents. When we view the other side as moral monsters or intellectually incompetent fools, we are not likely to listen to their dissenting perspectives. To productively manage our disagreements, we must cultivate mutual respect for one another as moral agents. We can understand debate across diferent views as engaged in the same project if all are committed to respecting each person within the political 27 He lists a number of sources of disagreement between reasonable citizens, many of which are epistemically controversial. Nevertheless, his core point is generalizable in a way that shows there are many reasons why we disagree that extend beyond epistemic irrationality. 28  Recognition respect can be distinguished from appraisal respect, which is at play when we assess the merits of an individual who has shown excellence at a particular pursuit (Darwall 1977, 2006).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

274  Elizabeth Edenberg system. Tis is an issue of how we treat one another in political life, not a matter of assessing the relative epistemic capacities of our fellow citizens. Recognition respect for the moral standing of our fellow citizens can serve as a more stable basis for political cooperation than the more contingent appraisals of respecting one another as epistemic agents. How might this shif make a tangible diference in productively managing political disagreements on social media? As noted, the structural features of the networks through which we engage as citizens in political disagreement signifcantly impact what people believe and how they interact. Attempts to rethink how we can more productively and respectfully structure our interactions both on- and ofine must pay attention to both the diversity of the groups of people with whom we interact and the design of social media platforms that facilitate so many politically relevant interactions. Structural change needs to target the source of our problems. As I have argued in this chapter the source of our problems is not ultimately a failure of individuals’ (or groups’) epistemic responsibility. In closing, I will sketch three core lessons for those who seek to bring our divided society together. First, we should begin by acknowledging the inevitable persistence of political disagreements and fnd ways to manage them by structuring dialogue that fosters mutual respect even when parties continue to disagree and fail to arrive at a resolution. Political legitimacy does not require everyone to agree with the outcome of a political decision. Instead, we should seek legitimate procedures to arrive at just political decisions that protect human rights, while leaving open productive ways for those who dissent to continue to make their case. Second, we need to draw out those areas of moral agreement foundational to any fair cooperative endeavor. While fghting false and manipulative claims is certainly important, more foundational divisions now prevent citizens from viewing one another as morally upright, responsible, and reasonable. Studies of polarization on social media show that the “vilifcation of one’s opponents” lies at the root of contemporary polarization (Horwitz and Seetharaman,  2020).29 To repair these divisions, we need to focus on cultivating empathy, mutual understanding, and moral respect for our fellow citizens. We should fnd ways to explicitly highlight common values (without discounting the diferences in how these values are integrated into diferent worldviews). Highlighting shared values can serve as a foundation upon which we can interrogate our diferences. Tis is done more productively when both sides approach the discussion with respect for one another as responsible, reasonable, and well-­motivated moral agents. Te call to develop mutual respect for fellow citizens may seem overly optimistic— especially in a deeply divided society like ours where it is all too common to view those with whom we disagree as moral monsters or intransigent fools. Yet we must

29  Facebook’s Common Ground team who were specifcally tasked with combatting polarization found that the core problem is the “vilifcation of one’s opponents.” In response, the team proposed a range of potential solutions designed to “increase empathy, understanding, and humanization of the ‘other side’.” However, this team was ultimately disbanded, and the newsfeed team was reorganized as the company’s priorities shifed “away from societal good to individual value” (Horwitz and Seetharaman, 2020).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Problem with Disagreement on Social Media  275 not forget that many shared values can form the basis for bridge-­building among citizens: freedom, equality, mutual respect, basic human rights. Despite the many conficts over how these overarching values should be specifed, weighted, and applied in specifc situations, these values are important parts of just about every moral and religious system. Tey are also touchpoints for the United Nations Declaration of Human Rights.30 Without a doubt, we will continue to disagree about how best to respect human rights, freedom, and equality for all. Nevertheless, it is possible to fnd some overlapping consensus on what qualifes as a plausible conception of our shared values. Of course, we should not be overly idealistic in trying to design solutions to our current crisis. Emphasizing our common moral ground need not make us blind to how bad actors are likely to try to manipulate our political systems for their beneft. Tere will continue to be those who fan the fames of division and hate. Any proposal to bridge our divides should seek to build common ground on the basis of what is best in people, without discounting the worse aspects of human nature and the quest for power. However, as a society we will be far more resistant to these provocations when we have cultivated mutual respect for one another and mutual understanding of our common values in spite of our many continued disagreements. Tese frst two lessons apply broadly to political disagreement on- and ofine. Te third and fnal lesson from this chapter suggests a method for redesigning social media to help bridge our political divides. Te underlying goals powering social media’s algorithms need to be rethought if we are to have any hope of repairing our society. Rather than optimizing algorithms for engagement, growth, and time spent on the platform, we should build from a foundation where our agreement about fundamental values like freedom, equality, and mutual respect orient the ways we continue to disagree. Until we start focusing on how to cultivate moral respect for people who deeply disagree—any solutions will be insufcient because they will not be rooted in the moral commonalities that can unite us. Cultivating moral respect for one another and fostering productive ongoing dialogue across disagreements are needed to ground other eforts to improve social media’s epistemic landscape and build individuals’ epistemic responsibility. Yet this is not a task for an armchair philosopher (or computer scientist) alone. To efectively redesign social media requires input from a wide variety of people representing diferent disciplinary perspectives (including philosophy, computer science, law, political science, psychology, history, communications, and science and technology studies) and a wide range of diferent social perspectives. Including people from many diferent walks of life will enable us to have a full picture of the challenges and design a better system.

30  Unfortunately, there are those who also give only lip service to the shared values or respecting human rights while committing violent atrocities. But the shared values gives a small point of moral leverage from which the international community can condemn the violations and appeal to our shared humanity to respect them.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

276  Elizabeth Edenberg

6. Conclusion When we focus too heavily on the epistemic roots of our political divisions, we miss a more promising moral basis for cooperation grounded in building mutual respect for our fellow citizens across division. It is hard to build bridges across social divisions when we view one another as systematically misguided, dupes of media, pawns of ill-­motivated political agents, or perhaps just plain stupid. Placing the responsibility on individuals to cultivate more responsible epistemic practices and virtues likewise continues to treat the problem as one of misguided individuals who, if only they were to be more diligent, could come around to see the errors of their ways. Society cannot bridge the divides in our political system without mutual respect between citizens who nevertheless disagree with one another on many important political questions. When each party thinks their political opponents are either fools or moral monsters, the promise of shared governance is threatened. Instead, cultivating mutual respect for the moral status of our fellow citizens—in spite of our continued disagreements—is crucial for bringing our polarized society together.31

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Aikin, S.  F. and R.  B.  Talisse. 2019. Why We Argue (and How We Should): A Guide to Political Disagreement in an Age of Unreason, Second edition. New York: Routledge. Anderson, E. 1995. “Feminist Epistemology: An Interpretation and a Defense.” Hypatia 10(3): 50–84. Anderson, E. 2010. Te Imperative of Integration. Princeton, NJ: Princeton University Press. Anderson, E. 2021. “Epistemic Bubbles and Authoritarian Politics.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Breakstone, J. et al. 2019. “Students’ civic online reasoning: A national portrait.” Stanford History Education Group & Gibson Consulting. Available at: https://purl.stanford. edu/gf151tb4868 (last accessed Sept. 15, 2020). Brennan, J. 2016. Against Democracy. Princeton, NJ: Princeton University Press. Broussard, M. 2018. Artifcial Unintelligence: How Computers Misunderstand the World. Cambridge, MA: Te MIT Press. Cassam, Q. 2019. Vices of the Mind. Oxford: Oxford University Press. Christensen, D. and J. Lackey. 2013. Te Epistemology of Disagreement: New Essays. Oxford: Oxford University Press. Coady, C. A. J. 1992. Testimony: A Philosophical Study. Oxford: Oxford University Press. 31  Tanks to Michael Hannon, Emily McGill, Gina Rini, Rebecca Tuvel, and two anonymous referees for OUP for extremely helpful comments on earlier drafs of this chapter. Tanks as well to Aaron Ancell, Brian Berkey, Étienne Brown, Lukas Chandler, August Gorman, Jonathan Healey, Maggie Little, Sydney Luken, Jeroen de Ridder, Rob Simpson, students in my Social Media and Democracy courses, participants in the Political Epistemology Workshop at Georgetown, and audiences at the Political Epistemology Seminar Series at the Institute of Philosophy in London, the American Philosophical Association session on Political Epistemology, and the Political Epistemology Conference in Amsterdam for helpful discussions about the ideas in the chapter.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

The Problem with Disagreement on Social Media  277 Collins, P. H. 1989. “Te Social Construction of Black Feminist Tought.” Signs: Journal of Women in Culture and Society 14(4): 745–73. Crowell, C. 2017. “Our approach to bots and misinformation.” Twitter Blog (June 14, 2017). Available at: https://blog.twitter.com/en_us/topics/company/2017/OurApproach-Bots-Misinformation.html (last accessed Sept. 15, 2020). Darwall, S. 1977. “Two Kinds of Respect.” Ethics 88(1): 36–49. Darwall, S. 2006. Te Second-Person Standpoint: Morality, Respect, and Accountability. Cambridge, MA: Harvard University Press. de Ridder, J. 2021. “Deep Disagreement and Political Polarization.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Dotson, K. 2014. “Conceptualizing Epistemic Oppression.” Social Epistemology 28(2): 115–38. Edenberg, E. 2021. “Political Disagreement: Epistemic or Civic Peers?” In M.  Hannon and J.  de Ridder (eds.), Routledge Handbook on Political Epistemology. New York: Routledge. Eubanks, V. 2017. Automating Inequality: How High-Tech Tools Profle, Police, and Punish the Poor. New York: St. Martin’s Press. Feldman, R. and T. Warfeld. 2010. Disagreement. Oxford: Oxford University Press. Fricker, M. 2007. Epistemic Injustice: Power and the Ethics of Knowing. Oxford: Oxford University Press. Gutmann, A. and D. Tompson. 1996. Democracy and Disagreement. Cambridge, MA: Harvard University Press. Hannon, M. 2021. “Disagreement or Badmouthing? Te Role of Expressive Discourse in Politics.” In E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford University Press. Harding, S. 2004. Te Feminist Standpoint Reader. New York: Routledge. Hedden, B. 2019. “Reasons, Coherence, and Group Rationality.” Philosophy and Phenomenological Research 99(3): 581–604. Horwitz, J. and D. Seetharaman. 2020. “Facebook Executives Shut Down Eforts to Make the Site Less Divisive.” Te Wall Street Journal (May 26, 2020). Available at https:// www.wsj.com/articles/facebook-knows-it-encourages-division-top-executives-nixedsolutions-11590507499 (last accessed Sept. 15, 2020). Kappel, K. 2017. “Fact-dependent policy disagreements and political legitimacy.” Ethical Teory and Moral Practice 5(4): 391–413. Kappel, K. 2018. “Tere Is No Middle Ground for Deep Disagreements about Facts.” Aeon Magazine. Accessed at: https://aeon.co/ideas/there-is-no-middle-ground-fordeep-disagreements-about-facts Kearns, M. and A.  Roth. 2020. Te Ethical Algorithm: Te Science of Socially Aware Algorithm Design. New York: Oxford University Press. Kuran, T. 1997. Private Truths, Public Lies: Te Social Consequence of Preference Falsifcation. Cambridge, MA: Harvard University Press. Lackey, J. 2008. Learning from Words: Testimony as a Source of Knowledge. Oxford: Oxford University Press. Lynch, M. P. 2016. Te Internet of Us: Knowing More and Understanding Less in the Age of Big Data. New York: Liveright. Lynch, M. P. 2019. Know-It-All Society. New York: Liveright.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

278  Elizabeth Edenberg Lynch, M.  P. 2021. “Political Disagreement, Arrogance, and the Pursuit of Truth.” In  E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Mason, L. 2018. Uncivil Agreement: How Politics Became Our Identity. Chicago: University of Chicago. Medina, J. 2013. Te Epistemology of Resistance. Oxford: Oxford University Press. Mills, C. 2007. “White Ignorance.” In S.  Sullivan and N.  Tuana (eds.), Race and Epistemologies of Ignorance. Albany, NY: State University of New York Press, pp. 13–38. Mosseri, A. 2017. “Working to Stop Misinformation and False News” Facebook Media (April 7, 2017) Available at: https://www.facebook.com/facebookmedia/blog/workingto-stop-misinformation-and-false-news (last accessed Sept. 15, 2020) Muchnik, L., S.  Aral and S.  J.  Taylor. 2013. “Social Infuence Bias: A Randomized Experiment.” Science 341(6146): 647–51. Mutz, D.  C. 2006. Hearing the Other Side: Deliberative versus Participatory Democracy. New York: Cambridge University Press. Nguyen, C. T. 2020. “Echo chambers and epistemic bubbles.” Episteme 17(2): 141–61. O’Connor, C. and J.  O.  Weatherall. 2019. Te Misinformation Age: How False Beliefs Spread. New Haven, CT: Yale University Press. O’Neil, C. 2016. Weapons of Math Destruction: How Big Data Increases Inequality and Treatens Democracy. New York: Broadway Books. Pariser, E. 2011. Te Filter Bubble: How the New Personalized Web Is Changing What We Read and How We Tink. New York: Penguin Books. Pew Research Center. 2020. U.S.  Media Polarization and the 2020 Election: A Nation Divided. Washington, DC: Pew Research Center. Available at: https://www.journalism. org/2020/01/24/u-s-media-polarization-and-the-2020-election-a-nation-divided/[last accessed Sept. 15, 2020.] Rawls, J. 2005. Political Liberalism: Expanded Edition. New York: Columbia University Press. Rini, R. 2017. “Fake News and Partisan Epistemology.” Kennedy Institute of Ethics Journal 27(2): 43–64. Rini, R. 2021. “Weaponized Skepticism.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Singer, D. J. et al. 2019. “Rational Social and Political Polarization.” Philosophical Studies 176(9): 2243–67. Sinnott-Armstrong, W. 2018. Tink Again: How to Reason and Argue. New York: Oxford University Press. Stony Brook University, Center for News Literacy. 2020. “Digital Resource Center: Organizations Promoting News Literacy.” Available at: https://digitalresource.center/ organizations-promoting-news-literacy (last accessed Sept. 15, 2020). Sunstein, C.  R. 2002. “Te Law of Group Polarization.” Journal of Political Philosophy 10(2): 175–95. Sunstein, C. R. 2017. #Republic: Divided Democracy in the Age of Social Media. Princeton, NJ: Princeton University Press. Talisse, R.  B. 2019. Overdoing Democracy: Why We Must Put Politics in its Place. New York: Oxford University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

The Problem with Disagreement on Social Media  279

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Talisse, R. B. 2021. “Te Problems of Polarization.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Tugend, A. 2020. “Tese Students Are Learning About Fake News and How to Spot It: News literacy instruction is fourishing in the wake of the 2016 election as worries about fake news grow.” New York Times (Feb. 20, 2020). Available at: https://nyti. ms/2v23SKj (last accessed Sept. 15, 2020). Vaidhyanathan, S. 2018. Antisocial Media: How Facebook Disconnects Us and Undermines Democracy. New York: Oxford University Press. Young, I. M. 2000. Inclusion and Democracy. Oxford: Oxford University Press. Zuckerberg, M. 2019. “Standing for Voice and Free Expression.” Te Washington Post. (Oct. 17, 2019). Available at: https://www.washingtonpost.com/technology/2019/10/17/ zuckerberg-standing-voice-free-expression/(last accessed Sept. 15, 2020).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

15

When Should We Disagree about Politics?

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Jennifer Lackey

We are inundated with political information. With the internet and social media part of the daily lives of many around the globe, and people frequently sharing not only their political views but also multiple news stories with political content, most of us fnd ourselves with seemingly limitless opportunities for disagreement. Moreover, these opportunities ofen involve matters of enormous signifcance, such as climate change, Russian interference with U.S.  elections, sexual assault, separating children from their parents at the U.S.–Mexico border, and many other topics that have life-­changing consequences. At the same time, we have a broad range of obligations, from the mundane and practical, such as grocery shopping and taking out the trash, to the more normatively weighty, such as keeping our promises, caring for loved ones, and living responsibly, both in belief and in action. It cannot be expected of us that we spend all day, every day, disagreeing with ­people about politics. My aim in this chapter is to explore in greater detail when we should, from an epistemic point of view, disagree about politics by frst asking the general question: when do we have the epistemic duty1 to object to assertions we take to be false or unwarranted? I will begin by highlighting two features of the duty to object that reveal that it is best understood as an imperfect, rather than a perfect, duty, and hence that there are imperfect epistemic duties, in addition to moral ones. I then examine in detail one specifc account of imperfect moral duties: Liam Murphy’s collective view that includes what he calls the Compliance Condition. Such a view understands imperfect duties as belonging to groups or collectives, but denies that we need to “pick up the slack” from non-­complying members. Afer showing that we should reject the Compliance Condition, especially on epistemic grounds, I sketch a view according to which the duty to object is an imperfect epistemic one that belongs to groups. In the last section, I apply these considerations specifcally to the political domain and highlight the ways in which distinctive or pressing issues arise when we disagree about political matters.

1  For reasons that will become apparent later in the chapter, I will talk about there being the “duty” to object. But for those who don’t like talk of duties, this can be substituted with “obligations,” “demands,” “normative force,” or “normative pressure,” which I will ofen use interchangeably with duties.

Jennifer Lackey, When Should We Disagree About Politics? In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Jennifer Lackey. DOI: 10.1093/oso/9780192893338.003.0016

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

When Should We Disagree about Politics?  281

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1.  Te Duty to Object as an Imperfect Epistemic Duty Broadly speaking, there are at least two sources of the normative pressure at work in the duty we have to object to what we take to be false or unwarranted.2 Te frst, and perhaps more familiar, is moral. What people say can produce life-­altering harm to others or be the cause of grave injustices. A false eyewitness identifcation can rob a person of his freedom and ultimately his life. Racist and sexist comments can not only undermine a person’s sense of self, but also breed hatred and violence within communities. Since being good moral agents involves preventing at least some serious harms, especially when doing so would cost very little,3 speaking out against assertions that perpetuate such harms is surely called for. In addition to the moral, the other source of the pressure to object is epistemic. It is a familiar fact that we have duties as individual knowers—I should, for instance, try to believe in accordance with the available evidence, choose reliable sources of information, avoid dogmatism and close-­mindedness, and so on. But what is less familiar to us is that, as members of communities or broader social environments, we might also have epistemic obligations to others, both individuals and collectives. For instance, just as I have a moral duty to keep you from starving if all it would take is handing you an apple from my kitchen table, so, too, I have an epistemic duty to prevent you from improperly believing if all it would take is a minor correction on my part. Indeed, whatever we take to be of epistemic value—truth, knowledge, understanding, evidence, and so on—there is simply no reason why we would have obligations to promote these ends only in ourselves. If it is valuable for me to believe truly, then it is also valuable for you to believe truly. If I have distinctively epistemic obligations to ensure this end in myself, why would I not, at least sometimes, have epistemic duties to assist others in this end?4 At the same time, the duty to object difers in important ways from some other paradigmatic kinds of duties, both epistemic and moral. Te frst is that the duty itself seems subject to direct infuence by what others are doing, especially those in the conversational context in question.5 In the epistemic realm, if I have a duty to 2  Much of the material in this frst section is developed in greater detail in Lackey (2018). 3  Principles of this sort are prevalent in the literature on moral obligations, but perhaps the most well-­ known is found in Singer (1972). 4  I should emphasize two points here. First, I want to make clear that I aim to remain as neutral as possible on particular moral and epistemological views. Tus, while I talk about promoting ends in myself and others, these ends might be, for instance, acting or believing virtuously, acting so as to satisfy the categorical imperative, believing in accordance with the evidence, maximizing happiness or truth, and so on. Second, my strategy here is, at least in part, to shif the burden to those who espouse the doxastic and the intrapersonal theses of epistemic duties to explain why we ought to promote epistemic ends with respect to only my own beliefs. However, while a full discussion of this lies beyond the scope of this chapter, I should note that there are also positive arguments that show that epistemology should be expanded to include our duties to be responsive to the epistemic needs of others. One way this can be done is to follow Reed (2018), who argues that epistemic normativity is best explained as arising from epistemic practices, many of which are distinctively social in nature. Individuals have an extremely strong practical reason to belong to these practices insofar as they structure the way we share information with others. On this view, epistemology is itself a social phenomenon—to reduce epistemology to the egocentric perspective is to artifcially constrain the scope of the epistemic. 5  I say “especially” rather than “only” those in the conversational context because I want to remain neutral on how precisely such contexts ought to be individuated. For instance, even if one wants to say

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

282  Jennifer Lackey believe in accordance with the evidence, I am not more or less obligated to do so if you are. I cannot, for instance, look at how careful my husband is in the formation of his political beliefs and take this as a license for forming my own political beliefs with disregard for relevant evidence. Of course, I might rely on my husband’s testimony as a reliable source of information regarding politics, but this does not at all diminish my own duties as an epistemic agent. I need to be just as sensitive to the evidence as I  would be if he were not so careful—it is just that his testimony is some of the ­evidence on which I rely when I responsibly form my political beliefs. Te same is true in the moral realm. If I make a promise to you, my duty to keep it is unafected by whether others around me do the same. Suppose, for instance, that Bill and I both make a promise to Rose to visit her in the hospital. Bill’s keeping his promise, or failing to do so, in no way afects my duty to keep my promise to Rose. Sure, Rose might be less upset if Bill shows up when I don’t because at least someone came to see her, but this does not afect the responsibility that I have to visit Rose and the corresponding claim that she has on me. Should Rose express her resentment or disappointment at my not keeping my promise, it clearly will not help for me to say, “But Bill kept his.” In contrast, the duty to object can be, and ofen is, directly afected by what others are doing. If I am at a social gathering and someone says that Farmhouse closes at 11:00 PM, which I know to be false, my duty to object diminishes if you or others correct the record. Indeed, the duty might be altogether eradicated if either your objection is particularly powerful—perhaps you say, “I was at Farmhouse last night and I’m certain that it stays open until midnight”—or multiple people in our group object. Te same is true with respect to the moral pressure to speak up. If a potentially damaging statement is made about a colleague at a party that I take to be groundless—say, that she did not follow university policy in the handling of a student’s complaint—my obligation to weigh in is afected by whether others do so. If your correction is sufcient to clarify or even settle the matter, then my obligation to also object might be eliminated. Indeed, we can imagine circumstances in which there is moral pressure to not say anything. Suppose that the damaging statement was made out of ignorance or unintentional carelessness, and the speaker seems deeply embarrassed afer being called out. My adding my own objection might be seen as “piling on” criticism in an unnecessary and callous manner. All of the examples discussed thus far focus on the duty to object being diminished or eliminated by what others do, but it can also be made more pressing.6 For instance, there are quite a few insightful and compelling pieces that have been written lately about the dangers of Donald Trump’s misogyny, racism, and utter disregard for the truth. Such articles have been written both by members of my profession and that I am not part of the national conversational context about current politics, I want to leave room for my duty to object to Trump’s views nonetheless being afected by what those who are part of that conversational context are saying. 6  I will be talking about duties as if they come in degrees. For instance, I will speak about duties being greater, more pressing, diminished, and so on. Tis way of framing matters is intended to capture the intuitive point that normative pressures vary. For instance, there is greater normative pressure for me to save my daughter from drowning than there is for me to keep my promise to put the lid on the jar of peanut butter. If talk of duties coming in degrees is problematic for some, it can be substituted with “normative pressure” or the like.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

When Should We Disagree about Politics?  283 by those in the broader social and political community. Given this, I feel no need to write an article or op-­ed condemning Trump and his views because others have already done so, and have done so very efectively. But now suppose that no one had. Suppose that Trump’s lies and bullshit and hatred were met by complete silence. Here, I seem not only to have the duty to object, but also a very pressing one. Tis is at least in part because my actions are more likely to have substantial causal powers— there is a greater chance that speaking out would add new ideas and challenges to the conversation, while failing to do so would permit and perhaps even sanction Trump’s ideology. For both epistemic and moral reasons, then, my duty in such a scenario is great, as it seems crucial to furthering both truth- and goodness-­related goals. For the sake of ease of expression, let us understand this feature of the duty to object as follows, where the duty at issue can be either epistemic or moral: others: One’s duty to object can be directly infuenced by whether others object, especially those in the conversational context in question.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

others highlights the extent to which the duty to object is sensitive to social environments and the interpersonal relationships found in them.

Another feature of the duty to object that is not shared by many other duties is that it can be directly afected by social status. By “social status,” I include not only those properties that contribute to diferences in power, paradigmatic examples of which are race, gender, and class, but also properties that are more epistemic in nature, such as authority and expertise, which ofen accompany professional roles. For instance, my epistemic duty to consider relevant counterevidence is not impacted by the fact that I am a woman or a professor. Of course, such social positions might put me in touch with more or diferent bodies of evidence and might bear on what counts as “relevant,” but I don’t have less of a duty to consider the evidence that is available than does the President of Northwestern, nor a greater duty than do my undergraduate students. Te same is true of my moral duty to keep my promise to, say, take you for your dialysis treatment. Sure, positions of powerlessness might make it difcult to make promises in the frst place, and there might be more conditions that provide excuses for not keeping them—such as unreliable childcare, a disconnected phone, and so on—but the duties themselves do not vary. In contrast, if a tenured, white, male professor hears a fellow colleague make a clearly sexist remark, his duty to object might be greater than that of his Black, female, junior colleague. Tis might be true both epistemically and morally. Let us consider the former frst: with great power in a domain ofen comes greater authority, and thus an increased likelihood that one’s testimony will have an efect. So, if we assume that the sexist remark in question is false and that one of our aims as epistemic agents is to promote the truth, then the white professor objecting to it might have more epistemic impact in producing true beliefs, both at the individual and the collective level.7 Tat this is the case is supported by a recent study by Kevin Munger (2016), in which he looks at the impact of calling out racist harassment 7  As should be clear, I am assuming neither that the promotion of truth is the only epistemic aim, nor that all epistemic aims are consequential in nature.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

284  Jennifer Lackey online via Twitter. To this end, he used “bots” to object to harassers, varying their identities not only by in-­group (white man) and out-­group (Black man) membership, but also by the number of Twitter followers each bot has. Munger found that subjects who were sanctioned by high-­follower white males signifcantly reduced the use of racist slurs, leading to the following title of an article in Te Atlantic: “Why Online Allies Matter in Fighting Harassment: A clever experiment with Twitter bots shows that telling people not to be racist can work—but only if it comes from someone infuential and white.”8 In this same article, the author writes:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Tere’s a reason why higher-­status members of these communities bear a larger share of the responsibility for speaking out against racist or bigoted speech,” says Betsy Levy Paluck, a psychologist at Princeton University. “Tis isn’t just a moral judgment but an empirical regularity that’s been coming out of many research programs: People with higher status are infuencing norms, and with that infuence comes responsibility. If anyone says, I’m not a role model, that’s a wish, not a fact. (Munger, 2016)

Here are other examples of this kind: a pediatrician speaking out about the safety of vaccines, a prosecutor objecting to police misconduct, and a university ofcial condemning inefective sexual assault policies on campus. In each case, the duty to object might be greater for the person in question than it is for the average citizen in large part because that person’s objecting is likely to lead to more true, and fewer false, beliefs. Similar considerations apply at the moral level. Surely there is more moral pressure for the tenured, white, male professor to object to the sexist remark than there is for his junior colleague, both because he has the social standing to bring about positive change and because there is so little risk of harm for him. He has political, professional, and economic advantages that make him far less vulnerable. He does not have to worry about being regarded as stereotypically angry or whiny, and he does not risk losing his job and fnancial stability. For the sake of ease of expression, let us understand this feature of the duty to object as follows, where the duty at issue can be either epistemic or moral: social status: One’s duty to object can be directly infuenced by one’s social status.

What this feature makes clear is that with greater power comes greater responsibility to speak up against both epistemic and moral wrongs.9 8  Source: http://www.theatlantic.com/technology/archive/2016/11/why-­online-­allies-­matter-­in-­fghting-­ harassment/507722/, accessed December 7, 2016. 9  Of course, there is also ofen a greater cost for those who speak out against those who have more power than they have. In discussing whistleblowing, for instance, Peggy DesAutels writes, “Tose with power use the race, class, and gender of resisters in a variety of ways, depending on the circumstances, as marginalizing strategies. In some cases, gender becomes more salient, in others race, and in others class. And of course, all three overlap in complex ways. . . . . If a whistleblower has socially-­marginalized traits, these traits are sure to elicit trait-­specifc retaliatory measures as a way to emphasize just how much the whistleblower always has been and always will be marginalized in our society—as a way to emphasize that the whistleblower was never really an insider afer all” (DesAutels, 2009, p. 231). Similarly, Lisa Tessman

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

When Should We Disagree about Politics?  285 Both others and social status are supported by our practices of praise and blame. We are far less likely to criticize someone for silence when everyone else has already loudly protested, and we are more likely to do so when no one has. Imagine, for instance, that I recount to a friend that you reported that a colleague had plagiarized a passage in his book, that I knew this was merely a rumor, and yet I remained completely silent. No doubt upon asking me why I didn’t say anything, she would have quite diferent reactions to, “Well, everyone else immediately objected, and so I didn’t think further good would come from my doing so, too,” and “I didn’t want to be the only person to say something.” Similar considerations apply to social status: I might regard the actions of a Black undergraduate student who speaks out against racist activities on her campus as supererogatory and deserving of signifcant praise, while a tenured professor might simply be fulflling expectations when she does so. Indeed, it is generally met with disapprobation when those in power remain silent in the face of confict and allow the most powerless and vulnerable to put themselves at risk by doing all of the talking. others and social status also point to a fundamental diference between a duty like promise-­keeping and one like objecting: the latter is deeply social in a way that the former is not. Whereas the obligation to keep a promise is unafected both by the social world beyond that of the promiser and promisee, the duty to object is inextricably tied to social relations. In Lackey (2018), I argue that these two features of the duty to object are shared with the duty of charity. I have the duty to be charitable, perhaps by donating to the poor, but surely it is afected both by what others are doing and by my social status. If no one in my community is donating to a local economically disadvantaged family’s fund to pay for expenses associated with a medical emergency, then there might be greater pressure for me to do so than if everyone were. If I come upon a homeless man on the street in frigid weather, it may be less obligatory for me to help him if four others are already coming to his aid. Mark Zuckerberg and Bill Gates have greater obligations to engage in philanthropy and charitable giving than I do, and I have greater ones than do my incarcerated students at Stateville Correctional Center. Te dissimilarities between the duties to object and to keep promises, and the similarities between the duties to object and to be charitable, point to a potentially deeper explanation. Promise-­keeping is regarded as a classic perfect duty, while charity is an imperfect one. Tere are at least two dimensions of imperfect duties that are crucial for them to truly be imperfect in nature (Schroeder,  2013). First, discretion and latitude are allowed in their fulfllment. Charity, for instance, might require that I donate to the poor, but it does not specify to whom or how much. I may fulfll this duty by sending $200 to Oxfam every month or by sending $500 to Habitat for Humanity twice per year. Contrast this with the classic perfect duty of promise-­keeping, where there is no discretion or latitude regarding how I satisfy it. If I promise to visit Rose in the argues that resisting oppression ofen involves what she calls “burdened virtues”: “What I think of as the burdened virtues include all those traits that make a contribution to human fourishing—if they succeed in doing so at all—only because they enable survival of or resistance to oppression (it is in this that their nobility lies), while in other ways they detract from their bearer’s well-­being, in some cases so deeply that their bearer may be said to lead a wretched life” (Tessman, 2005, p. 95).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

286  Jennifer Lackey

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

hospital on Tuesday, then I have a duty to do just this. It will not do to instead visit George in the hospital on Tuesday, or to visit Rose at home on Tursday. Of course, I might build latitude directly into the content of a promise—“I promise to visit you at some point next week”—but this is not the same as there being latitude respect to its fulfllment. Second, it is not only possible to do more than is required with respect to imperfect duties, but doing so is also morally valuable. S.  Andrew Schroeder writes, “you can be more gracious, merciful, charitable, or benefcent than morality demands. You can’t, on the other hand, keep more of your promises than morality requires, nor can you repay more than what you owe. You can, of course, do more for a friend than you’ve promised and you can give more money to a creditor than you agreed to, but the surplus in each case is no longer regarded as an instance of promise-­keeping or debt-­repayment. Rather, it is typically described as benefcence or gratitude” (2013, pp. 560–1). Both of these are true of the duty to object. Tere is, for instance, certainly latitude in the fulfllment of this duty. I may object on Facebook, in a published article, at the dinner table, on the subway, in a blog post, and so on. But surely it is not necessary that I object in all of these places all the time. Moreover, someone may object more than is required of them and this may have additional normative value. A graduate student, for instance, who frequently objects to instances of sexual harassment and assault on her campus—ofen at great expense to herself—is clearly doing more than is required of her, and this has additional moral and epistemic value. She is, for instance, promoting both moral and epistemic goods in her community through her objections, and this renders her deserving of praise. Te duty to object, thus, can be understood as imperfect in nature. When the focus is on the promotion of epistemic ends—such as truth, wisdom, believing in accordance with the evidence, the cultivation of intellectual virtues, and so on—the imperfect duty is epistemic. I now want to turn to a deeper exploration of the best way to understand the imperfect epistemic duty to object.

2.  Groups and Compliance One way of understanding imperfect duties is in collective terms—as belonging to groups rather than to individual agents.10 On this view, the duty to engage in, say, charitable giving is one that is directed at us collectively, as a community or society. Liam Murphy, for instance, proposes what he calls a cooperative conception of benefcence, according to which: benefcence can be understood in terms of a shared cooperative aim. Let me try to explain the diference between a cooperative aim and an individual or noncooperative aim. If two of us have individual aims to promote the good, what I should do may depend on what you do. For what I can best do to promote the good can vary depending on what you are going to do, and, as already noted, I might be best able to achieve

10  See, for instance, Cohen (1981), Murphy (1993, 2003), and Schroeder (2013).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

When Should We Disagree about Politics?  287 my aim by cooperating, in a secondary way, with you. But if you cannot cooperate, or if you fail to promote the good, this in no way afects my understanding of, or commitment to, my own aim. If we both have a cooperative aim to promote the good, by contrast, we do not see ourselves as engaged in separate solitary enterprises. Te other person’s relevant behavior afects the attitude of each to his own requirements.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Indeed, we can say that if we all have a cooperative aim to promote the good, each of us does not, strictly speaking, aim to promote the good. Each sees himself as working with others to promote the good. Tus the best way to describe the aim of each might be: “to promote the good together with others.” (Murphy, 1993, pp. 285–6)

On Murphy’s view, then, the duty in question is a collective one and the distribution of each share of responsibility to individual members of the groups is based on what would most promote the good. A central advantage of this cooperative conception of benefcence is that it provides the resources for responding to a classic objection to act consequentialism: namely, that it is overdemanding. In the face of world poverty, for instance, the obligation of those of us in the First World to promote the good seems to demand that we sacrifce most of our resources for the sake of others, and this strikes many as requiring too much. But according to Murphy, it is typically only in situations of partial compliance with act consequentialism—where not everyone is doing their fair share—that the view is excessively demanding. In cases of full compliance, the view ofen requires far less of each agent. If, say, everyone in the First World were now doing their part in responding to world poverty, then it is not the case that each of us would have to give up most of our resources for the sake of others. For Murphy, then, shifing over to understanding benefcence in cooperative terms opens the door to a kind of act consequentialism that does not demand that we take on more when others fail to do their part. In particular, if benefcence is a cooperative project, then “it would be natural to resist taking on, in addition to one’s own share of the burdens of this . . . project, the shares of noncomplying agents” (Murphy, 1993, pp. 267–8). Indeed, Murphy takes the satisfaction of a “Compliance Condition” to be necessary on any principle of benefcence, according to which the demands on agents should not increase as compliance with the principle by other agents decreases. Tus, Murphy’s view ultimately holds that morality should require of each agent no more than it would demand if everyone were fulflling their duties: “Each agent is required to act optimally to perform the action that makes the outcome best except in situations of partial compliance with this principle. In situations of partial compliance it is permissible to act optimally, but the sacrifce each agent is required to make is limited to the level of sacrifce that would be optimal if the situation were one of full compliance” (Murphy, 1993, p. 280). While this framework provides some resources for responding to concerns about morality being overdemanding, it comes at what I take to be a debilitating cost. Consider the following variant of Rescue: suppose that you and I are walking past a shallow pond, and we see two children drowning whom we could easily rescue. On Murphy’s view, you and I are obligated to do only our fair share here—which we can assume amounts to each of us saving one child—and no more. Suppose further, then, that you run in and save one of the children while I stand by idly watching the

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

288  Jennifer Lackey situation unfold. If I continue to stand on the shore watching the second child drown, you are now under no obligation to run in and save her, even if it would cost you very little, as that would involve picking up the slack that results from my noncompliance.11 But this seems wildly counterintuitive. Imagine any one of us not criticizing you for your inaction simply because you only had to do your “fair share.” Responses have been ofered to this objection—some of which amount to simply biting the bullet by admitting that your saving the second child would indeed be supererogatory12—but none will be able to deal with the core problem that arises via the Compliance Condition. If what each of us is obligated to do is never afected by the compliance, or complete lack thereof, of others, then there will always be variant Rescue cases to face. Tat is, there will be situations in which others fail to do what they are supposed to do and we clearly seem obligated to step up and take on more, especially when doing so comes quite easily to us.13 Tis is no less true when it comes to the duty to object. Suppose that the obligation to challenge false accusations of professional misconduct by my colleagues falls on my department as a group rather than on me individually. Two such accusations— one against Ben and another against Bill—are raised at a social gathering that a fellow department member and I are attending. If I call into question the claim about Ben and my fellow department member remains silent, surely I should speak out about Bill, too. From an epistemic point of view, this is especially clear if my evidence on behalf of their innocence is equal, or even greater with respect to Bill’s. Moreover, it would hardly assuage Bill’s anger or sadness or feelings of betrayal for me to say, “Well, I spoke out about Ben, and so I’d done my fair share. It’s not my fault that our colleague remained silent.” Tis raises serious concerns about the inclusion of the Compliance Condition in a conception of imperfect duties—either epistemic or moral; for such a condition is fundamentally at odds with the arguments on behalf of others ofered earlier. But there is also reason to think that this conception of imperfect duties is incompatible with social status. To see this, consider the case of Ben and Bill again: our obligations to object seem deeply impacted by a number of under-­described features of the case, all of which we might say fall under the distribution of goods, epistemic and otherwise. Are Ben and Bill both tenured? Is either a member of an underrepresented group? Are both department members who are expected to object regarded as equally senior and authoritative? Does either occupy a precarious position within the university because of race, gender, and so on? Imagine, for instance, that Bill is a Black, untenured faculty member, while Ben is white and has an endowed chair. My obligation to speak out against a claim that Bill plagiarized seems even greater than it is that Ben did, in part because the risks to Bill are potentially more severe. Similarly, if my fellow department member is, say, untenured and Latina and already feeling vulnerable, then it might be obvious that I ought to challenge the accusations about both Ben and Bill while she remains silent. 11 Tis sort of case is discussed in Singer (1972), Cohen (1981), Murphy (1993,  2003), and Schroeder (2013). 12  See, for instance, Cohen (1981) and Murphy (1993, 2003). 13  See Tadros (2016) for a similar argument.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

When Should We Disagree about Politics?  289 Of course, these diferences in status could be factored into the initial division of duties among group members. If speaking out would cost you a great deal and me nothing at all, then our obligations may vary at the outset. If challenging false claims against Ben would bring about little that is positive, while doing so in the case of Bill would be the diference between, say, his getting tenure and being terminated, this might be accounted for from the start.14 But here is the lingering problem for this view of imperfect duties: even if the distribution of goods in part determines the division of obligations among members of groups, compliance, or lack thereof, can redistribute goods in ways that bear on the duties themselves. Suppose that I am the only person in my department to speak out about my colleague sexually harassing one of our graduate students. I am a senior faculty member and so let us say that I have a greater duty to do this than some others, particularly my less senior colleagues. However, my compliance here, and the lack of it by my colleagues, redistributes some of the relevant goods in ways that matter for our obligations. Upon objecting, my being a senior faculty member may become far less salient than does my being a woman, and this might bear directly on the credibility aforded to me. Te fact that I am speaking out about sexual harassment, which women are frequently accused of lying about, might further damage my epistemic and moral reputation. Tat my colleagues remained silent about this matter might infate their apparent trustworthiness, as they might be taken to be standing by someone who is being wrongfully accused. I might sufer attacks on social media and trolling as a result, which impact my psychological wellness, and my standing at the university might be threatened. In all of these ways, my compliance and the absence of it by my colleagues are directly connected with our corresponding statuses, which, in turn, impact our obligations. Tere is one further problem for the Compliance Condition in relation to the duty to object that I would like to highlight:15 doing only my share can make matters worse than not doing anything at all, especially when there is a lack of compliance on the part of others. Suppose that two suspects who are presumed guilty in a recent murder case are placed in front of you and me, and we both know that neither is the man we saw feeing the scene of the crime. Assuming that there is no reason for anything but an equal division of labor between us, you and I are each obligated to object to one of the men being charged with the murder. Now suppose that I object to the frst suspect being charged, am silent with respect to the second, and you say nothing at all about either. If the police ofcer knows that we both got an excellent view of the perpetrator, then my objecting to the presumed guilt of only the frst suspect might reasonably be regarded as implicating the guilt of the second. Indeed, the implicature that follows from my doing my share might stand with or without your compliance. Tus, my doing my part, according to the Compliance Condition, can make the situation epistemically and morally worse—the authorities might take themselves to have support for their false belief that the second suspect is guilty, which could beget massive additional false beliefs downstream in the courts, the 14  Similarly, in Rescue, if you can swim while I cannot, you may be obligated from the beginning to save both of the drowning children. 15  I am grateful to Jefrey Sanford Russel for a helpful conversation here.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

290  Jennifer Lackey public, and so on; and, the suspect might end up falsely imprisoned for a crime he did not commit, losing just about everything in life that is valuable. What all of this shows, then, is that Murphy’s model of the imperfect duty of benefcence cannot simply be adapted to understand the duty to object. But notice: all of the objections raised above focus on the Compliance Condition rather than the collective aspect of the view. Might we still take on board the core insight that imperfect duties—specifcally, the duty to object—might helpfully be understood in collective terms?16

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Te Imperfect Duty to Object as a Collective Duty Understanding imperfect duties in collective terms is a promising model for our purposes, as it provides an explanation for what we fnd in others; namely, that the duty to object can be afected by what other members of the community are doing. Contrary to what is suggested by Murphy’s Compliance Condition, if you and I agree to paint someone’s house, it is our responsibility to do so. If you do not do your share, perhaps because of either an emergency or sheer negligence, I have to take on more in order to fulfll the agreement. In other words, I have not fulflled my commitment for us to paint your house if I have done only my share and you have not done yours. For my commitment was not to do only my share of our painting your house but, rather, for us to paint your house, where there might be many ways for this to be distributed. Similarly, if others in my community are doing more or less than their share in setting the record straight, this directly impacts what my epistemic duty is. Groups or collectives can be understood here very broadly and heterogeneously. Some groups are highly formal where one becomes a member through appointment or election, such as the Supreme Court of the U.S., Congress, and the administration of an institution. Other collectives have a formal structure, but membership is extended through invitation, admission, or “open enrollment,” such as fraternities, schools, and churches. Tere are some groups without a formal structure where there is no choice about membership, such as children, Latinos, rape victims, and those with Down Syndrome, while others are as voluntary as they come, such as dog owners, stamp collectors, and ethical vegetarians.17 Some groups have varying degrees of structure and voluntariness, such as U.S.  citizens, cancer survivors, and heart surgeons and some are so broad that the parameters of membership are murky, such as global citizens. Finally, there are some groups to which we all belong in one form or another—such as being members of the moral, political, or epistemic community. With respect to the duty to object in particular, membership in one group or another might be more or less salient at diferent times, which can directly impact the normative pressure to speak out in that context. If, for instance, there is a 16  Notice that I deliberately speak of “collective” duties rather than “cooperative” ones. Tis is because the latter, but not the former, seems intimately connected with issues of compliance, which is precisely what I am rejecting. 17  To be clear, I mean here the group that includes all children, the group that includes all Latinos, and so on.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

When Should We Disagree about Politics?  291 conversation about a highly visible case of sexual assault on my campus, my ­membership in the Northwestern community is more salient than is my U.S. citizenship, and thus my obligation to say something is greater. In contrast, if I am in Europe and people are talking about Trump, then my being American is more salient than is my connection with Northwestern, and so there is more normative pressure to make clear my resistance to him than there is for, say, a citizen of Brazil. With these points in mind, if we conceive of the duty to object as a collective one, how should we understand what this amounts to? Here is the start of an answer: the shares of responsibility to object in a given collective should be distributed at least in part on the basis of the goods each member possesses. Restricting ourselves to just the epistemic dimension of the collective duty, we might say, frst, that the distribution of epistemic goods among the group’s members in part determines the epistemic share each should shoulder. Epistemic goods here should be understood broadly to include not only the paradigmatic aims of inquiry, such as truth, knowledge, wisdom, understanding, and justifed belief, but also features that are ofen connected with the achievement of these aims, such as authority, expertise, and credibility.18 So, for example, suppose that it is the community’s collective duty to object to a false and damaging accusation made against one of our members. What is the responsibility that each of us bears? Tis will depend on who has relevant knowledge about the situation, how well-­supported the beliefs in question are, whether we are regarded as authorities on the matter, how much credibility we each have on the topic, how great an epistemic impact our testimony will have, and so on. For instance, those who know more about the situation have a greater obligation to weigh in than do those who are ignorant of the matter; those who understand the details in question have more responsibility than those who do not; and those who are regarded as credible on an issue relevant to the case have a stronger duty than those who are not. It is helpful to see that there is an immediate positive result of this view: understanding the epistemic duty to object in terms of the distribution of epistemic goods increases the likelihood that these very epistemic goods will be promoted. If those with, say, knowledge and authority have more of a responsibility to object than those with false beliefs, and people are doing what they ought to do, then there is a greater chance that knowledge will be promoted. Note, however, that when I argued on behalf of status, I included broader social statuses beyond merely the epistemic. Tere are at least two ways in which this fgures into the picture I am sketching here. First, social status ofen has a direct bearing on epistemic status: white police ofcers are ofen aforded greater credibility than Black suspects, professionals are ofen regarded as having more authority than non-­professionals, and so on. Tus, the distribution of epistemic goods is ofen inextricably linked with social status. Second, we are obviously far more than just 18  It should be noted that with all of these goods, there is a diference between the possession of them, and the recognition that they are possessed. One might in fact be an expert on a given topic, for instance, but I want to say that it is a further epistemic good to be recognized as an expert. Tis is because the recognition enables one to play a critical role in the epistemic community by producing knowledge in others.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

292  Jennifer Lackey epistemic beings, and the goods we have go beyond just those with epistemic value. As I said at the start of this chapter, the normative pressure to object comes from at least the epistemic and the moral realms, and so the social status of the members in our groups will clearly afect the risks, and the corresponding harms, involved in their speaking out. While I have argued for this at length earlier in the chapter, the point that I want to emphasize here is that a general and comprehensive view of the duty to object will include all of the ways in which our place in the broader social world impacts our responsibilities. Tus, while you might have more epistemic goods than I do, your social status might be far less secure than mine, and so my overall duty to object—which goes beyond just the epistemic dimension—might be greater than yours. My view, then, is that the duty to object can helpfully be understood as an imperfect one, where this is characterized in collective terms. For the sake of clarity, let us call this the imperfect duty view (IDV), which can be broadly understood as follows: imperfect duty view: Te duty to object is a collective one, where each member’s individual share of the responsibility is determined in large part by (i) the distribution of goods, especially social and epistemic statuses, and (ii) whether the other members of the collective do their share. Te greater (lesser) the goods, the greater (lesser) the duty; the more (less) that some members do, the less (more) that other members have to do.19

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

I put this in general terms, but the specifc epistemic version of the duty to object would restrict the goods and statuses in question to only those with epistemic signifcance.20 Let us now turn to applying the Imperfect Duty View to disagreements specifcally about politics.

4.  When Should We Disagree about Politics? When should we object about political matters? Te answer of the Imperfect Duty View has several steps. First, a group of which I am a part needs to take the assertion in question to be false or unwarranted. Te duty to object is a collective one, so it is our duty to object, and I as an individual have a part of this duty. Tere are countless 19  One might worry that my view is promoting silencing (see Hornsby (1995) and Langton (1993)) or testimonial injustice (see Fricker (2007)) by encouraging those with privilege and power to ofer more assertions than members of marginalized groups. But notice: my view is compatible with there being other duties that can pull in other directions. For instance, there may be the duty to be silent, which can have both moral and epistemic dimensions. Moreover, my view does not in any way prevent or discourage speakers with a lower social status from objecting—it merely says that their doing so is supererogatory and thus deserving of praise, while the same action might be obligatory for a speaker with a higher social status. 20  I develop this view in greater detail in Lackey (2018).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

When Should We Disagree about Politics?  293 groups, and each of us is a member of many. I, for instance, am an American, a member of the Northwestern community, a Democrat, a prison educator, an academic, a philosopher, a mother, and so on. Which of these groups is responsible for objecting, and thus which of my group memberships is relevant, will depend on the content of the assertion and its salience to these groups. If, for instance, someone makes a false claim about a Northwestern student, then it is the responsibility of the Northwestern community to set the record straight. My part in this group responsibility is determined by what other members of my community are doing, as well as by my social status. Since the domain we are concerned with here is political in nature, my membership in groups with political salience will be most relevant, such as being American, being a liberal, and so on. One of the central epistemic aims of objecting is to promote epistemic ends in our communities. So as to avoid debates about what the central epistemic end is, I will just assume pluralism for the purposes of this chapter. In this sense, it is best to promote truth at some times, knowledge at others, and understanding, believing in accordance with the evidence, cultivating virtues, and so on at still others. Since we do not always know when these ends have been achieved, a good epistemic rule of thumb is this: the more resistant to counterevidence an assertion is, the more objections are needed. It is, for instance, almost certainly not epistemically sufcient for one member of a group to object to a racist assertion, but this may be adequate if the assertion is about the time a local restaurant closes. But it is important to note that the aim of objecting is not always met merely by achieving one of these epistemic ends. It may also be crucial to contribute to the communal pool of evidence, even when these ends have been met. Suppose, for instance, that I object to an assertion, and all of the relevant members of the ­conversation or community change their minds and come to have the knowledge in question. It may still be epistemically valuable for others to object. Further objections may deepen the available evidence, make the formed beliefs more resistant to counterevidence, strengthen the epistemic positions of the community members, and so on. Tis is especially valuable when beliefs are particularly vulnerable to attack, which is true of many political beliefs. Even if everyone in your community forms the correct belief about, say, a new gun law, it may be helpful for others to also add their voices to the body of available evidence on the matter. Tis will strengthen the epistemic status of the members’ beliefs, enable more efective resistance to misleading counterevidence, deepen the kind of considerations on behalf of the matter in question, and so on. In this way, the value of objecting is not in any way reducible to, or equivalent with, the mere changing of particular minds. For even when minds are changed, the depth and breadth of the evidence available to the community can have epistemic benefts that go far beyond its current members. Relatedly, even if it is known that objections will be ignored or dismissed, there is epistemic power in the very act of ofering testimony. If I object to your belief that global climate change is not related to human activity, I have thrust counterevidence upon you, whether you accept it or not. In this way, I have provided you with what is

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

294  Jennifer Lackey sometimes called a normative defeater:21 I have given you evidence that you should take into account, even if you fail to do so. Tis will impact the appropriate epistemic assessment of your beliefs—as being justifed or unjustifed, rational or irrational, and so on. It should also be noted that aims can be both short-­term and long-­term. Suppose, for instance, that two colleagues have the same area of research specialization, but diferent levels of social status—one is a young, female assistant professor, while another is a senior, male distinguished faculty member with numerous publications. Let us stipulate that the senior faculty member has more true beliefs in the area in question simply because he has had more years to do research. If the goal were simply the maximization of true beliefs in the community in the short-­term, it would follow that the senior faculty member is obligated to object nearly every time something is taken to be false or unwarranted because his objections would bring about more true beliefs than those of his junior colleague’s. However, it is important to notice that there are epistemic goals besides the maximization of short-­term ends. Te junior colleague should at least be ofered the opportunity to object in some cases, both to build up a track-­record of credibility and to develop a positive reputation within her professional community. Tis would enable her to make epistemic contributions with appropriate receptivity, which will maximize, say, true beliefs in the long-­term since, for instance, her colleague cannot be everywhere at once, will presumably retire long before she does, and so on. Tus, even if objections from those with lower social status are ofen supererogatory, the overall fourishing of the epistemic community would not be achieved by only the most powerful and expert among us objecting. Efective objections will also be highly sensitive to the best evidence we have about human psychology, and this will be especially relevant in political matters. While contributing to the communal pool of evidence—regardless of whether minds will be changed—is indeed valuable, it must also be factored in whether the objection is likely to make the overall epistemic situation worse of. Suppose, for instance, that one can reliably predict that one’s objection will be wildly polarizing, perhaps because of what is said, by whom it is said, to whom it is said, where it is said, and so on. Political polarization has increased across the globe, and so there are certainly occasions and contexts where it is pretty clear that objecting to a political matter will simply reinforce or legitimate the target view. When the polarization is reasonably determined to be greater than the epistemic beneft of contributing counterevidence to the communal pool, the duty to object can certainly be diminished, even substantially.22 In addition to polarization, there are also considerations of trolling and providing a platform or engagement with views or people that perhaps should not be acknowledged. We ofen hear the advice to not “feed the trolls” or to not give 21  For discussions involving what I call normative defeaters, approached in a number of diferent ways, see BonJour (1980, 1985), Goldman (1986), Fricker (1987, 1994), Chisholm (1989), Burge (1993, 1997), McDowell (1994), Audi (1997, 1998), Williams (1999), BonJour and Sosa (2003), Hawthorne (2004), Reed (2006), Lackey (2008), and Goldberg (2017). What all of these discussions have in common is simply the idea that evidence can defeat knowledge (justifcation) even when the subject does not form any corresponding doubts or beliefs from the evidence in question. 22  See Talisse (2021) for more on polarization.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

When Should We Disagree about Politics?  295 particular speakers a platform, such as racists or bullies. Such input can be understood as not only moral, but epistemic as well. Feeding trolls and debating racists, for instance, may give the appearance of epistemic legitimacy to views that are utterly groundless. We have seen, then, that the duty to object about political matters is an imperfect epistemic one, and that what our share is of this duty is determined by a number of factors, some of which can pull in diferent directions. Tere might be some unhappiness with the messiness or absence of very specifc instructions about when we are obligated to object on such a view, but I regard this as a virtue. Politics is messy, and our ability to promote epistemic ends in our communities should be sensitive to the reality of our lived experiences in the world.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Audi, R. 1997. “Te Place of Testimony in the Fabric of Knowledge and Justifcation.” American Philosophical Quarterly 34: 405–22. Audi, R. 1998. Epistemology: A Contemporary Introduction to the Teory of Knowledge. London: Routledge. BonJour, L. 1980. “Externalist Teories of Epistemic Justifcation.” Midwest Studies in Philosophy 5: 53–73. BonJour, L. 1985. Te Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press. BonJour, L. and E.  Sosa. 2003. Epistemic Justifcation: Internalism vs. Externalism, Foundations vs. Virtues. Oxford: Blackwell Publishing. Burge, T. 1993. “Content Preservation.” Te Philosophical Review 102: 457–88. Burge, T. 1997. “Interlocution, Perception, and Memory.” Philosophical Studies 86: 21–47. Chisholm, R.  M. 1989. Teory of Knowledge, Tird edition. Englewood Clifs, NJ: Prentice-Hall. Cohen, J. 1981. “Who Is Starving Whom?” Teoria 47: 65–81. DesAutels, P. 2009. “Resisting Organizational Power.” In L. Tessman (ed.), Feminist Ethics and Social and Political Philosophy: Teorizing the Non-Ideal. Berlin: Springer, pp. 223–36. Fricker, E. 1987. “Te Epistemology of Testimony.” Proceedings of the Aristotelian Society, supp. vol. 61: 57–83. Fricker, E. 1994. “Against Gullibility.” In B. K. Matilal and A. Chakrabarti (eds.), Knowing from Words. Dordrecht: Kluwer Academic Publishers, pp. 125–61. Fricker, M. 2007. Epistemic Injustice: Power & the Ethics of Knowing. Oxford: Oxford University Press. Goldberg, S. 2017. “Should Have Known.” Synthese 194: 2863–94. Goldman, A.  I. 1986. Epistemology and Cognition. Cambridge, MA: Harvard University Press. Hawthorne, J. 2004. Knowledge and Lotteries. Oxford: Oxford University Press. Hornsby, J. 1995. “Disempowered Speech.” Philosophical Topics 23: 127–47. Kvanvig, J. L. 2003. Te Value of Knowledge and the Pursuit of Understanding. Cambridge: Cambridge University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

296  Jennifer Lackey

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Lackey, J. 2008. Learning from Words: Testimony as a Source of Knowledge. Oxford: Oxford University Press. Lackey, J. 2018. “Te Duty to Object.” Philosophy and Phenomenological Research 101(1): 35–60. Langton, R. 1993. “Speech Acts and Unspeakable Acts.” Philosophy and Public Afairs 22: 293–330. McDowell, J. 1994. “Knowledge by Hearsay.” In B. K. Matilal and A. Chakrabarti (eds.), Knowing from Words. Dordrecht: Kluwer Academic Publishers, pp. 195–224. Munger, K. 2016. “Tweetment Efects on the Tweeted: Experimentally Reducing Racist Harassment.” Political Behavior 39: 629–49. Murphy, L.  B. 1993. “Te Demands of Benefcence.” Philosophy & Public Afairs 22: 267–92. Murphy, L. B. 2003. Moral Demands in Nonideal Teory. Oxford: Oxford University Press. Reed, B. 2006. “Epistemic Circularity Squared? Skepticism about Common Sense.” Philosophy and Phenomenological Research 73: 186–97. Reed, B. 2018. “Practical Interests and Reasons for Belief.” In C.  McHugh, J.  Way, and D.  Whiting (eds.), Normativity: Epistemic and Practical. Oxford: Oxford University Press; 200–20. Schroeder, S. A. 2013. “Imperfect Duties, Group Obligations, and Benefcence.” Journal of Moral Philosophy 11: 557–84. Singer, P. 1972. “Famine, Afuence, and Morality.” Philosophy and Public Afairs 1: 229–43. Tadros, V. 2016. “Permissibility in a World of Wrongdoing.” Philosophy and Public Afairs 44: 101–32. Talisse, R. 2021. “Problems of Polarization.” In E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Tessman, L. 2005. Burdened Virtues: Virtue Ethics for Liberatory Struggles. Oxford: Oxford University Press. Williams, M. 1999. Groundless Belief: An Essay on the Possibility of Epistemology, Second edition. Princeton, NJ: Princeton University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

16

Disagreement or Badmouthing? Te Role of Expressive Discourse in Politics Michael Hannon

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

1. Introduction Is Barack Obama the founder of ISIS? Was he born in the United States? Does the United Kingdom send £350 million to the European Union each week? Would the world be safer if everyone owned a gun? Is global warming going to be catastrophic if we don’t take immediate action? Tese questions provoke strong disagreement. So do plenty of others. Political opponents cannot agree on matters concerning the economy, foreign afairs, education, energy, health care, the environment, privatization, and immigration. In the U.S., nearly half of all Republicans and Democrats say they “almost never” agree with the other party’s positions (Doherty et al., 2016). When it comes to politics, there seems to be no end to the number of issues over which people disagree. Political disagreement is ofen a good thing for a healthy democracy. We expect values and preferences to difer in a pluralistic society, and reasonable citizens understand that people of good will can disagree about moral and political issues (Rawls, 1993). However, partisan disagreements have spread beyond political values and even include disputes about obvious matters of fact (Sinnott-­Armstrong, 2018; Bartels, 2002). In the U.S., for example, Democrats and Republicans disagree about whether the GDP has gone up or down, whether unemployment rates are better or worse, how many immigrants entered the U.S.  illegally, and many other topics. Consider the issue of climate change. Te extent and causes of climate change are scientifc questions that should be settled independently of one’s political beliefs. Yet politics seems to drive our beliefs about the facts instead of the facts driving policy. Tis is a signifcant problem for democratic politics. If partisanship is shaping our perceptions of reality, then democratic decisionmaking becomes incredibly difcult. Without agreement on the facts, voters will be unable to hold representatives accountable, to productively deliberate with others, and to fnd political compromise. But is political disagreement as extensive and deep as many have claimed? In this chapter I will argue that many apparent political disagreements are not genuine disagreements. I draw on three sources of evidence to justify this claim, described below. First, new evidence suggests that voters are increasingly polarized in terms of their attitudes towards each other, even though there has been comparatively little

Michael Hannon, Disagreement or Badmouthing? The Role of Expressive Discourse in Politics. In: Political Epistemology. Edited by: Elizabeth Edenberg and Michael Hannon, Oxford University Press (2021). © Michael Hannon. DOI: 10.1093/oso/9780192893338.003.0017

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

298  Michael Hannon polarization on the issues. As Lilliana Mason (2018) puts it, politics is increasingly characterized by “uncivil agreement.” Voters have grown more partisan, angry, and biased against their political opponents, but these reactions have almost nothing to do with one’s opinions about the issues. We are simply behaving as if we disagree. Second, survey data tends to exaggerate the level of disagreement in politics. When surveyed about political issues, people ofen deliberately misreport their beliefs as a way to express their attitudes. Tis is called “expressive responding.” People engage in this behaviour for one of two reasons: either partisans know the truth but prefer to “cheerlead” when there is nothing to gain from accuracy, or they are ignorant on the issue and they ofer a congenial answer as their best guess. Either way, survey responses are not entirely sincere. Tus, we should not interpret these responses as evidence that partisans are unable to agree. Tird, voters ofen claim to have policy convictions when, in fact, they do not have robust political beliefs at all. Here I follow the democratic “realists” who argue that people vote largely on the basis of partisan loyalties, not sincere policy preferences (Achen and Bartels,  2016). Although many citizens will describe themselves as “liberal” or “conservative,” they actually lack stable beliefs ftting these ideological self-­descriptions. What seems like deep political disagreement is actually superfcial and inauthentic. All this suggests that political disagreement is neither as deep nor as extensive as many have thought. What follows from this fact? Tere are several signifcant implications that I will explore in the second half of this chapter. For example, I will argue that insincere disagreement explains why debates ofen go so poorly, why people seem to hold blatantly contradictory beliefs, and why it is ofen so difcult to correct false beliefs. I will also discuss some positive and negative implications of the idea that political disagreement is ofen illusory. On the positive side, I will argue that political surveys tend to overstate the level of political misinformation; that motivated reasoning is not distorting our perceptions of reality as ofen as many scholars claim; and that there is less disagreement over the facts than survey data suggest. Tis is good news. Te bad news is that voters are not supporting policies based on their actual content; we cannot decrease polarization by reasoned debate; and people are not genuinely interested in engaging with their political opponents.

2.  Division without Disagreement According to a common view of politics, voters tend to choose the political party that best matches their own interests and issue positions. Tis is the “folk theory of democracy” (Achen and Bartels, 2016). Imagine a voter who thinks that corporate and individual tax cuts will help create a booming economy and that there is a critical need for an immigration policy that secures the border and limits migration. Such a person is likely to vote Republican, according to the folk theory, because the Republican Party will better represent this voter’s interests and values. Te folk theory predicts that party afliation is strongly linked to issue positions, since voters will choose to support whichever party best refects their political preferences.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Disagreement or Badmouthing?  299 Te folk theory of democracy is intuitive, rational, and widely accepted. It is also largely incorrect. In Uncivil Agreement, Lilliana Mason argues that many citizens do not base their voting decisions primarily on the proximity of the policy positions of parties to the voters’ own positions. Rather, their decisions are based on social identity. A “social identity” involves

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

a subjective sense of belonging to a group that is internalized to varying degrees, resulting in individual diferences in identity strength, a desire to positively distinguish the group from others, and the development of ingroup bias. (Huddy et al., 2015, p. 3)

Social identity theory is a powerful foundation to study partisanship and political behavior. On this model, the strength of a person’s partisanship can derive from a number of infuences that have nothing to do with political issues; in particular, partisan strength seems to be rooted in social group memberships, social networks, and cultural identity. Tese infuences may increase the strength of partisanship without any corresponding increase in the extremity of issue positions (Mason, 2018 and Iyengar et al., 2012).1 Tis captures the current state of American politics. On the one hand, Democrats and Republicans are increasingly polarized in the sense that they dislike each other more than ever, ascribe negative traits to members of the other side, and even claim that they would be upset if their children married someone from the other party (Iyengar et al., 2012). On the other hand, there has been almost no increase in the extremity of issue positions in the mass public (Iyengar et al., 2019; Iyengar et al., 2012; and Fiorina et al., 2008).2 To borrow some terminology from Iyengar and his colleagues (2012), there has been an increase in “afective polarization” without a corresponding increase in “issue-­ based polarization.” Afective polarization is characterized by increased partisan strength, partisan bias, political activism, and anger, while issue-­based polarization occurs when citizens move from moderate issue positions to more extreme ones. Although there has been increased issue-­based polarization at the level of political elites (e.g. members of Congress), there is little evidence that ordinary citizens increasingly disagree on the issues.3 Rather, they simply dislike, even loathe, each other more. 1  “Extremity” refers to a person’s movement from a moderate policy position to a less moderate one, as well as the strength with which they hold their view (Mason, 2013, p. 142). 2  Abramowitz and Saunders (2008) contest this claim. See Fiorina et al. (2008) for a reply. 3  According to the Pew Research Center (2014), the partisan gap has increased on some issues over the past few decades. For example, when asked whether “government regulation of business usually does more harm than good,” the partisan gap has more than doubled from 1994 to 2004. I will not deny that there is evidence of a partisan gap on some issues. However, these diferences may not refect changes in genuine beliefs (as I will argue in Section 3). Rather, these changes refect changes in the social returns of cheerleading for one’s political team. Tis fts nicely with the recent empirical work showing there has been a large increase in afective polarization. As the strength of partisanship increases, people become more motivated to engage in cheerleading behavior (Bullock et al., 2015). Tis also explains why the partisan “gap” is most pronounced during campaign seasons, since elections may make more salient the need to support one’s party (Iyengar et al., 2012).

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

300  Michael Hannon Tis suggests that partisans have grown increasingly distant and hostile toward each other even though their policy disagreements are not profound. In fact, many Democrats and Republicans do not difer that much on matters of public policy. Rather, they continue to hold relatively inconsistent policy attitudes, and have done so for decades (Converse, 1964; Achen and Bartels, 2016; Kinder and Kalmoe, 2017; Mason, 2018). For example, Mason measured Americans’ preferences on six major issues—immigration, the Afordable Care Act, abortion, same-­sex marriage, gun control, and the relative importance of reducing the defcit or unemployment—and found that identifying as a “liberal” or “conservative” only explained a small part of their issue positions. It seems that people are polarized by labels such as “liberal” and “conservative” (and what they imagine their opponents to be like) more than they are by actual disagreements over topics like immigration, abortion, and taxes. As a result, Americans have become angrier with their political opponents while not disagreeing with them on many issues. Why, then, do liberals and conservatives hate each other so much? If these feelings are not rooted in policy disagreements, what are they based on? Te proliferation of partisan news outlets and high-­choice media environment is frequently blamed for the current polarized environment (Lelkes et al.,  2017). But this explanation might get things backwards: those who are the most polarized are also more motivated to watch partisan news; thus, partisan news may not be the cause of afective polarization.4 It is also common to blame the Internet and social media for echo chambers, flter bubbles, and polarization; but the relationship between Internet access and afective polarization has also been contested (Boxell et al., 2017). Whatever the exact causes of polarization might be, it is widely acknowledged that the strengthening of partisan identities has little to do with the issues and almost everything to do with group loyalty and party identity. Once we identify with a particular party, we are highly motivated to protect and advance our group’s status. Tis is identity politics at its worst. Democrats and Republicans tend to hate each other but this hatred has almost nothing to do with their opinions on the issues. Tey dislike the other team simply because they are the other team. As a consequence, we have an electorate that is increasingly divided and raring to fght, yet there is a lack of any substantive policy reasons to do so. Although Democrats and Republicans conceive of themselves as disagreeing over substantive issues like tax policy, health care, and government regulation, the political confict in America today is not really about these things. It is rather about team identifcation and winning for its own sake. I will return to this topic in Section 4.

4  See Arceneaux and Johnson (2013) and Prior (2013). In contrast, Levendusky (2013) fnds that exposure to partisan news makes those with extreme attitudes even more extreme. It seems plausible that partisan media consumption can be part of a feedback loop that increases afective polarization. Tat is, afective polarization gets started via other channels, but media consumption and online “echo chambers” exacerbate it.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Disagreement or Badmouthing?  301

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

3.  Political Expressivism As section 2 illustrates, people ofen behave as though they disagree even when there is not much disagreement. Tis is because partisanship can afect our attitudes towards others without necessarily afecting our beliefs about the relevant issues. Partisanship can also lead people to say things that create the appearance of disagreement. In particular, people may deliberately misreport their political beliefs as a way to express their attitudes. Tis is called “expressive responding” or “cheerleading” (Bullock et al., 2015). Sometimes it is obvious that partisans are just cheerleading. Consider what Trump supporters say when asked to compare photos of his 2017 inauguration crowd and Obama’s in 2009. In a survey of almost 700 American adults, participants were shown a crowd picture from each inauguration and asked a very simple question: “Which photo has more people?” Although only one answer is clearly correct, Trump supporters were seven times more likely (compared to Clinton voters) to say that the half-­empty photo of Trump’s inauguration had more people. Trump supporters with college degrees were the most likely to answer incorrectly: 26 per cent of them gave the clearly wrong answer. Do these people really believe there are more people in the obviously half-­empty photo? It would be mistaken to interpret their responses in this way. Instead, some Trump supporters clearly decided to express their support for Trump rather than to answer the question factually. As Elizabeth Anderson (Chapter  1 in this volume) puts it, “it is a way of showing those smug liberal academics [who were conducting the survey] that Trump voters will stand their ground in repudiating insults toward their group.” Tey were not making a factual claim; their answer was expressive. Expressive responding may also explain why approximately one in seven Americans will say that Obama is “the antichrist.”5 Do these people really believe this? Maybe some do. But a likely scenario is that such reports ofen refect partisan cheerleading rather than genuine belief. How ofen do people misreport their beliefs? Probably a lot. A seminal fnding of new research in political behavior is that what seems like factual disagreement is ofen just partisan cheerleading or badmouthing. For example, John Bullock and colleagues (2015) fnd that partisans tend to give more accurate (and less partisan) responses to politically charged questions when ofered monetary incentives to do so. As a result, the gap between Democrats and Republicans in response to factual questions sharply decreases. More specifcally, small payments for correct answers reduced partisan divergence by at least 60 percent. Tey reduce by 80–100 percent when participants are paid both for correct responses and a smaller amount for admitting they do not know the correct response. Te dramatic efects of a small incentive for accuracy—or a smaller incentive to admit that one does not know the answer—suggests that survey responses ofen refect “the expressive value of making statements that portray one’s party in a

5  https://www.theguardian.com/world/2013/apr/02/americans-­o bama-­a nti-­c hrist-­c onspiracy­theories.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

302  Michael Hannon

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

favorable light” (Bullock et al., 2015, p. 521). As Gary Langer, former chief pollster for ABC News, aptly remarks: “some people who strongly oppose a person or proposition will take virtually any opportunity to express that antipathy . . . not to express their ‘belief,’ in its conventional meaning, but rather to throw verbal stones.” Tis fnding is supported by multiple independent studies.6 Markus Prior and colleagues (2015) asked members of the public about objective economic conditions, such as whether the level of employment has gotten better or worse. Tey found that supporters of the current president’s party tended to report more positive economic conditions than its opponents, but this tendency was signifcantly reduced when survey-­takers were fnancially motivated to answer factual questions accurately.7 If these survey responses refected actual beliefs, then paying partisans to answer correctly should not afect their responses. Yet it does. Te observed gaps between Democrats and Republicans are substantially reduced with relatively small payments. Tis suggests that partisans “do not hold starkly diferent beliefs about many important facts” (Bullock et al., 2015, p. 522). Further, it indicates that partisans have the capacity to acknowledge inconvenient truths and are willing to report them when motivated to do so. Without adequate incentives, however, the motivation to give an answer that supports one’s political party may outweigh the motivation to give an accurate response. Tis can be for one of two reasons: either survey-­takers have accurate but uncongenial information and they prefer to give congenial but inaccurate answers, or they are ignorant on the issue and they ofer a congenial answer as their best guess.8 Either way, survey responses are not revealing misinformation or political disagreement. Tis fnding is incredibly signifcant. Public opinion polls are consistently showing that partisans are unable to agree on the facts. For example, Republicans are more likely than Democrats to say that the defcit rose during the Clinton administration; 6  See also the studies by Huber and Yair (2018) and Khanna and Sood (2018). Peterson and Iyengar (2020) found that partisan cheerleading infates divides in factual information, but only modestly. Tis is consistent with my general hypothesis that the extent and depth of political disagreements is likely overstated; I do not claim that political disagreements are entirely or primarily insincere. Moreover, this study may underestimate the extent of cheerleading by failing to adequately incentivize participants in the treatment condition. As the authors note, their study used a smaller fnancial incentive than the higher rewards in previous studies; further, this study assessed responses to issues that were especially divisive in nature. As a result, incentives for accuracy may have actually provided people with an opportunity for more powerful expression. When someone wants to express strong support for a politician or political party, they may do so by spurning the opportunity to receive monetary compensation for reporting their true belief. Work by Philip Tetlock (2003) suggests that “sacred values” are usually held to be incommensurable with—and tainted by—fnancial reward. Tus, if people count support for a politician or party as a sacred value, they will likely reject the opportunity for fnancial reward in exchange for accuracy. 7  An alternative explanation, suggested by Neil Levy and Robert Ross, is that respondents are simply pretending to hold certain beliefs in order to get the reward. On this interpretation, these studies show that people are aware of the mainstream view and will claim to believe it when incentivized to do so. While I agree this is an important concern, it does not obviously apply to the studies I am discussing. In the cases of Prior et al. (2015) and Bullock et al. (2015), participants were provided with a mix of questions that favored both Republicans and Democrats, and where there was no obvious “mainstream view” regarding the answers (e.g. “Is the economy doing better or worse?,” “Has unemployment increased or decreased?,” etc.). 8  See Bullock et al. (2015) and Prior et al. (2015) for evidence of the frst explanation, and Luskin et al. (2013) for evidence of the second explanation.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Disagreement or Badmouthing?  303 Democrats are more likely than Republicans to say that infation rose under Reagan (Bartels, 2002, p. 519). Similar patterns exist when they are asked factual questions about healthcare (Nyhan and Reifer,  2010), foreign policy (Jacobson,  2010), and social services (Jerit and Barabas,  2012), among other issues. When people are surveyed about factual issues, they are more likely to report having beliefs that are favorable to their existing beliefs and attachments than beliefs that are unfavorable. Tese patterns are ordinarily taken as evidence that partisanship afects factual beliefs about politics.9 Democrats and Republicans are allegedly seeing “separate realities” (Kull et al., 2004). An alternative explanation is that such patterns merely refect a desire to praise one party or condemn another. Tus, instead of assuming that the public is misinformed, we should assume that the public is misinforming us. In doing so, these survey responses mask shared, bipartisan beliefs about factual matters. Tis is likely not just an American phenomenon. Nearly half of the British public still claim to believe that the UK sends £350m to the EU each week, despite p ­ ersistent attempts to debunk this myth.10 A new study by the Policy Institute at King’s College London found that 42 percent of people who had heard the claim still believe it is  true, whilst only 36 percent thought it was false, and 22 percent were unsure. According to this study, conservative voters and Brexit voters are particularly susceptible to the misinformation, with 54 percent and 61 percent of each buying the claim. If we take these fgures at face value, we are led to conclude that nearly half of the British public continues to be misinformed about the issue because they continue to believe that this claim is true. Tis is precisely how Professor Bobby Dufy, director of the policy institute that carried out this research, interprets these results. In an interview with Te Independent, he says, “Tese misperceptions raise important questions about the basis of our decision-­making . . . the fact that diferent groups see the same realities so diferently shows how divided we are.” Are these people seeing the same reality diferently? As I have suggested, we should be wary of taking answers to factual questions with partisan implications at face value, since they are ofen contaminated by the motivation to root for one’s team. People believe one answer, but they give a diferent answer to support their party. It represents nothing more than partisan bad-­mouthing and the joy of cheerleading.

4.  Voters without Beliefs So far I have discussed two reasons why the extent and depth of political disagreement may be overstated. First, increasing levels of polarization tend to refect our attitudes towards our political opponents but not necessarily our policy preferences or issue positions. Second, the partisan gap revealed by survey data sometimes refects team cheerleading and cheap talk, not genuinely held political beliefs.

9  Tose who take survey responses at face value include Campbell et al. (1960), Kull et al. (2004), Jerit and Barabas (2012), Shapiro and Bloch-­Elkon (2008), and Jacobson (2010). 10  https://www.kcl.ac.uk/policy-­institute/research-­analysis/the-­publics-­brexit-­misperceptions.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

304  Michael Hannon I now want to consider a third—and perhaps more radical—reason to think that political disagreement is ofen illusory. Te idea, briefy put, is that voters typically do not have many stable political beliefs. Indeed, many of their political “beliefs” may not be beliefs at all.11 Politics today is largely identity politics: citizens vote primarily on the basis of partisan loyalties that are grounded in social identity, not sincere policy preferences.12 Christopher Achen and Larry Bartels defend this view, which they call “political realism.” Tey describe the view as follows:

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

voters choose political parties, frst and foremost, in order to align themselves with the appropriate coalition of social groups. Most citizens support a party not because they have carefully calculated that its policy positions are closest to their own, but rather because “their kind” of person belongs to that party. (Achen and Bartels, 2016, p. 307)

Te true psychological basis for voting behavior, they claim, is not individual ­political preferences but group identity. People do not frst identify the issues they care about and then align themselves with the political party that best refects these preferences. Rather, people frst identify themselves with a group and then vote according to this group allegiance. For realists, politics resembles sports and voters are like sports fans. Consider the Toronto Raptors (a basketball team) and their fans. People from Toronto—like myself—become Raptors fans because we are from Toronto. We do not frst form strong opinions about basketball and then examine the teams on ofer and fnally select a team to support based on how well that team realizes our pre-­existing preferences or values.13 Rather, we cheer for the Raptors because that team is connected to our demographic identity. On the realist view, political afliation is psychologically equivalent to sports team loyalty (at least for most citizens). Just as one’s loyalty to a particular basketball team is not a refection of one’s pre-­existing ideological commitments about basketball, one’s political afliation is also not a refection of one’s political, moral, or otherwise ideological commitments. Rather, the typical voter 11  See Hannon and de Ridder (2021) for a more detailed discussion. 12  Tis claim difers from the explanations ofered in the previous two sections. In Section 2, I argued that people largely agree on the issues even though they dislike each other. In Section  3, I argued that partisan cheerleading masks shared beliefs. Now, I am exploring the idea that partisans ofen lack stable beliefs about political issues; they are what Mason (2018) calls “ideologues without issues.” As a result, there is some tension between the claims defended in the previous two sections and the claim I am now exploring. In particular, Sections 2 and 3 imply that political opponents do have shared beliefs about many issues, whereas I am now suggesting that partisans may not have beliefs on many political issues. Let me provide two quick replies to this worry. First, we needn’t view these explanations as all or nothing. People like Mason may be right that when citizens do have beliefs about political issues, these beliefs ofen overlap with the views of their opponents; but it may also be true that many apparent political beliefs are not genuine beliefs. It simply depends on which beliefs we are investigating. Second, these explanations need not be compatible for my argument to go through. I may be interpreted as ofering a variety of plausible explanations for illusory political disagreement, but it is not necessary for my conclusion—that many apparent political disagreements are not genuine disagreements—that these explanations are compatible with each other. 13  I borrow this type of example from Brennan (Chapter 8 in this volume). Somin (2013, pp. 78–9) also likens voting behavior to that of sports fans.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Disagreement or Badmouthing?  305 becomes attached to a political “team” largely due to accidental historical circumstances.14 Tey are simply born into it, just as children typically inherit the religious afliations of their parents.15 Tis point calls for two important qualifcations. First, some voters may initially get their political allegiances by enculturation, but this does not preclude them from genuinely coming to believe some (or a lot) of the things their political party stands for. Tis may occur through a process of refection on the attitudes they inherited from their community. Second, political realism is implausible as an account of political elites and people who devote their careers to politics. Tese people surely have genuine beliefs about the issues. My point, however, is that many people are not like this. According to a vast range of empirical research, the typical voter in the U.S. today resembles a sports fan.16 Admittedly, many citizens will describe themselves as “conservative” or “liberal.” But most people actually lack stable beliefs ftting these ideological self-­descriptions. Tey are not deeply committed to their proclaimed ideologies. As Jason Brennan (Chapter 8 in this volume) puts it,

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Teir beliefs, if we can even call them that, do not refect real commitments about the good or just, or about how the world works. Instead, expressing political “beliefs” is largely equivalent to wearing sports team colors; they are a form of conspicuous display intended to show membership in what are, for that voter, socially advantageous groups. For them, advocating a policy is like wearing the Patriots’ blue and silver or waiving the Steelers’ terrible towel. . . . Teir commitment is no deeper than Patriots fans’ commitments to blue, red, and silver; if the parties were to change policy platforms, most of their “ideological” voters would claim they agree with the changes, and some would claim they believed such views all along.

To illustrate, Brennan asks us to consider how many Republicans switched their “views” on numerous economic issues when Trump was elected.17 Tey went from pro-­free trade to protectionist very quickly afer the 2016 election.18 Tis suggests that even apparently key issues like free trade are, at bottom, just proclamations intended to demonstrate group membership. Similarly, when Democrats say “I’m pro-­regulation,” they may not be expressing a belief about appropriate responses to

14  Brennan (2017; this volume), Achen and Bartels (2016, pp. 213–66), and Campbell et al. (1960) all defend this claim. 15  Tis is substantiated by a vast amount of empirical work. See Greenstein (1965), Jennings and Niemi (1981), and Sears (1983). 16  Te analogy with sports fans is not perfect: there seem to be important meta-­cognitive diferences between sports fans and politically partisan individuals. In particular, sports fans are likely aware of the fact that many of their belief reports are just instances of expressive responding (e.g. “Manchester is the best team ever!”) and that they pick what teams they support on the basis of fairly arbitrary factors (e.g. upbringing). Tanks to Robert Ross and Neil Levy for making this point, which deserves to be explored in more detail (elsewhere, alas). 17  https://www.vox.com/science-­and-­health/2017/11/17/16585982/psychology-­memory-­polls-­trump 18  According to Pew (2017), Republican support for free trade dropped from 56 percent in 2015 to 36 percent in 2017.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

306  Michael Hannon market failure; rather, they are just expressing a commitment to seeing the Democrats win. Te implicature is “Hurray, Democrats!” (Brennan Chapter 8 in this volume). Indeed, people will support whatever policy or platform they think is backed by their party. Geofrey Cohen’s (2003) work provides a striking example of this. He ran a study in which participants were told about two welfare programs: a harsh (extremely stingy) welfare program and a lavish (extremely generous) one. When Democrats were told that their in-­group party supported the harsh policy, they approved of it. When Democrats were told that their party supported the lavish policy, they approved of that instead. Te same thing happened with Republican participants. All that mattered was which party was said to support the program; it made little diference what the actual content of the policy was.19 Moreover, the participants were unaware of this bias. When asked to justify their decision, the participants felt they were responding to the program’s objective merits and insisted that party considerations were irrelevant. Relatedly, studies have found that people are unable to justify political positions they claim to feel strongly about. For example, people who claim to believe deeply in cap and trade ofen have little idea about what these policies actually entail (Brennan Chapter  8 in this volume; Fernbach et al., 2013). What looks like the expression of a genuine belief is just a proclamation intended to display partisan afliation. What does all this mean for political disagreement? It means that when a typical Democrat or Republican expresses their “beliefs” about some political issue, we should not necessarily regard this as evidence that they genuinely disagree with their opponents. Rather, this behavior may be symbolic, expressing loyalty to their political team and a desire to see that team win, without any deeper commitment to what that team stands for. Tus, many people cannot quite be said to “believe” in their party’s ideology or platform; they only express commitment to it. Consequently, their ideological “disagreements” are superfcial and inauthentic. You might insist that these states of mind still qualify as beliefs. Perhaps voters do sincerely believe, say, that government regulation is good, even if they are unable to say anything else about the issue. Similarly, in the case of expressive responses (see Section 3), you could think that voters believe one thing and then, when promoted by fnancial incentives, they revise and update their beliefs. People may simply reconstruct their beliefs from moment to moment. Perhaps there is a sense in which we can call these mental states “beliefs.” Tink of them as fragile beliefs. Whether or not fragile beliefs really are beliefs is a verbal dispute that I would prefer to avoid. My worry is that these “beliefs” would not be sufciently robust to serve as a satisfactory basis for democratic theory. A much thicker notion of belief lies at the heart of liberal democratic theory. Tis is evident from the frst page of Robert Dahl’s (1971) classic book, Polyarchy. He writes, “a key characteristic of a democracy is the continued responsiveness of the government to 19  One might suggest that this is the result of the following heuristic: voters know that a particular party shares their general ideological outlook, so they use this as a way to determine what they would think about other issues. However, Gabriel Lenz (2012) tested this hypothesis and found little support for it.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Disagreement or Badmouthing?  307

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

the preferences of its citizens.” But the aim of a liberal democracy is surely not to aggregate just any wishy-­washy individual preferences into collective choice, nor to include those that are merely expressive proclamations. It is for a similar reason that political theorists worry about framing efects on political preferences, which seem to show that individuals do not have “real political preferences” (Kelly, 2012, p. 21). As Bartels (2003, p. 49) writes, the beliefs of citizens “are not sufciently complete and coherent to serve as a satisfactory starting point for democratic theory, at least as it is conventionally understood.” Liberal democratic theorists have assumed that people have defnite preferences and that a democratic government must respond appropriately to those (real) preferences. Tis is why Philip Converse, in his foundational work on the nature of belief systems in mass publics, said that citizens “do not have meaningful beliefs” on many political issues (1964, p. 245). Te instability of fragile “beliefs” (if we choose to call them that) would therefore still create signifcant problems for attempts to discern public preferences on political issues. Tus, the distinction between a thicker notion of belief and a thin sense of belief is not merely a theoretical nicety. Further, contemporary philosophers of mind generally use the term “belief ” to refer to the attitude we have when we regard something as true. As revealed by the literature on expressive responding, however, citizens who make factual claims do not actually regard the content of these assertions as true. Tus, expressive responses would not qualify as beliefs even according to the standard defnition of “belief ” in contemporary Anglophone philosophy. To be clear, I am not saying that people never choose political parties on the basis of their beliefs or policy preferences. My claim is only that in a vast range of cases, the folk theory of democracy does not ft the general picture of political reasoning that emerges from decades of empirical research across multiple disciplines. Tis literature suggests that an adequate account of partisan identity must give a central role to social identity, tribal allegiances, and expressive discourse.

5. Implications What are the implications of this research? Suppose I am right that there is less disagreement in politics than we have been led to believe. What follows? In the remainder of this chapter, I will argue that we can explain some puzzling aspects of contemporary political life by appealing to the hypothesis that political disagreements are ofen illusory and merely the refection of identity-­expressive discourse. In particular, I will argue that this hypothesis sheds light on the following issues: • why people seem to disagree despite access to unequivocal evidence; • why attempts to correct false beliefs sometimes backfre; • why political debates ofen go so poorly; • why it can be difcult to correct mistaken beliefs; • why disagreement ofen seems irresolvable; • why people are unaware of the most basic facts about political issues on which they express opinions;

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

308  Michael Hannon

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

• why people ofen seem to hold blatantly contradictory beliefs; • why people ofen do not read the news posts they share; • why motivated reasoning is sometimes a poor explanation for polarization and misinformation; • why it is a mistake to classify many political disagreements as “deep disagreements.” Let us start with why people seem to disagree about well-­established facts even when the evidence is unequivocal and easy to access. Tis behavior might seem illogical, but my account provides a straightforward explanation: people are simply making claims about factual issues to signal their allegiance to a particular ideological community. When a Republican says that Trump’s inauguration photo has more people, they are not actually disagreeing with those who claim otherwise. Tey are just cheerleading. Relatedly, the use of factual claims for tribal signaling or identity-­ expression may explain why attempts to correct false beliefs sometimes appear to backfre (Nyhan and Reifer, 2010). If factual corrections are interpreted as challenges to our “team,” the correction will seem to “backfre” because people will reply by expressing their loyalty. Tis also helps us understand why debates ofen go so poorly. According to an optimistic view, political disagreement is a good thing because it allows citizens to encounter diverse perspectives, consider the value of alternative points of view, and evaluate their opinions in light of counterarguments. For these reasons, theorists such as Aristotle, Mill, Dewey, and Arendt extolled the benefts of deliberation and disagreement in politics. Why, then, do real life political disagreements swifly devolve into heated partisan rancour? It is because these disputes are not generally treated as opportunities to exchange reasons or make arguments. Rather, they are opportunities for cheering and booing. Recall the analogy with sports. When fans cheer for their team, this is not an exercise in rational deliberation. Tey are just expressing loyalty to their team. If political disagreement is similarly tribal, then we should view partisan claims about immigration, health care, and the like in a similar light. Tey are not conclusions articulated on the basis of reasons, but rather proclamations akin to “Yay, team!” and “Boo, the other guys!” As such, political disputes are not aimed at rational resolutions. And when partisans do give reasons or arguments for their views, these are most likely the product of rationalization, confabulation, and post-­hoc reasoning (Haidt, 2012). As the empirical evidence shows, political attitudes are ofen not the products of careful reasoning. Further, it might be true that they are not supposed to be.20 When people cheer for the L.A. Lakers or Toronto Raptors, this is not supposed to be an exercise in rational deliberation. It would be misguided to complain that a Raptor’s fan’s enthusiasm for his team does not refect a sober appraisal of the team’s recent performance. Tat would miss the point. If politics is also about expressing team loyalty, then complaining that someone’s views on global warming are not grounded in the facts may also be of the mark. Teir views on these issues are not articulated conclusions but expressive proclamations. 20  Bloom (2016, p. 236) makes this argument.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Disagreement or Badmouthing?  309 In reply, one might argue that even if politics is not typically an exercise in rational deliberation, it ought to be. We should treat political disagreements as an opportunity to exchange reasons, consider the value of alternative perspectives, and evaluate our opinions in light of counterarguments. So even if politics does resemble sports, it should not. We can therefore criticize individual citizens for failing to meet this standard—or so the objection goes. However, political views may share another common property with views about sports teams—their truth does not really matter. As Paul Bloom observes,

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

If I have the wrong theory of how to make scrambled eggs, they will come out too dry; if I have the wrong everyday morality, I will hurt those I love. But suppose I think that the leader of the opposing party has sex with pigs, or has thoroughly botched the arms deal with Iran. Unless I’m a member of a tiny powerful community, my beliefs have no efect on the world. Tis is certainly true as well for my views about the fat tax, global warming, and evolution. (Bloom, 2016, p. 237)

While this idea is not uncontroversial, it has many defenders. Te idea is that voters are “rationally ignorant” (Somin,  2013; Bloom,  2016; Brennan,  2017). A failure to gather evidence, attend to data, and consider counterarguments in the political domain does not refect stupidity, laziness, or irrationality. It refects how many of us make sense of politics: we care more about team loyalty than the truth because, for us, politics is not really about truth (Bloom, 2016, p. 237). Tis would explain why people are unaware of even the most basic facts about the issues, policies, and politicians that they express opinions about. It would also explain why it is ofen so difcult to correct false beliefs (or “beliefs”). If our political claims are expressions of cheerleading, then providing partisans with correct information may do little to change their minds (Bullock and Lenz, 2019). When our views are not based on the facts or aiming at truth, we should not expect them to be rationally revisable in light of the evidence or reasoned argumentation.21 Tis would also explain why disagreements ofen seem irresolvable: we cannot resolve issues when there is no genuine disagreement. Te tendency to signal allegiance by making claims about factual issues would also explain why people ofen seem to hold blatantly contradictory beliefs. David Dunning (2014) surveyed roughly 500 Americans and found that over 25 percent of liberals (but only 6 percent of conservatives) endorsed both the statement “President Obama’s policies have already created a strong revival in the economy” and “Statutes and regulations enacted by the previous Republican presidential administration have made a strong economic recovery impossible.” Both statements are pleasing to liberals and thus may induce cheerleading behavior, even though it is impossible for Obama to have already created a strong recovery that Republican policies have rendered impossible. Among conservatives, 27 percent (compared to just 10 percent of

21  I am not saying that arguments and evidence never resolve political disagreements; only that they are ofen inefective.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

310  Michael Hannon liberals) agreed both that “President Obama’s rhetorical skills are elegant but are insufcient to infuence major international issues” and that “President Obama has not done enough to use his rhetorical skills to efect regime change in Iraq.” Tis seems contradictory: if Obama’s rhetorical skills are insufcient, why should he be criticized for not using them to infuence the Iraqi government? Dunning and Enns explain these results by appealing to psychological bias. Alternatively, we may regard these contradictory claims as expressions of attitude rather than empirical assertions. Te idea that political disagreements are ofen superfcial has a lot of explanatory power. Once we realize that political disagreement is ofen illusory and merely a refection of identity-­expressive discourse, we can make a lot more sense of our current political context. Our tendency to engage in identity-­expressive discourse would also explain why people do not ofen read the news posts they share. Just as the primary function of certain claims about factual issues is to signal allegiance, the primary function of the communicative act of news-­post sharing may also be expressive (Lynch, 2019).

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

6.  Additional Upshots If this argument is correct, it has several additional upshots. First, it suggests that standard survey research methods are fawed.22 Te bulk of survey research assumes that respondents provide truthful answers when asked questions. If this assumption were correct, then conventional survey methods would provide evidence of real and deeply held diferences in assessments of political facts. But as the literature on expressive responding makes clear, these partisan gaps are ofen illusory. Te appearance of factual disagreement in politics is, to some extent, an artifact of survey measurement. Admittedly, we do not know the precise extent to which voter “misinformation” and factual “disagreement” are merely a refection of identity-­expressive discourse. Nevertheless, the research I have surveyed strongly indicates that many alleged disagreements are not real disagreements, and that what seems like a misinformed public is not in fact so. Second, the appeal to “motivated reasoning” is sometimes incorrect. Tis theory says that people who come across the same information will walk away with diferent beliefs about what the evidence supports, since partisanship leads us to process factual information in biased ways (Taber and Lodge, 2006). Tis is by far the commonest explanation for why voters are misinformed and increasingly polarized. However, the theory of motivated reasoning presumes that the misinformation documented by survey researchers is an accurate refection of what individuals believe. I have tried to throw this assumption into doubt. When Trump voters point to a half-­empty inauguration crowd photo and say it has more people than an obviously full photo, this is not the result of a cognitive processing error. Motivated reasoning has not led them to believe that which fies in the face of

22  Bullock et al. (2015) and Prior et al. (2015) make this point.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Disagreement or Badmouthing?  311

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

unambiguous photographic evidence. What looks like motivated reasoning is just political cheerleading. Tis also explains why political “misperceptions” are the most common among the most political engaged. It is commonly assumed that the most strongly partisan are also the most likely to be biased and engage in motivated reasoning. Tis may be true, but this explanation ignores the fact that the most strongly partisan are also the most likely to engage in cheerleading and other forms of expressive responding. Tus, the theory of motivated reasoning may ofen misdiagnose partisan cheerleading as biased cognitive processing. Tis leads me to another point. It is commonly said that many political disagreements are “deep disagreements” (de Ridder Chapter  12 in this volume; Lynch, 2010; Kappel, 2018; Aikin, 2019). A deep disagreement occurs when two or more people not only disagree about the facts, but also disagree about how best to form beliefs about those facts. For example, two people may disagree about the causes and consequences of climate change because they have diferent underlying commitments about what counts as good evidence, how to weigh diference sources of evidence, who the experts are, and so forth. When the disputing parties have fundamentally diferent epistemic commitments, it will be difcult (if not impossible) to reach a rational resolution. According to Klemens Kappel (2018), most societal disagreements are deep disagreements. Tis allegedly explains why political disagreements are so intractable. However, I have argued that many cases of political disagreement are not genuine disagreements; thus, they cannot be deep disagreements. While I do not dispute the claim that political disagreements may sometimes be “deep,” many of them may not be. Instead of thinking of these as deep disagreements, we should instead think they are relatively shallow. Tis would also explain why these disputes are not rationally resolvable. As I argued above, political disputes are ofen not aimed at rational resolutions; they are just opportunities to root for one’s team.

7.  Good News and Bad News Tese conclusions have a variety of positive upshots. Here’s one: the extent to which voters are misinformed is overstated. Although a large number of people will say that Obama is the antichrist, that he founded ISIS, and so forth, many of these people do not genuinely believe these things. Our worries about voter incompetence have been partly driven not by voter misperceptions but rather by our misperceptions about voters. Tis is reassuring. A person’s willingness to occasionally disregard factual information is far less pernicious than being misinformed, since genuinely believing incorrect information would preclude doubt and obstruct the attainment of truth. I do not deny that voters are ofen ignorant.23 One of the best-­established fndings from decades of research in political theory is the extent to which ordinary citizens 23  It is common to distinguish ignorance (a lack of information) from misinformation (false or inaccurate information.). When the public is misinformed, then tend to confdently hold false beliefs. As a result, misinformation is ofen a greater obstacle to educating people with facts.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

312  Michael Hannon are ignorant of politics.24 Indeed, the studies on expressive responding by Bullock and his colleagues reveal that people will ofen admit their ignorance on many factual political questions. But this, too, is reassuring. When partisans are motivated to give accurate responses, they display the capacity to acknowledge their own ignorance.25 Te problem is that people are unwilling to admit their lack of knowledge under ordinary survey conditions. In addition to overstating the actual amount of factual disagreement in politics (and the extent to which voters are misinformed), survey responses may also exaggerate the degree to which partisanship distorts our perception of the facts. As mentioned earlier, the theory of “motivated reasoning” assumes that the misinformation documented by survey researchers accurately refects what voters truly believe. But this is an inadequate diagnosis in many cases. It turns out that citizens have the capacity to perceive reality in a less partisan way than many have claimed. Tis may be good news for democracy. If there were genuine disagreement over basic factual issues, then the possibility of democratic deliberation and compromise would be slim.26 In short, people might not be as dumb or as biased as is commonly assumed. What appears to be stupidity or irrationality is ofen just cheerleading. Now for the bad news. An unfortunate consequence of identity-­based polarization is that we ofen cannot resolve partisan confict by reasoned debate or educating people about the issues. If Mason is right that Democrats and Republicans are afectively polarized despite their agreement on many issues, then attempting to resolve political disagreement by closing partisan gaps on policy issues is misguided. Tis is a problem for deliberative conceptions of democracy. If our disagreements are not based on genuine reasons or arguments, then we cannot engage with each other’s views. An even larger worry is that the facts do not seem to matter. Individuals do not generally choose to support a political party on the grounds that it best represents their interests, preferences, or values. Quite the opposite. Partisans will edit their list of reasons for holding particular attitudes in order to defend the position that is faithful to their party. Tus, the typical voter’s political “views” have little intellectual value. As Somin (2013), Achen and Bartels (2016), Mason (2018), Anderson (Chapter 1 in this volume), Brennan (Chapter 8 in this volume), and many others have argued, our political identities are only loosely based on our own interests and issue positions. We are happy to cheerlead and protect our group’s status as if we had the facts, but the facts do not play a substantive role in shaping our political attitudes or beliefs. It also seems that we are not really interested in genuinely engaging with the other side. When policy debates are just battles between “us” versus “them,” we no longer select policies based on their actual content and we lose all motivation to reach a compromise. Consider what happened with the Afordable Care Act. When 24  For surveys, see Delli Carpini and Keeter (1996) and Somin (2013). 25  Interestingly, there may still be a failure of self-­knowledge, since many partisans claim that they “almost never” agree with the other party’s position (Doherty et al., 2016). Tus, we may ofen think that we disagree even when we do not. 26  However, it is not clear what hope there is for deliberative democracy if “political realism” is true and people rarely have genuine political beliefs.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Disagreement or Badmouthing?  313 constructing the Act, Democrats incorporated Republican ideas in hopes of winning bipartisan support, but they received none. For Republicans, it was not about policy. It was about denying their opponents a win and giving themselves a campaign issue. A compromise would have been seen as an unnecessary concession to “the enemy.” Tis has implications for policy-­making. If the political divide cannot be bridged by creative new policies that incorporate ideas from both the lef and right, there is little possibility for bipartisan lawmaking. Policy becomes about one side getting its agenda through and scoring a victory. Instead of constructing bipartisan policies, then, it seems we must try to reduce partisan antipathy by reducing the strength or alignment of political identities. Partisan cheerleading also corrupts public discourse. When empirical language is appropriated to make expressive claims, it generates confusion about what people are actually saying. As Anderson (Chapter 1 in this volume) puts it, “populist political discourse hijacks empirical discourse—the grammar of assertion—for expressive purposes, overtaking spaces normally reserved for empirical policy discussion.” Tis harms public discourse by infecting the public domain with misleading information, thereby corrupting human knowledge. Tis, too, has policy implications. When we mistakenly interpret expressive discourse literally, we tend to reply with empirical arguments. For instance, if liberal democrats interpret vehement denials of anthropogenic climate change as out of touch with reality, they will criticize their opponents for not engaging with the evidence. Tis may further antagonize the other side because their opponents will interpret liberals as calling them stupid. I will end this section by mentioning a rather strange upshot of my argument. I have argued that politics is characterized by less disagreement than we thought. On this basis, one might conclude that politics has more agreement than we thought. But this does not follow. Indeed, my argument may explain away genuine agreement just as easily as it explains away genuine disagreement.27 For example, two people who claim to be pro-­regulation may not actually have any settled beliefs on this matter; they may simply make this claim to express partisan support. In general, whenever two people seem to agree on some political issue, they may be expressing similar pro-­attitudes without any corresponding beliefs on the matter. Tus, my central claim that there are fewer political disagreements than we thought does not imply that there is more agreement in politics.28 We now have to be unsure about whether people who seem to agree are actually agreeing or just cheering for the same side.

8.  Concluding Remarks I have no solutions to these problems.29 Te aim of this chapter is not to recommend solutions but rather to change how we look at the problem of political disagreement. 27  Ballantyne (2016, p. 759) makes a similar point in the context of verbal disputes in philosophy. 28  However, Mason (2018) says there is more issue-­based agreement than surveys typically suggest. 29  Several solutions have been proposed. For example, correcting misperceptions about party supporters tends to reduce animus toward the other side (Ahler and Sood,  2018). Mason (2018) suggests that partisan news media should be compelled to present opposing partisans in more sympathetic ways. We

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

314  Michael Hannon In our polarized political climate, it is easy (and common) to conclude that politics is rife with disagreement, including factual disagreement. Tis is allegedly supported by a wealth of survey data. I have argued, however, that many political disagreements are not really disagreements. What appears to be disagreement is ofen cheap talk and partisan cheerleading. I have also explored the implications of this idea. As we have seen, the very factors that explain why political disagreement is superfcial may also explain why debates ofen go so poorly, why it is difcult to correct false beliefs, and why people seem to hold blatantly contradictory views. Te expressive nature of political engagement also raises doubts about standard survey research, the theory of motivated reasoning, and the view that political disagreements are “deep disagreements.” Te news is not all bad, though. As I have argued, political surveys tend to overstate the level of political misinformation, motivated reasoning is not distorting our perceptions of reality as ofen as scholars have claimed, and there is less disagreement over the facts than we thought. Ultimately, I hope to have highlighted the importance of thinking carefully about the concept of “disagreement” for future research in politics.30

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

References Abramowitz, A. I. and K. Saunders. 2008. “Is Polarization a Myth?” Te Journal of Politics 70(2): 542–55. Achen, C. and L.  Bartels. 2016. Democracy for Realists. Princeton, NJ: Princeton University Press. Ahler, D. and G.  Sood. 2018. “Te Parties in Our Heads: Misperceptions about Party Composition and Teir Consequences.” Te Journal of Politics 80(3): 964–81. Aikin, S. 2019. “Deep Disagreement, the Dark Enlightenment, and the Rhetoric of the Red Pill.” Journal of Applied Philosophy 36(3): 420–35. Ancell, A. 2019a. “Political Irrationality, Utopianism, and Democratic Teory.” Politics, Philosophy and Economics. DOI: 10.1177/1470594x19889108 Ancell, A. 2019b. “Te Fact of Unreasonable Pluralism.” Journal of the American Philosophical Association 5(4): 410–28. Anderson, E. 2021. “Epistemic Bubbles and Authoritarian Politics.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Arceneaux, K. and M. Johnson. 2013. Changing Minds or Changing Channels? Chicago: University of Chicago Press. Ballantyne, N. 2016. “Verbal Disagreements and Philosophical Skepticism.” Australasian Journal of Philosophy 94(4): 752–65.

may also try to shif the salience of partisan identities. In general, it seems the only way to reduce partisan antipathy is by reducing the strength of political identities. Tus, another upshot is that we should aim to depoliticize issues as much as possible, since partisan cheerleading trivializes important matters. Talisse (2019) suggests that we should be less politically engaged, since talking to each other about politics tends to activate our political identities. Te better thing to do is engage in non-­political activities with members of the opposing side. 30  Tanks to Jeroen de Ridder, Elizabeth Edenberg, Adam Gibbons, Neil Levy, and Robert Ross for comments on an earlier draf.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Disagreement or Badmouthing?  315 Bartels, L. M. 2002. “Beyond the Running Tally: Partisan Bias in Political Perceptions.” Political Behavior 24: 117–50. Bartels, L.  M. 2003. “Democracy with Attitudes.” In M.  MacKuen and G.  Rabinowitz (eds.), Electoral Democracy. Ann Arbor: University of Michigan Press. Bloom, P. 2016. Against Empathy: Te Case for Rational Compassion. New York: Ecco. Boxell L., M. Gentzkow, and J. M. Shapiro. 2017. “Greater internet use is not associated with faster growth in political polarization among US demographic groups.” PNAS 114(40): 10612–17. Brennan, J. 2017. Against Democracy. Princeton, NJ: Princeton University Press. Brennan, J. 2021. “Does Public Reason Liberalism Rest on a Mistake? Democracy’s Doxastic and Epistemic Problems.” In E.  Edenberg and M.  Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Bullock, J. and G.  Lenz. 2019. “Partisan Bias in Surveys.” Annual Review of Political Science 22: 325–42. Bullock, J., A. Gerber, S. Hill, and G. Huber. 2015. “Partisan Bias in Factual Beliefs about Politics.” Quarterly Journal of Political Science 10(4): 519–78. Campbell, A., P. E. Converse, W. E. Miller, and D. E. Stokes. 1960. Te American Voter. Oxford: Wiley. Cohen, G. 2003. “Party Over Policy: Te Dominating Impact of Group Infuence on Political Beliefs.” Journal of Personality and Social Psychology 85: 808–22. Converse, P. E. 1964. “Te Nature of Belief Systems in Mass Publics.” In D. E. Apter (ed.), Ideology and Discontent. New York: Te Free Press. Dahl, R. 1971. Polyarchy: Participation and Opposition. New Haven, CT: Yale University Press. Delli Carpini, X. M. and S. Keeter. 1996. What Americans Know About Politics and Why It Matters. New Haven, CT: Yale University Press. de Ridder, J. 2021. “Deep Disagreement and Political Polarization.” In E. Edenberg and M. Hannon (eds.), Political Epistemology. Oxford: Oxford University Press. Doherty, C. J. Kiley, and B. Jameson. 2016. “Partisanship and Political Animosity in 2016.” Pew Research Center report. Accessed at: https://www.people-press.org/2016/06/22/ partisanship-and-political-animosity-in-2016/ Dunning, D. 2014. “We Are All Confdent Idiots.” Pacifc Standard. Accessed at: https:// psmag.com/social-justice/confdent-idiots-92793. Fernbach, P.  M., T.  Rogers, C.  R.  Fox, and S.  A.  Sloman. 2013. “Political Extremism Is Supported by an Illusion of Understanding.” Psychological Science 24(6): 939–46. Fiorina, M. P., S. J. Abrams, and J. C. Pope. 2008. “Polarization in the American Public: Misconceptions and Misreadings.” Te Journal of Politics 70: 556–60. Greenstein, F. 1965. Children and Politics. New Haven, CT: Yale University Press. Haidt, J. 2012. Te Righteous Mind. London: Random House. Hannon, M. and J. de Ridder. 2021. “Te Point of Political Belief.” In M. Hannon and J. de Ridder (eds.), Te Routledge Handbook of Political Epistemology. Routledge. Huber G. A. and O. Yair. 2018. “How Robust Is Evidence of Perceptual Partisan Bias in Survey Responses? A New Approach for Studying Expressive Responding.” Presented at the Midwest Political Science Association Annual Meeting, Apr. 5–8, Chicago, IL. Huddy, L., L.  Mason, and L.  Aaroe. 2015. “Expressive Partisanship: Campaign Involvement, Political Emotion, and Partisan Identity.” American Political Science Review 109(1): 1–17.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

316  Michael Hannon Huemer, M. 2016. “Why People Are Irrational about Politics.” In J. Anomaly et al. (eds.), Philosophy, Politics, and Economics: An Anthology. Oxford: Oxford University Press. Iyengar, S., Y.  Lelkes, M.  Levendusky, N.  Malhotra, and S.  J.  Westwood. 2019. “Te Origins and Consequences of Afective Polarization in the United States.” Annual Review of Political Science 22: 129–46. Iyengar, S., G.  Sood, and Y.  Lelkes. 2012. “Afect, Not Ideology: A Social Identity Perspective on Polarization.” Public Opinion Quarterly 76: 405–31. Jacobson G. C. 2010. “Perception, Memory, and Partisan Polarization on the Iraq War.” Political Science Quarterly 125: 31–56. Jennings, M. and R. Niemi. 1981. Generations and Politics: A Panel Study of Young Adults and Teir Parents. Princeton, NJ: Princeton University Press. Jerit, J. and J. Barabas. 2012. “Partisan Perceptual Bias and the Information Environment.” Te Journal of Politics 74: 672–84. Kappel, K. 2018. “Tere Is No Middle Ground for Deep Disagreements about Facts.” Aeon Magazine. Accessed at: https://aeon.co/ideas/there-is-no-middle-ground-fordeep-disagreements-about-facts. Kelly, J.  T. 2012. Framing Democracy: A Behavioral Approach to Democratic Teory. Princeton: Princeton University Press. Khanna, K. and G. Sood. 2018. “Motivated Responding in Studies of Factual Learning.” Political Behavior 40(1): 79–101. Kinder, D. and N. Kalmoe. 2017. Neither Liberal nor Conservative: Ideological Innocence in the American Public. Chicago: University of Chicago. Kull, S., C.  Ramsay, S.  Subias, and E.  Lewis. 2004. “Te Separate Realities of Bush and Kerry Supporters.” Program on International Policy Attitudes. Lelkes, Y., Sood, G., and Iyengar, S. 2017. “Te Hostile Audience: Selective Exposure to Partisan Sources and Afective Polarization.” American Journal of Political Science 61(1): 5–20. Lenz, G. 2012. Follow the Leader? Chicago: University of Chicago. Levendusky, M. 2013. How Partisan Media Polarize America. Chicago: University of Chicago Press. Luskin, R.  C., G.  Sood, and J.  Blank. 2013. “Te Waters of Casablanca: Political Misinformation (and Knowledge and Ignorance).” Unpublished manuscript. Lynch, M. 2010. “Epistemic Circularity and Epistemic Disagreement.” In A.  Haddock, A. Millar, and D. Pritchard (eds.), Social Epistemology. Oxford: Oxford University Press. Lynch, M. 2019. Know-It-All Society. New York: Liveright. Mason, L. 2013. “Te Rise of Uncivil Agreement: Issue Versus Behavioral Polarization in the American Electorate.” American Behavioral Scientist 57(1): 140–59. Mason, L. 2015. “I Disrespectfully Agree: Te Diferential Efects of Partisan Sorting on Social and Issue Polarization.” American Journal of Political Science 59: 128–45. Mason, L. 2018. Uncivil Agreement: How Politics Became Our Identity. Chicago: University of Chicago. Nyhan, B. and Reifer, J. 2010. “When Corrections Fail: Te Persistence of Political Misperceptions” Political Behavior 32: 303–30. Peterson, E. and Iyengar, S. 2020. “Partisan Gaps in Political Information and Information-Seeking Behavior: Motivated Reasoning or Cheerleading?” American Journal of Political Science. https://doi.org/10.1111/ajps.12535.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Disagreement or Badmouthing?  317

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Pew Research Center. 2014. “Political Polarization in the American Public.” Accessed at: https://www.people-press.org/2014/06/12/political-polarization-in-the-americanpublic/. Pew Research Center. 2017. “Continued Partisan Divides in Views of the Impact of Free Trade Agreements.” Accessed at: https://www.pewresearch.org/fact-tank/2017/04/25/ support-for-free-trade-agreements-rebounds-modestly-but-wide-partisan-diferencesremain/ Prior, M. 2013. “Media and Political Polarization.” Annual Review of Political Science 16: 101–27. Prior, M., G.  Sood, and K.  Khanna. 2015. “You Cannot be Serious: Te Impact of Accuracy Incentives on Partisan Bias in Reports of Economic Perceptions.” Quarterly Journal of Political Science 10(4): 489–518. Rawls, J. 1993. Political Liberalism. New York: Columbia University Press. Schafner, B. and S.  Luks. 2018. “Misinformation or Expressive Responding? What an Inauguration Crowd Can Tell Us about the Source of Political Misinformation in Surveys.” Public Opinion Quarterly 82(1): 135–47. Sears, D. 1983. “Te Persistence of Early Political Dispositions: Te Roles of Attitude Object and Life Stage.” In L.  Weaver and P.  Shaver (eds.), Review of Personality and Social Psychology. London: Sage Publications. Shapiro, R.  Y. and Y.  Bloch-Elkon. 2008. “Do the Facts Speak for Temselves? Partisan Disagreement as a Challenge to Democratic Competence.” Critical Review 20(1): 115–39. Sinnott-Armstrong, W. 2018. Tink Again. Harmondsworth: Penguin. Somin, I. 2013. Democracy and Political Ignorance. Stanford, CA: Stanford Law Books. Taber, C.  S. and M.  Lodge. 2006. “Motivated Skepticism in the Evaluation of Political Beliefs.” American Journal of Political Science 50: 755–69. Talisse, R. 2019. Overdoing Democracy. Oxford: Oxford University Press.

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved. Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Index For the beneft of digital users, indexed terms that span two pages (e.g., 52–53) may, on occasion, appear on only one of those pages.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

2016 Presidential election  16–17, 33–5, 38, 41, 57, 262–3 Aberdein, Andrew  253–4 accountability  161–3, 262, 264 electoral  161, 163–4 individual  263–4, 268–9 meaningful 161–3 social 264 Achen, Christopher  136–7, 166, 176–7, 238–40, 304 Afordable Care Act  300, 312–13 agency epistemic  91, 156–9, 163, 172, 177, 259–60, 262–3, 265–8 Ahlstrom-Vij, Kristofer  4, 6 Alexander, Michelle  82–8, 92 algorithms  267–9, 275 “alphabet people”  141–5, 147–51 Althaus, Scott  180–1, 193–4 American National Election Survey (ANES)  187–8 Anderson, Derek  80 Anderson, Elizabeth  2, 107n.8, 252, 260n.2, 301 anti-democratic  2, 31–2, 43 anti-paternalism 99 strong 100–1 weak 101 Appiah, Kwame Anthony  137 Arendt, Hannah  2, 244, 308 Aristotle  167, 308 arrogance  251–5, 281 factional 255–6 intellectual  5, 167, 245–6, 252, 254–6, 268, 273–4 tribal  246, 252–3, 256, 262, 264, 268–9 white 252–5 assertion  20–1, 23, 25, 51–6, 59, 61, 89, 115–16, 292–3, 313 authoritarian  2, 18, 26, 35, 37, 42–3, 117–18, 141, 253, 255–6 authoritarianism  35–6, 117–18 aversive racism  183, 191–2 backfre efect  220 badmouthing 301

Ball, James  49, 56–7 Bartels, Larry  130, 136–7, 166, 176–7, 237–40, 304, 306–7, 312 Bayesian 80 belief  12–13, 40, 52, 79, 137, 139, 144, 182–3, 185–6, 209 collective 20–1 content  214–15, 219–21 degree 214–15 extreme  214–15, 218–21 fragile 306–7 group 21–2 polarization 209–24 responsible 262 bell hooks  92 benefcence  286–7, 290 Black Lives Matter  76, 89–90, 252–3 Bloom, Paul  309 Brady, Tom  142 Brazil  26, 290–1 Brennan, Jason  4, 6, 60, 120, 237, 305–6 Brexit  31, 37–9, 44, 49, 57–9, 61, 150, 181–2, 303 British Election Study (BES)  187–8, 190 Brown v Board of Education  87 Bullock, John  24–5, 301, 311–12 bullshit  2–3, 49–52, 54–9, 61–2, 282–3 strategic  50, 58 campaigns, political  49, 130, 259–62, 265 capitalism  137, 148 Carpini, Michael C. Delli  187–8 Cassam, Quassim  2–3, 261–3 challenger explosion (NASA)  212 chauvinist  211, 214–15 cheerleading  6, 24–5, 298, 301, 303–5, 308–14 Christiano, Tomas  6–7, 108 citizens’ assemblies  169 civic peer  270–1 civil rights movement  84 climate change  11–12, 14–15, 22, 25, 27, 164, 232–3, 250–1, 253, 280, 293–4, 297, 311, 313 global warming  11–12, 211–12, 214, 297, 308–9 Clinton, Hillary  13, 16–17, 32–3, 301–3

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

320 Index

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

closed-mindedness 2 coercion  141, 146–7, 149, 157 cognitive bias  2, 13–16, 18, 24, 27 asymmetries  12, 15–17, 25 framing efects  306–7 myside bias  234–5 Cohen, G.A.  52 Cohen, Geofrey  306 cold war  31 common good  119–20, 122–7 communism  12, 23, 117–18 community  17, 32–3, 86, 102, 115, 158–9, 163, 168, 170–2, 175–8, 261, 270–2, 285–6, 290–4, 305, 309 political  5, 162–4, 167, 177–8, 259–60, 282–3 competence  3–4, 31, 40–1, 97–8, 101, 103–6, 111, 119–20, 173–4 voter 105–6 competence gap  3–4, 99, 101, 104–5, 111 compliance condition  280, 287–90 Condorcet Jury Teorem  226 conservatism  15–16, 196, 200, 211–12, 303 conspiracy theory  1–2, 20, 23, 26, 218, 244, 253 Converse, Philip  307 cooperation  5, 19, 44, 209–10, 259–60, 272–3, 276 political  5, 259–60, 273–4, 276 corroboration view  219–21 counterfactual test  185 COVID-19  12, 26, 244–5, 253–5, 268n.19 cultural cognition theory  12–16, 26 culturally antagonistic memes  15–16, 24, 26–7 Dahl, Robert  306–7 Davis, Evan  49, 56–7 deception  22, 42, 44, 56–7, 270 deliberation  5–7, 21, 171–2, 212, 226–7, 235–7, 240–1, 246, 265–6, 308–9, 312 political  3, 64–5, 67, 70, 73–5 deliberative polling  171 democracy  3–5, 12, 19–20, 23–4, 31, 44–5, 97–9, 102–4, 106, 109, 111, 114, 116–18, 129, 131, 136, 141–3, 148, 157, 159–61, 194, 222–3, 246, 250–1, 256, 264–5, 270, 297, 306–7, 312 anti-democratic  35–6, 101, 117 challenges to  2–6, 12, 31, 34, 97, 104–5, 108, 112, 114, 117, 132–3, 159–61, 169, 222–4, 226, 246–7, 255, 312 defenses of  4, 98, 104, 108, 111–12, 115, 118–19, 121, 123, 160, 227, 244 democratic discourse  2, 23–4, 31, 42–3, 45 democratic culture  31–2, 42, 44 divided  261, 263–4, 269 epistemic  1, 6, 12, 98, 112, 115–16, 226–7 folk theory of  137, 298–9, 307

instrumental justifcation of  3, 97–8, 102, 106, 245–6 intrinsic justifcation of  115–16, 119–21 see also, elections right to  98–9, 109 Democratic Party  17–19, 142–3 de Ridder, Jeroen  5–7, 245, 247 Dewey, John  308 dignity  223, 271 disagreement  2, 4–5, 12, 15, 68, 112, 116, 132, 135–6, 140, 226, 246, 248, 251, 253, 255, 259, 269, 272, 300, 307 about facts  11, 132, 246, 259, 271–2, 298, 314 about values  11–12, 116, 228, 232, 270–1, 275 deep  5, 68, 115–17, 119–20, 167, 226–41, 247–9, 254, 298, 308, 311, 314 epistemic  5, 226–7, 234–5, 245–52, 255, 260 epistemology of  1, 6–7, 269 insincere  6, 298, 300–1, 308 moral  135, 250 ordinary 227 peer  229–30, 263 persistent  259, 273–4 political  5–6, 68, 140, 246, 259–60, 271–2, 274, 297, 302, 308–11, 313–14 reasonable  4, 135, 150–1, 231, 273 religious 140 disenfranchisement  105, 111, 123 by property  117–18 by race  118–19 of women  117–18 disinformation  17, 33–4, 37, 42–3, 164–5, 260 dissent  13, 19–20, 85, 89–92, 141, 260–1, 264–5, 273–4 diversity  7, 17–19, 21–2, 120, 177 of group  18, 174, 274 Diversity Trumps Ability  226 dogmatism, see Arrogance, Intellectual Dotson, Kristie  76–9, 82, 89–91 Downs, Anthony  121–3, 127–31 Downsian model  114, 123–7, 130–1 Dufy, Bobby  303 duty  115, 119, 281–7, 289, 291–2 epistemic  6, 280–1, 286, 290 perfect  5–6, 280, 285–6 imperfect  5–6, 280, 285, 290, 292 to object  5–6, 281–6, 288–95 D’Ancona, Matthew  49 echo chambers  1, 5, 89, 259, 259n.1, 300 Edenberg, Elizabeth  5–7 egalitarian  3–4, 6, 14, 114–17, 121–2, 129, 148, 156–7, 173, 178 see also equality

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Index  321 elections  4, 19, 23–4, 31, 36–7, 39, 103–4, 125–6, 136, 157, 159, 161, 163–8, 176–8, 261, 280 electoral representative systems  4, 156–7, 161, 163, 168–70, 175–8 epistemic pathologies of  4, 157–9, 178 elites  18–19, 21–2, 24–5, 173, 222, 299, 305 empathy 274 entertainment  44–5, 166 epistemic  1–3, 12, 18, 21–3, 26, 31–2, 37, 43–5, 55–6, 76–7, 86, 90, 104, 112, 137, 157, 162–6, 169, 176, 209, 226, 230, 244–5, 255, 260–2, 265, 268, 272, 282–3, 291 circularity 248 crisis  1–2, 26, 262 frames  79–80, 92 norms  20, 26–7, 79, 228 principles  5, 228–33, 238–40, 247–8, 250, 254 rationality  140, 231–2 responsibility  259–60, 264–70, 274–5 spillovers 250 value  78, 98, 112, 137, 281, 286, 291–2 epistemic bubbles  2, 6, 11–18, 20–3, 25–7, 260–1 epistemic exclusion  78–9, 90 First-order  78, 90 Second-order  78–9, 90 Tird-order  76–9, 90–1 epistemic peer  229–30, 263, 271n.23 see also disagreement, epistemology of epistemological system  3, 6, 76–92 resilient  76, 81–2, 84, 91 epistemology  1–2, 6–7, 22, 27, 40–1, 239–40, 259–60, 269, 272 political  7, 26, 49, 57–8, 62 social  12, 18, 20–2, 27, 81 epistocracy  3–4, 97–9, 101–6, 108–12, 117–18, 121, 123, 173 equality distributive 106 political  4, 99, 107–8, 110–11, 115–16, 121–2, 132, 156, 223 moral 108 relational  106, 109 decisional  106–8, 110 political  4, 99, 107–8, 110–11, 115–16, 121, 125, 132, 156, 223 right to  99 status 115–17 see also inequality see also egalitarian Estlund, David  3–4, 6–7 EU Referendum Identity Model  190–1, 198 evidence  12–13, 16, 22, 25, 43, 80, 121–2, 128–9, 143, 150, 159–60, 164, 172, 182–3, 185, 213, 240–1, 297, 308

anecdotal  180, 182–3, 192–3, 232, 236–7 statistical  180, 182–3, 192–3 experts  1–2, 4, 14, 19–20, 22–6, 53, 103–4, 114, 116, 118, 157, 162–7, 169–73, 175–6, 228–30, 232–3, 240, 244–5, 248, 250–1, 253, 266, 270–2, 283, 291, 311 distrust of  14, 244–5, 253 expressive responding  298, 301, 307, 310–12 see also cheerleading extremism  5, 16–17, 212–22, 268, 299 Facebook  17, 32–5, 37, 39–42, 142, 262, 268, 286 facts  12, 120, 246, 259, 297, 307–8 alternative  1, 61–2 empirical  3, 64–7, 71, 73, 75 fact-checking 259 moral 65 normative  64–8, 70–5, 230 fairness  111–12, 124 procedural 110 fake news  1–2, 17, 19–20, 22–3, 26–7, 31, 33, 36, 39–44, 261–3, 268 Fauci, Anthony  244–5, 254–5 feminism  81, 215 Floyd, George  76 Foucault, Michel  80 Frankfurt, Harry  50–1, 54–9, 62 free speech  19–20, 35–6, 110–11, 145, 226, 268 freedom  35–6, 42–3, 98–100, 102, 110–11, 116, 145, 156, 228–30, 246, 271, 273–5, 281 Freeman, Samuel  144–5 Fricker, Miranda  81–2, 86 Friedman, Jefrey  180 Gates, Bill  253, 285 Gaus, Gerald  135, 143–7 general will  98–9, 102 Gilbert, Margaret  20 Gilen, Martins  130 Goldman, Alvin  158, 231–2 Guerrero, Alexander  4, 6–7, 104 Habermas, Jürgen  151 Hannon, Michael  6–7 Hanson, Robin  142 hate speech  2–3, 61–2 hierarchy  106, 117, 180, 182–3 social  98, 108–9 Hobbes, Tomas  117 humility 174 intellectual  262, 264–6, 268–70 identity politics  6–7, 18, 137, 237, 299–300, 304 identity-expressive discourse  2, 6, 23–5, 307, 310 see also expressive responding

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

322 Index identity-protective cognition  14–15, 264 ideology  17–18, 82, 84, 137, 139, 142, 145, 147, 149–50, 253–5, 282–3, 306 purity 25 ignorance  4, 6–7, 21–2, 51, 53–5, 72, 122, 129, 140, 151, 158–63, 168, 173, 177, 180, 193, 252, 282, 311–12 rational  123–4, 127 voter  1, 4, 50, 160, 163, 165, 176, 181, 186, 193–4, 237 Ilyin, Ivan  35–6, 42–5 immigration  4, 19, 22, 24, 59–60, 150, 169–70, 180–2, 184, 187–94, 227–8, 230, 246, 297–8, 300, 308 attitudes  183, 191–2 “Good Immigrants”  180–1, 184 U.K.  4, 146, 181, 188 in-group/out-group thinking  166–7, 176–7, 270–1 incompetence  3–4, 99 voter  157, 311 inequality  107–8, 110–12, 117, 120–1, 125, 128–30, 137, 168, 227–8 information  2, 11, 13, 15, 40, 44, 114, 150, 163–4, 172, 216–17, 260, 267–8, 303 adequate 114–15 economics of  123, 125, 127, 132 “free information”  128–32 institutional sources of  264–5 low-information decision-making  4, 114, 132–3 political  140, 166, 263, 280 see also misinformation and disinformation well-informed  122, 126–7, 145, 150, 159, 186 information cascades  13 injustice  21–2, 72, 85, 168, 281 epistemic hermeneutical  78n.4, 81 testimonial 78n.2 Internet Research Agency (IRA)  32–5, 37–41, 44 irrationality  1, 309, 312 Iyengar, Shanto  237, 299 Jamieson, Kathleen Hall  33–4 Jim Crow Laws  77, 82–4, 86–7, 92 justice  86, 109, 119–20, 135, 156, 212 procedural 109 Rawls’s principles of  110–11 see also, injustice justifcation  83, 85, 87–8, 112, 124–5, 127, 132, 140, 144, 262–3 political  3, 6, 64–6, 68, 70–4, 112, 135, 145–7 pro tanto  64–5 public  137, 139, 141, 143–4, 146–7, 149–51

symmetry view  65, 69 asymmetry view  65, 69 of democracy  97–8, 102–4, 106, 114, 117, 157, 161 Kaepernick, Colin  252–3 Kahan, Dan  12–14, 16, 22, 24–6, 235 Kantianism 147–8 Kappel, Klemens  230–1, 311 Keeter, Scott  187–8 knowledge  1, 22, 34, 40, 79–80, 123, 156, 168, 192, 195, 230–1, 261–2 political  49, 125, 128, 147, 187–90, 192–4, 305n.17 Lackey, Jennifer  5–7, 52–3, 285 Langer, Gary  301–2 legitimacy  16, 18–20, 62, 140, 272, 294–5 political  6–7, 156, 274 liberal principle of  135–6, 141, 144–6 liberalism  5, 116–17, 135–6, 140–1, 143–51, 259–60, 273 lies  2, 37–8, 42, 50, 56–7, 62, 282–3 liars  35–6, 40, 42–4, 57 lying  50, 56–7, 289 lottocracy  4, 104, 169 lottocratic systems  4, 156–7, 169–70, 173, 175–6, 178 agenda setting  171 epistemic virtues of  104, 169, 172–4, 176 learning phase  170–1, 173, 175–6 Lynch, Michael P.  5–7, 37, 167, 230–1, 261–2, 264–70 MacCan, Donnarae  88 manipulation  2–3, 12, 165–6, 176, 270 political 33 Mason, Lilliana  142, 297–300, 312 mass incarceration  19, 83–4, 87, 92 Medina, José  87–8 Migration Advisory Committee  181, 184 Mill, John Stuart  110–11, 308 Mills, Charles  83, 87–8, 168 misinformation  1, 42, 140, 151, 159–60, 162–5, 176–7, 181, 271–4, 298, 303, 308, 310–12, 314 see also disinformation motivated reasoning  22, 24–6, 140, 162, 164, 298, 310–12, 314 Mueller, Robert  32–5 Munger, Kevin  283–4 Murphy, Liam  280, 286–8, 290 National Health Service (NHS)  57–8, 182 news  15–17, 32–3, 38–41, 137–8, 166, 176, 249, 259–61, 267, 300–2, 310

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Index  323 Nguyen, C. Ti  89, 260n.2 Nixon, Richard  55 non-interference principle  100–1, 104, 106 normative defeater  293–4 normative framework  227–30 epistemic (e-framework)  231–4, 236, 239–41 moral (m-framework)  232–4, 236, 239–41 shared  227, 229–30 norms  2, 6, 12, 18, 20–4, 26–7, 40–1, 45, 77, 79, 85–6, 126, 135–6, 143–6, 148–9, 156, 173, 228–9, 231–2, 246, 263, 284 of assertion  23, 52 of belief  21 of testimony  40, 261–5

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Obama, Barack  16, 23–5, 61–2, 252–3, 261, 297, 301, 309–10 oppression  3, 76–8, 81–2, 86, 89–90, 168 epistemic  76–9, 82 Pariser, Eli  261 partisanship  38–9, 41, 43–4, 189–90, 192, 198–9, 237–8, 263, 297–304, 310–12 partisan identity  11, 26, 300, 307 party identity model  189–91 paternalism  3–4, 97–106, 108 epistrocratic  98, 100 legal  100, 104–6 Peter, Fabienne  3, 6–7 philosophy of race  81 Plato  1, 117, 147 pluralism  19, 143, 293 reasonable 227–8 Pohlhaus, Gaile  82, 86 polarization  1, 5, 11–12, 209–11, 234, 238–41, 245, 264–6, 268, 274, 298, 303 afective  210, 221–2, 245, 249, 299–300 belief 209–24 cognitive  5, 234, 236–8, 245–7, 249–51, 255–6 group  12–13, 15–16, 18, 26, 172, 210n.2, 211, 236 informational account  216, 218 mechanisms 213 political  1–2, 4–5, 209–12, 222, 227, 234, 238, 294 practical  5, 234, 236, 238–41 social comparison account  216–18 platform  209–10, 245 police brutality  19, 76, 86, 90–2, 252–3 political capture  161–3 political ignorance  1, 4, 6–7, 114, 121–3, 129, 140, 150–1, 157–63, 165, 173, 176–7, 181, 186, 193, 237, 311–12 political irrationality  14, 23, 78, 140, 163, 184, 221–4, 230–1, 233–6, 238, 240, 252, 293–4, 309, 312

political knowledge tests  187–8 political liberalism  5, 141, 259–60, 273 see also public reason political liberties  110–11 political participation  125, 176–7, 189, 237 populism  2, 18–22, 24–7, 136–7 lef-wing 18–19 right-wing 18–19 populist model of democracy  136 post-truth  1–3, 49–50, 55–6, 61–2 Postman, Neil  166 preferences  131, 137–9, 160–1, 163–4, 180–1, 183–9, 191–4, 239, 297–8, 300, 303–4, 306–7, 312 real/genuine 186 unreasonable 184 Prigozhin, Yegveniy Viktorovich  32–3 Prior, Markus  302 Project Lakhta  32, 34–5, 39 propaganda  1–3, 31, 42–3, 59–61, 260 afective 60 demagogic  50, 59, 62 identity 60–1 political 60 racist 62 Russian propaganda  37, 42 public discourse  20, 32, 256, 313 public reason  4, 71, 135–7, 139–47, 149–52 Putin, Vladimir  35–7, 42–3 qualifed acceptability requirement  112 Quong, Jonathan  135, 145, 149 racism  19, 21, 76–7, 86–91, 183, 187, 191–2, 211, 227–8, 245, 282–3 Railton, Peter  79–80, 194 rationality  22, 27, 140, 156, 177, 216, 227, 231–2, 234, 256 Rawls, John  5, 65, 110–12, 135, 137, 141, 144–7, 151, 227–8, 259–60, 271–4 realism  136–41, 144, 146–7, 149–50, 304–5 reasonable  4, 32, 41, 43, 52–3, 119–20, 126, 135, 144–7, 151, 226, 233–4, 263, 273–4 people  118–19, 146, 190 doctrines 141 pluralism  143, 227–8 Reconstruction 118–19 Republican Party  12, 15–19, 22, 32–3, 139, 142–3, 236–7, 297–302, 305–6, 309–10, 312–13 respect  3, 27, 145, 256, 264–5, 269–74 appraisal  270n.20, 273n.28 epistemic  269–70, 272–3 moral  259–60, 268–9, 271–5 mutual  5, 256, 259–60, 272–6 recognition  270n.20, 273–4

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

324 Index

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

responsibility  37, 174, 268–9, 282, 284, 290–3 epistemic  259–60, 264–9, 275 retrospection test  185–7 rights  18–20, 26, 115–16, 120, 148, 190, 223 collective 102 democratic  117, 223 economic 116–17 human  19, 118–19, 274–5 individual  19–20, 35 liberal  116–17, 145 Rini, Regina  2–3, 6–7, 261–4, 266–9 Rousseau, Jean-Jacques  19–20, 98–9, 102 Russia  2, 32–9, 42, 44 KGB 34 interference in U.S. election  34, 39–40, 280 Internet Research Agency (IRA)  32–5, 37–41, 44 Project Lakhta  32, 34–5, 39 Russian intelligence  33–4 Russian social media operations, see also social media  2, 32, 37–9 Schumpeter, Joseph  121–2 self-governance  3–4, 102 collective 98–9 individual 102 sensibility  4, 156–8, 163–6, 168–9 single-issue lottery-selected legislature (SILL)  170–6, 178 skepticism  2, 6, 32, 37, 42, 232–3, 270 climate change  26, 232–3, 246, 250, 313 Pyrrhonian  248, 251 testimonial 41–4 weaponized  32, 42, 44–5 social identity  17–18, 21, 82, 217, 219–20, 239, 299, 304, 307 social media  2, 6, 15, 17, 31–2, 39–40, 42, 44, 220, 255–6, 259–65, 267–9, 275 Russian social media operations  2, 32, 37–9 bots  2, 33, 37–9, 44 trolls  2, 27, 37–8, 44 Twitter  32–5, 37–40, 42, 244, 262, 283–4 Facebook  17, 32–5, 37, 39–42, 142, 262, 268, 286 media literacy campaigns  259, 262–3, 265 fact checking  259 testimony  40–2, 45, 264 epistemic challenges  260 epistemic solutions  260, 265–6 moral challenges  260 moral solutions  268–9 social status  13–14, 167, 283–5, 288, 291–4 Somin, Ilya  238, 312

Sommers, Samuel  21 sortition 169 sovereignty  19–20, 97–8, 136 individual 100 sports fans  142, 147, 238, 304–5, 308–9 sports team  138–9, 142, 147, 166, 239, 304–5, 308–9 standpoint theory  168 Stanley, Jason  59–60 Steele, Jennifer  4, 6–7 subjection, legal  98–9, 103–5, 107–8 asymmetrical  99, 104–8 right against  3–4, 98–9, 103–6 symmetrical  98–100, 103–4, 106, 109 unequal 107–8 sufrage  103–5, 111–12 universal  103–4, 107, 117–19 for women  118–19 see also disenfranchisement Sunstein, Cass  12–13, 15, 26, 260–1 Talisse, Robert  5–7, 11, 234–6, 245 Taylor, Breonna  76 Tesich, Steve  50, 55–6 testimony Doxastic Challenge  139, 146 social media  262–3, 265 testimonial skepticism  41–4 testimonial reputation  264, 268–9 testimonial norms  261–5 Tomasi, John  141, 144–5 Toole, Briana  3, 6–7 Trump, Donald  11–12, 16–20, 22, 24–5, 32–4, 39, 41, 43–4, 49, 51–8, 60–2, 143, 244–5, 252–3, 261, 282–3, 290–1, 301, 305–6, 308, 310–11 trust  1–2, 35–6, 40, 116, 177, 232–3, 249–50, 264, 270 truth  1–2, 5, 24–7, 37, 50, 53, 55–6, 58, 61–2, 74, 137, 194, 240–1, 244, 246, 249, 262, 270–1, 298, 309, 311 post-truth  1–3, 49–50, 55–6, 61–2 moral  65–6, 74 Tufekci, Zeynep  37 Twitter  32–5, 37–40, 42, 244, 262, 283–4 U.S. Congress  16, 39, 140, 167, 170, 174, 290, 299 U.S. House of Representatives  159 U.S. Senate  38, 159, 244 U.S. Supreme Court  290 uncertainty  3, 6, 40, 64–71, 75 belief 66–8 empirical  6, 64–5, 67, 69, 74

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.

Index  325 normative  3, 65–6, 68–71, 74–5 political 66–9 practical 67–8 substantial  3, 64 understanding  1–2, 4, 6, 15–18, 22, 49–50, 53–4, 77, 79–80, 83, 85–6, 88–9, 91–2, 97, 115, 120–1, 123, 130, 132, 156, 168–9, 176, 180, 192–3, 223, 230, 245–6, 268, 273–4, 281, 286–7, 290–1, 293 unions  118–19, 130–1 United Nations Declaration of Human Rights 274–5 US v Internet Research Agency  31n.1, 32n.2, 33nn.4, 7, 34n.13 utilitarianism  148, 230

virtue  14, 70, 107–8, 145, 213, 223, 229–30, 273, 286, 293, 295 epistemic  1, 4, 176, 264–5, 276 voting  36–9, 97, 99, 102, 104–7, 110–11, 118–19, 122, 124, 130, 137–8, 161–2, 171, 181, 184–6, 237, 244, 299, 304 disenfranchisement  105, 118–19 plural  105–6, 110–11 right to vote  102–3, 105–6, 110–11, 115, 121 universal sufrage  103–4, 107, 117–19 voters  4, 6, 19, 24–7, 33–5, 58–9, 61, 97, 99, 105–7, 109, 122–4, 130, 136–40, 142–3, 150, 159, 163, 166, 173, 190, 238, 297–9, 301, 303–6, 309–12 Voting Rights Act of 1965  118–19

Vallier, Kevin  143, 146–7 values  1, 11–13, 18–19, 67–70, 85, 114, 116, 122, 131–2, 138, 142–4, 156–7, 173, 178, 187, 191, 194–5, 223, 227–8, 230, 232, 236, 240, 251, 259, 264, 270–1, 274–5, 297–8, 304–5, 312 vice 1 epistemic  261–4, 266–7 Viehof, Daniel  99 Vietnam War  212 violence  61, 165, 246, 252–5, 281 epistemic 89

Watergate 55 Weatherson, Brian  69–70 Westwood, Sean  237 white supremacy  3, 6, 76–7, 82–92 will of the people  4, 19–20, 23–4, 181, 193–4 wisdom of crowds  226 Wittgenstein, Ludwig  236, 248, 254 Wright, Gavin  118–19 xenophobia 61

Copyright © 2021. Oxford University Press USA - OSO. All rights reserved.

Zuckerberg, Mark  285

Political Epistemology, edited by Elizabeth Edenberg, and Michael Hannon, Oxford University Press USA - OSO, 2021.