Advances in Experimental Philosophy and Philosophical Methodology 9781474223218, 9781474223232, 9781474223225

Citation preview

Advances in Experimental Philosophy and Philosophical Methodology

Advances in Experimental Philosophy Series Editor: James R. Beebe, Associate Professor of Philosophy, University at Buffalo, USA Editorial Board: Joshua Knobe, Yale University, USA Edouard Machery, University of Pittsburgh, USA Thomas Nadelhoffer, College of Charleston, UK Eddy Nahmias, Neuroscience Institute at Georgia State University, USA Jennifer Cole Wright, College of Charleston, USA Joshua Alexander, Siena College, USA Empirical and experimental philosophy is generating tremendous excitement, producing unexpected results that are challenging traditional philosophical methods. Advances in Experimental Philosophy responds to this trend, bringing together some of the most exciting voices in the field to understand the approach and measure its impact in contemporary philosophy. The result is a series that captures past and present developments and anticipates future research directions. To provide in-depth examinations, each volume links experimental philosophy to a key philosophical area. They provide historical overviews alongside case studies, reviews of current problems and discussions of new directions. For upper-level undergraduates, postgraduates and professionals actively pursuing research in experimental philosophy these are essential resources. Titles in the series include: Advances in Experimental Epistemology, edited by James R. Beebe Advances in Experimental Moral Psychology, edited by Hagop Sarkissian and Jennifer Cole Wright Advances in Experimental Philosophy of Language, edited by Jussi Haukioja Advances in Experimental Philosophy of Mind, edited by Justin Sytsma Advances in Religion, Cognitive Science, and Experimental Philosophy, edited by Helen De Cruz and Ryan Nichols

Advances in Experimental Philosophy and Philosophical Methodology Edited by Jennifer Nado

Bloomsbury Academic An imprint of Bloomsbury Publishing Plc

Bloomsbury Academic An imprint of Bloomsbury Publishing Plc 50 Bedford Square London WC1B 3DP UK

1385 Broadway New York NY 10018 USA

www.bloomsbury.com BLOOMSBURY and the Diana logo are trademarks of Bloomsbury Publishing Plc First published 2016 Paperback edition first published 2017 © Jennifer Nado and Contributors, 2016 Jennifer Nado has asserted her right under the Copyright, Designs and Patents Act, 1988, to be identified as the Editor of this work. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage or retrieval system, without prior permission in writing from the publishers. No responsibility for loss caused to any individual or organization acting on or refraining from action as a result of the material in this publication can be accepted by Bloomsbury or the editor. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library. ISBN: HB: 978–1–4742–2321–8 PB: 978–1–3500–4857–7 ePDF: 978–1–4742–2322–5 ePub: 978–1–4742–2320–1 Library of Congress Cataloging-in-Publication Data Advances in experimental philosophy and philosophical methodology / edited by Jennifer Nado. pages cm. -- (Advances in experimental philosophy) Includes bibliographical references and index. ISBN 978-1-4742-2321-8 (hb) -- ISBN 978-1-4742-2322-5 (epdf) -- ISBN 9781474223201 (epub) 1. Philosophy--Research. 2. Methodology. I. Nado, Jennifer, editor. B52.A325 2016 107.2--dc23 2015034320 Series: Advances in Experimental Philosophy Typeset by Fakenham Prepress Solutions, Fakenham, Norfolk NR21 8NN

Contents Notes on Contributors 1 Introduction Jennifer Nado 2 Experimental Philosophy, Noisy Intuitions, and Messy Inferences Jonathan M. Weinberg

vii 1

11

3 How to Do Better: Toward Normalizing Experimentation in Epistemology John Turri

35

4 Thought Experiments, Mental Modeling, and Experimental Philosophy Joshua Alexander

53

5 Gettier’s Method Max Deutsch

69

6 Intuitive Diversity and Disagreement Ron Mallon

99

7 Intuitions and the Theory of Reference Jennifer Nado and Michael Johnson

125

8 Intuitive Evidence and Experimental Philosophy Jonathan Jenkins Ichikawa

155

Index

175

Notes on Contributors Joshua Alexander is Associate Professor of Philosophy at Siena College, where he is chair of the philosophy department and director of cognitive science. His work focuses primarily on the nature of philosophical cognition and intellectual disagreement. He is the author of Experimental Philosophy— An Introduction (Polity 2012), and is currently working on a book on the philosophy and cognitive science of disagreement. Max Deutsch is Associate Professor of Philosophy and Chair of the Department of Philosophy at the University of Hong Kong. He works in the philosophy of language, the philosophy of mind, and philosophical methodology. His book, The Myth of the Intuitive, was recently published by MIT Press. Jonathan Jenkins Ichikawa is an Associate Professor of Philosophy at the University of British Columbia. He has published in epistemology, philosophical methodology, and philosophy of mind. He is the co-author (with Benjamin Jarvis) of The Rules of Thought (Oxford University Press) and editor of the forthcoming Routledge Handbook of Epistemic Contextualism. Michael Johnson is honorary Assistant Professor of Philosophy at Hong Kong University. He received his PhD from Rutgers University in 2011. His research focuses primarily on the metaphysics of mental and linguistic representation. Ron Mallon is Associate Professor of Philosophy and Director of the Philosophy-Neuroscience-Psychology Program at Washington University in St. Louis. He is currently working on a book on the social construction of human kinds. Jennifer Nado is Assistant Professor of Philosophy at Lingnan University, Hong Kong. She received her PhD from Rutgers University in 2011. She works

viii

Notes on Contributors

primarily on metaphilosophy and epistemology. She has published papers in journals such as Philosophy and Phenomenological Research, The British Journal for the Philosophy of Science, Philosophical Psychology, and Philosophy Compass. John Turri is a philosopher and cognitive scientist at the University of Waterloo. He directs the Philosophical Science Lab, where researchers use scientific methods, concepts, and findings to make progress on philosophical questions, new and old. Jonathan M. Weinberg is Associate Professor of Philosophy at the University of Arizona, where he is also a member of the Interdisciplinary Program in Cognitive Science.

1

Introduction Jennifer Nado

Five or ten years ago, an introduction for a volume entitled “Advances in Experimental Philosophy and Philosophical Methodology” would probably have opened with something like the following: From its inception, the most salient and controversial aspect of experimental philosophy has been its implications for traditional philosophical methodology. Experimental philosophy challenges traditional methods both through its findings, and through its very existence. The findings generated by experimental philosophers, many of which appear to suggest that intuitions are sensitive to factors such as cultural background and emotional state, challenge the legitimacy of appeal to intuition as an evidential source in philosophical theorizing. Further, however, the radical notion of philosophers using empirical methods drawn from psychology challenges the idea that philosophy is an exclusively armchair-driven discipline.

Summaries like this can be found throughout the literature—I’m quite sure my 2011 dissertation contained a roughly similar paragraph. In fact, in all honesty, the above is essentially the opening paragraph I wrote (more or less on autopilot) as I began to draft this introduction. But experimental philosophy’s implications have really never been quite so straightforward as such a paragraph suggests. Early work in what has been called the “negative program” in experimental philosophy tended to invoke something like the following very simple argument, which we might call the “Argument from Variation”: 1 Experimental studies have shown that intuitions vary as a function of inappropriate factors like cultural background, emotional state, and so forth.

2

Advances in Experimental Philosophy and Philosophical Methodology

2 The above variation shows intuition to be an unreliable guide to philosophical truth. 3 Philosophers must abandon their reliance on intuition-based methods of theorizing. Variations on this basic line of argument for example can be found in Alexander and Weinberg (2007), Swain et al. (2008); Joachim Horvath (2010) describes a near-identical argument as being the “master argument” of experimental philosophy.1 As experimental philosophy has matured, however, its practitioners have been confronted with the fact that this simple argument is in fact less straightforward than it appears. In particular, critics of experimental philosophy have pointed out that it involves several unstated and potentially problematic assumptions. First, the argument’s premises are stated simply in terms of “intuition,” suggesting that our ordinary, non-technical understanding of intuition will suffice to ground the argument. Yet it’s quite plausible that a casual understanding does not so suffice—debates over just what an intuition is predate experimental philosophy (see for instance the articles in DePaul and Ramsey 1998), but the variation argument gives them new urgency. As several critics have pointed out, it’s quite difficult to give a definition of intuition that comports fairly well with philosophers’ general usage of the term without causing the variation argument to become either self-undermining, or overly skeptical. As an example of the former problem, both George Bealer (1992) and Joel Pust (2001) have pointed out that arguments against the use of intuition plausibly invoke epistemological standards, concepts, and so forth, that appear to be backed up only by intuition. If we accept the conclusion of an anti-intuition argument, we will be forced to reject those standards— thereby undermining the very argument that led us to reject intuition. As an example of the latter problem, Timothy Williamson (2007) has noted that the cognitive abilities that comprise “philosophical intuition” also plausibly underlie a large proportion of everyday cognition as well. Reject intuition, and one ends up becoming a “judgment skeptic”—that is, one’s arguments against philosophical uses of intuition will generalize to a great many ordinary judgments. Indeed, a flat-out rejection of any belief justified only by intuition would plausibly require rejection of elementary logic and mathematics—and

Introduction

3

presumably, most experimental philosophers would rather hold on to (say) modus ponens and addition. A second difficulty is that the variation argument’s premises simply assume that the experimental findings—which initially used undergraduate subjects almost exclusively—generalize to professional philosophers. In other words, the argument implies that, if undergraduates are shown to be sensitive to order effects, we can conclude that philosophers are highly likely to be so as well. This assumption has been the subject of what is commonly called the “expertise defense” (see for example Kauppinen 2007; Ludwig 2007; Devitt 2011; Williamson 2011). Proponents of this defense claim that, since professional philosophers plausibly have a great deal of expertise in thinking about matters philosophical, there is simply no reason to assume that they will be subject to the biases and errors that have been found in “folk” intuition. Recently, experimental philosophers have attempted to combat this objection by conducting studies using philosophers as subjects—and results so far suggest that philosophers’ intuitions may exhibit biases as well (see e.g. Schulz et al. 2011; Schwitzgebel and Cushman 2012; Tobia et al. 2013). A proponent of expertise, however, might well reply that survey contexts do not prompt philosophers to exercise the capacities which constitute their special expertise—philosophers’ expertise might manifest only through extended reflection, for instance. Finally, the conclusion of the variation argument follows only if we assume that intuition really does play a central, evidential role in philosophical method. This is the most recent assumption to come under question, in the form of what I’ll call (following Cappelen 2012) the “Centrality” debate. “Centrality” is, in Cappelen’s words, the thesis that “Contemporary analytic philosophers rely on intuitions as evidence (or as a source of evidence) for philosophical theories” (Cappelen 2012: 3). It’s a common assumption on both sides of the intuition literature, but Cappelen and several others (Deutsch 2009, 2010; Earlenberg and Molyneux 2013; Ichikawa and Jarvis 2013; Ichikawa 2014) have recently begun to argue that it is false. They argue, for instance, that evidence in philosophy does not consist of intuitions—it consists of facts about philosophical phenomena like knowledge, causation, and so forth. And although philosophers have often characterized thought experiment judgments as mere appeals to intuition, closer examination

4

Advances in Experimental Philosophy and Philosophical Methodology

reveals that the relevant claims are typically backed up by argument. If this is right, then empirical studies of spontaneous, unreflective “intuitions” aren’t obviously relevant to what philosophers actually do. The three assumptions just surveyed are not unrelated. Williamson’s response to the “judgment skeptic” involves rejecting what he calls the “psychologization” of our evidence: that is, the tendency to assume that evidence in philosophy consists of intuitions. Similarly, Cappelen’s arguments against Centrality, at points, turn on doubts about the very coherence of the concept of an intuition. Moreover, although Cappelen criticizes proponents of the expertise defense for their commitment to the Centrality assumption, a rejection of Centrality raises similar concerns about issues of philosophical expertise. As noted earlier, a common response to the expertise defense has been to run more of approximately the same sort of survey that experimental philosophers have always used—only now using professional philosophers as subjects. But, if the opponents of Centrality are right, then the real difference between naïve subjects and professionals may well be that the former, but not the latter, form their beliefs about philosophical issues primarily by consulting intuition. Perhaps philosophers are subject to just the same biases in intuition as the “folk”—that’s of little consequence if the actual practice of philosophy, outside the context of the laboratory, primarily involves argumentation rather than brute appeal to intuition. Philosophical expertise, in other words, might consist precisely in the ability to employ sophisticated argumentation and complex reasoning in support of judgments that the folk standardly make “intuitively.” All this lends to a growing sense that experimental philosophy needs to reexamine its roots—to rearticulate just what the targets, aims, and methods of experimental philosophy really are. This volume serves as an attempt to do just that. Each of the authors in this volume offers a new perspective on the discipline of experimental philosophy, one that goes beyond the straightforward, armchair-burning pronouncements so frequently suggested by the early days of the movement. Not all are wholly optimistic—the volume contains its share of criticisms of experimental philosophy’s aims and selfimage. The contributions, in fact, frequently represent quite different visions regarding the proper role of experimentation in philosophy—from Turri’s call for the normalization of experiments in epistemology, to Deutsch’s pessimism, to many positions in between.

Introduction

5

Jonathan Weinberg’s chapter sets us off by offering an alternative to the standard reliability-based critique of the epistemological worth of intuitions. Weinberg notes that even a quite reliable source of evidence (of which intuition is, in fact, likely one) can nonetheless be untrustworthy, depending upon the particular inferences one makes from the source’s deliverances. Given this insight, Weinberg argues that metaphilosophers ought to be considering critiques not only of our use of intuition but of other aspects of our methodology as well. Weinberg notes, for example, that theories in philosophy tend to be “exception intolerant”—a single counterexample can often wholly overturn a theory. Even highly reliable sources of evidence may fail to be trustworthy enough for the generation of such theories. To address this problem, Weinberg suggests that philosophers might well consider adopting “exception tolerant” theories rather than searching for ways to improve the reliability of intuition. If they do so, however, they will also need to adjust the forms of inference they employ; namely, they will need to adopt modes of inference that work well on the “messy” sorts of data sets that x-phi suggests intuition generates. Much early negative work in experimental philosophy focused on discrediting intuition by way of demonstrations of bias or inappropriate variation. On such views, one might expect experimental philosophy to play an inherently short-lived role—once traditionalists had become convinced of their folly, appeals to intuition would be abandoned and the work of the experimental philosopher would be complete. In his chapter, however, John Turri offers a quite different criticism of traditional philosophical method. Turri notes that theories in epistemology are frequently judged by how well they comport with our commonsense views about the nature of knowledge, justification and so forth. Turri then reviews several recent studies which indicate that philosophers very frequently mischaracterize how ordinary folks think and talk about knowledge. The current philosophical practice of reliance on introspection and casual observation is, then, clearly insufficiently rigorous. Turri’s conclusion is a critical one, but it is not that intuition (or appeal to commonsense epistemology) be abandoned—it is that experimentation must become part of the standard practice in epistemology. Joshua Alexander’s contribution focuses not on the traditional target of intuitions, but on thought experiments. Alexander examines two competing accounts of how thought experiments work, both from the literature in

6

Advances in Experimental Philosophy and Philosophical Methodology

philosophy of science—the “argument view” and the “mental model” view. Examination of these accounts demonstrates the importance of studying philosophical cognition generally—for instance, the role imagination might play in understanding a thought experiment. The shift to a general focus on philosophical cognition demonstrates an expanded role for experimental philosophy. For, Alexander notes, experimental philosophy just is the empirical study of how we think philosophically. A focus on “intuition” obscures this broader role for x-phi. Alexander ultimately suggests a “cooperative” view of the relationship between experimental philosophy and traditional uses of the method of cases. The second half of the volume focuses broadly on the Centrality debate, leading off with a contribution by Max Deutsch. Deutsch, one of the initiators of the debate, here argues that Gettier’s 1963 article, “Is Justified True Belief Knowledge?,” supposedly one of the archetype examples of “intuitionbased” thought experimentation, in fact does not rely on intuitions as evidence. Indeed, Gettier offers arguments for his claims, not brute appeals to intuition—the primary argument being that the agents in the counterexamples have true beliefs only due to sheer luck, and that this sort of luck is inconsistent with knowledge. One might suppose that, despite Gettier’s intentions, at the very least the post-Gettier literature has treated Gettier’s counterexamples as involving appeal to intuition. But Deutsch argues that both the “Gettierology” literature that Gettier’s paper spawned and the current textbook discussions of Gettier’s cases fail to evidence any such interpretation. Further, contrary to popular belief, responses to Gettier’s paper themselves relied on argumentation, rather than intuition, in supporting their own conclusions. Ron Mallon’s contribution explores whether the impact of experimental philosophy would in fact be undermined by the failure of the Centrality thesis. Mallon focuses on the “argument from diversity”—that is, the argument that the observed variation in intuitive judgments across cultural, socioeconomic, or other groups of persons is prima facie a cause for concern about current philosophical methods. Mallon contends that the effectiveness of this argument is not dependent on the classical, “Platonic armchair”-style picture of philosophical method that opponents of Centrality reject. Mallon suggests that we view the argument from diversity as akin to the problem of

Introduction

7

disagreement among epistemic peers; the issue, then, becomes one of whether we are epistemically obligated to reduce our credence in a proposition in response to the disagreement. From this perspective, the argument remains effective against multiple construals of philosophical method, including naturalistic alternatives to the “Platonic armchair” view as well as views which reject any form of reliance on spontaneous “intuitive” judgment, no matter how characterized. In our joint contribution, Michael Johnson and I attempt to address the Centrality question on a smaller scale by conducting a “case study” of the use of intuitions in one particular field in philosophy—theories of reference. Interestingly, experimental philosophy has prompted a quite separate debate on this topic, one which we suggest has unexplored parallels with the incipient Centrality debate. When Machery et al. (2004) published findings suggesting that persons with East Asian cultural backgrounds are less likely than Westerners to exhibit anti-descriptivist intuitions in response to two of Kripke’s well-known thought experiments, several authors objected that the intuitions Machery et al. studied were of little relevance to the success of Kripke’s arguments. We examine arguments by Michael Devitt, Genoveva Martí, and others, ultimately concluding that the central data for a theory of reference consists not of intuitions, but of applications of terms to things. In other words, the central data consists of linguistic behavior, and can thus quite naturally be studied by way of the methods of experimental philosophy. Nonetheless, since intuition can be good (although not perfect) evidence for what one would say, armchair appeal to intuition can therefore be good (though indirect and non-ideal) evidence for a theory of reference. All of this, however, applies only to the theory of reference—we argue that intuition’s evidential status is more uncertain in other fields of philosophy. Jonathan Ichikawa, like Deutsch, rejects the traditional view that philosophers rely on intuitions as evidence. However, Ichikawa’s reasons differ from Deutsch’s (and indeed from those of other opponents of Centrality)—while Deutsch emphasizes the presence of argumentation in, e.g. Gettier’s work, Ichikawa approaches from the other direction, asking us to consider whether the absence of an occurrence of intuition really impacts an agent’s reasons to believe. Ichikawa argues that agents are often under rational pressure to accept

8

Advances in Experimental Philosophy and Philosophical Methodology

certain propositions even in the absence of any awareness of this pressure or inclination to accept those propositions. If this is so, then, although the occurrence of an intuition might lead to awareness of the rational pressure, it is hard to see how the occurrence of an intuition lends any justificatory weight. Though Ichikawa thus rejects the view that intuitions play an evidential role in philosophy, he—unlike most other opponents of Centrality—holds that experimental philosophy’s importance is not wholly undermined by a rejection of intuition-centered views of philosophical method. Ichikawa notes that, even if intuitions do not play an evidential role in philosophy, this does not rule out the possibility that psychological facts about intuitions may nonetheless be epistemically significant. The existence of order effects in intuition, for example, might lead us to doubt our ability to respond rationally to our reasons—without requiring the commitment that we are relying on those intuitions as evidence. This volume joins several others in the “Advances in Experimental Philosophy” series; its focus, however, is somewhat atypical. The advances with which this volume’s authors are concerned are not, for the most part, a matter of new, exciting empirical results—this volume contains significantly less discussion of actual experimental findings than its predecessors. Instead, the advances are metaphilosophical. They consist of the development and sophistication of different perspectives on the nature and relevance of the entire project of experimental philosophy. More broadly, they consist of new insights on just what philosophy is and ought to be. It’s an exciting time to be involved in this field; one gets the sense that philosophy’s self-image will likely be quite different twenty or so years down the line. It’s my hope and belief that experimental philosophy will continue to challenge, and thereby help shape, philosophers’ views about their own discipline.

Note 1 Interestingly, some of the classic early experiments (such as Weinberg, Nichols and Stich 2001 and Machery et al. 2004) fail to precisely fit this argument type, since they restrict their conclusions to epistemic intuitions and semantic intuitions respectively. However, the basic approach is otherwise similar.

Introduction

9

References Alexander, J. and Weinberg, J. (2007), “Analytic Epistemology and Experimental Philosophy,” Philosophy Compass 2 (1): 56–80. Bealer, G. (1992), “The Incoherence of Empiricism,” Proceedings of the Aristotelian Society 66: 99–138. Cappelen, H. (2012), Philosophy Without Intuitions, Oxford: Oxford University Press. DePaul, M. and Ramsey, W. (eds) (1998), Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry, Lanham, MD: Rowman and Littlefield. Deutsch, M. (2009), “Experimental Philosophy and the Theory of Reference,” Mind and Language 24 (4): 445–66. Deutsch, M. (2010), “Intuitions, Counter-Examples, and Experimental Philosophy,” Review of Philosophy and Psychology 1.3: 447–60. Devitt, M. (2011), “Experimental Semantics,” Philosophy and Phenomenological Research 82: 418–35. Earlenbaugh, J. and Molyneux, B. (2009), “If Intuitions Must Be Evidential then Philosophy is in Big Trouble,” Studia Philosophica Estonica 2 (2): 35–53. Horvath, J. (2010), “How (not) to React to Experimental Philosophy,” Philosophical Psychology 23: 447–80. Ichikawa, J. J. (2014), “Who Needs Intuitions? Two Experimentalist Critiques,” in A. Booth and D. Rowbottom (eds), Intuitions, Oxford: Oxford University Press, 232–56. Ichikawa, J. J. and Jarvis, B. (2013), The Rules of Thought, Oxford: Oxford University Press. Kauppinen, A. (2007), “The Rise and Fall of Experimental Philosophy,” Philosophical Explorations 10: 95–118. Ludwig, K. (2007), “The Epistemology of Thought Experiments: First Person Versus Third Person Approaches,” Midwest Studies in Philosophy 31: 128–59. Machery, E., Mallon, R., Nichols, S., and Stich, S. (2004), “Semantics, Cross-Cultural Style,” Cognition 92: B1–B12. Pust, J. (2001), Intuitions as Evidence, New York: Garland. Schulz, E., Cokely, E., and Feltz, A. (2011), “Persistent Bias in Expert Judgments About Free Will and Moral Responsibility: A Test of the Expertise Defense,” Consciousness and Cognition 20: 1722–31. Schwitzgebel, E. and Cushman, F. (2012), “Expertise in Moral Reasoning? Order Effects on Moral Judgment in Professional Philosophers and Non-Philosophers,” Mind & Language 27: 135–53.

10

Advances in Experimental Philosophy and Philosophical Methodology

Swain, S., Alexander, J., and Weinberg, J. (2008), “The Instability of Philosophical Intuitions: Running Hot and Cold on Truetemp,” Philosophy and Phenomenological Research 76: 138–55. Tobia, K., Buckwalter, W., and Stich, S. (2013), “Moral Intuitions: Are Philosophers Experts?” Philosophical Psychology 26: 629–38. Weinberg, J., Nichols, S., and Stich, S. (2001), “Normativity and Epistemic Intuitions,” Philosophical Topics 29: 429–60. Williamson, T. (2007), The Philosophy of Philosophy, Oxford: Blackwell. Williamson, T. (2011), “Philosophical Expertise and the Burden of Proof,” Metaphilosophy 42: 215–29.

2

Experimental Philosophy, Noisy Intuitions, and Messy Inferences Jonathan M. Weinberg

Much discussion about experimental philosophy and philosophical methodology has been framed in terms of the reliability of intuitions, and, even when it has not been about reliability per se, it has been focused on whether intuitions meet whatever conditions they need to meet to be trustworthy as evidence. But really that question cannot be answered independently from the questions: evidence for what theories, arrived at by what sorts of inferences? I will contend here that not just philosophy’s sources of evidence but also its inferential resources are in great need of closer examination.

1. The methodological inadequacy of “reliability” Consider the fundamental methodological question of when one should or should not trust a source of evidence that delivers the output P: to what extent should we go ahead and accept P on the basis of a source’s say-so? In trusting a source of evidence, we allow ourselves to accept P itself on the basis of its deliverance that P—always in a manner with the potential to be overturned by the vagaries of further research; and perhaps only with some barrage of qualifiers in here, such as pro tempore, or for the purposes of inquiry, or in accord with some conventional standard. Now, one might reasonably have hoped that we could just import the notion of reliability wholesale from epistemology into a methodological principle of when and how much to trust a source of evidence, namely, one ought trust a source about P to the extent that it is reliable about matters of P.

12

Advances in Experimental Philosophy and Philosophical Methodology

A source that delivers a higher proportion of truths is one that can be more responsibly relied upon in inquiry than one that delivers a lower proportion. This would of course be subject to our best information about subdomains or circumstances in which the source is of greater or lesser reliability, but the core idea would still be: one can responsibly make use of a source only if, and at most to the extent that, it is reliable. Something like this idea—that reliability is the determinative consideration when deciding when to make use of a source—is in the background in much recent discussion about methodology and intuitions. For example, it is at the heart of Boyd’s and Nagel’s (2014) defense of the appeal to intuitions in epistemology, and plays the sort of central role one would expect in Goldman’s (2010) discussion of philosophical methodology. And it seems like it would be nice, in at least two ways, if reliability really were the main determination of methodological trustworthiness. First, it can seem like a straightforward consideration to deploy (at least if we set aside theoretical concerns like those in the vicinity of the Generality problem). Second, it seems highly plausible that intuitions on the whole are in fact reliable, and indeed to try to argue otherwise is to put oneself in danger of collapsing into a very broad skepticism, should intuitions turn out to undergird a very broad swath of our cognition, as many philosophers have claimed (e.g. BonJour 1998; Bealer 1998). But things turn out to be not so simple. For any degree of reliability noticeably less than perfect is consistent with a high degree of untrust­ worthiness, and any degree of reliability noticeably greater than nil, in which at least some detectable information about the target is present, can be consistent with trustworthiness. I am arguing here for the methodological inadequacy of the notion of reliability. My claim is that if all we know about a source is its baseline reliability, that by itself cannot settle the question of whether and when to trust it. Low baseline reliability is consistent with trustworthiness, and high baseline reliability is consistent with untrustworthiness. Further sorts of information beyond baseline reliability are crucial in making determinations of trustworthiness, most particularly about the instrument itself, and about our general theoretical situation. Of the two directions, it is easier to see how a low reliability source might nonetheless be trustworthy, in the right setting and given the right additional information. Suppose that we know that the reliability of a given device is 51



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

13

percent, say, in terms of telling whether a given sample is an acid or a base. (For ease of exposition I will just consider cases where there each datum is a simple binary, but the argument easily expands, mutatis mutandis, to more complicated sorts of outputs.) But allow that we know that the device’s reports are both repeatable, and so, for example, it does not totally use up the sample in any trial. Moreover, the reports are independent even when applied to the same sample, and so, it will not necessarily just give the same answer every time, when re-applied to the same sample. Moreover, suppose we know it is not subject to any biases in its readings, in that its mistakes are a matter of chance and not, say, on average more likely to mistake an acid for a base than vice-versa. Under such assumptions, one can apply the device over and over again to the same sample, and, while each individual report may have a 51 percent chance of being correct, we know the average response of the device to any given sample will be increasingly likely to be true as the number of readings increases. This is a straightforward application of the Condorcet Jury Theorem: specify any degree of reliability you want to achieve, and you can determine easily how many trials you must perform on any given sample in order to be able to infer to its overall pH status to that degree.1 But note that we had to know more than just the baseline reliability of the source here—in particular, the independence of its deliverances is an essential part of the situation that allows this kind of aggregation to work. Any one reading may not be sufficiently trustworthy on its own, but we can build a trustworthy practice based on its readings when further founded on our knowledge of various aspects of the device, most particularly what kinds of errors it is or is not likely to make.2 So far, so good—it should not be surprising that further information of the right sort can allow us to leverage better inferences out of a highly noisy instrument. But what about the other direction? How can a highly reliable source nonetheless not be trustworthy? One easy way to arrive at such a situation is when the source’s output is wildly in conflict with other evidence we may already have; for example, if an apparently well-conducted experiment seems to indicate the reality of ESP, despite massive amounts of already existing evidence to the contrary. (See, e.g. Wagenmakers et al. 2011 response to Bem 2011.) But suppose that one has no specific evidence against some hypothesis H, and in fact H starts off at least as likely as every other rival

14

Advances in Experimental Philosophy and Philosophical Methodology

to H. And then a highly reliable source provides stark evidence for H. Should we not then accept H, at least for the purposes of inquiry, to the extent of that source’s reliability? But it turns out, again, that features of the theoretical situation can still make it inappropriate to extend our trust so far. I will use an extreme case to make the point stark. Let us start with a device much more reliable than the one in the previous case, in having 95 percent specificity and sensitivity for both acids and bases: out of every 100 samples of acids, it will correctly give the reading that it is an acid ninety-five times, and likewise for bases. But it is also unlike that first device, in that its results are not independent, should we re-apply it to the same sample: if it first outputs that a sample is an acid, it will always say that this is the case, no matter how many times we check. Thus we cannot ramp up its reliability by way of repeated observations, like we could with the first device. Now, suppose we have arrived at a point in inquiry where we have a set of fifteen samples, and, for each permutation of acid/base predictions for the members of that set, we have an equally probable hypothesis that makes exactly that prediction. Let us assume that we are certain that the truth must lie among those hypotheses, though, so their probabilities will sum to 1. So, one hypothesis predicts that they are all acids; another, that they are all bases; another, that the first four are acids and the latter nine are bases; and so on for all 2^15 permutations here. Now, we perform the measurements, and our device reads that all fifteen are acids. Let H be the hypothesis that makes exactly that prediction. So, its predictions appear to be exactly confirmed, and it is the only hypothesis whose predictions are so confirmed, since every other hypothesis will diverge from the readings for at least one sample, and many of them for more than one. It can thus seem that of course we should go ahead and endorse H, on such evidence, from such a highly reliable source. But a quick calculation reveals that, although we should correctly view H as decidedly more likely than any single one of its rivals, its probability on the evidence remains only (0.95)^15—just a bit above 46 percent. In fact, if we consider the set of hypotheses that diverge from the observations by either exactly one or exactly two readings, it is somewhat more likely that the correct hypothesis is to be found somewhere in that set, than it is that H is



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

15

the true one. We cannot trust the device here, with regard to H, certainly not anywhere near the level that the device’s high reliability would have antecedently suggested to us. That is not to say that the device has been epistemically worthless—far from it. Of what had originally been 2^15 hypotheses, we can consider these observations as having pretty much ruled out about 2^10 of them, and to have given us a sense of which of the remaining competitors are more likely or less so, with H having pride of place among the former. Beyond that, though, we must view our inquiry as still very open. The instrument is highly reliable, and while one, and only one, hypothesis predicts the observations that the instrument provides for us, nonetheless we find ourselves very far from arriving at an answer as to which hypothesis is the right one. Just how far depends on what sort of threshold of likelihood we consider acceptable. I had to use a dizzyingly large hypothesis space in order to get that ultimate probability below 0.5, because I wanted to find an example to give us that stark a mathematical result. Yet I expect that philosophers do not in general want to consider anything-above-0.5 as a sufficient degree of confidence for accepting a theoretical result. If we set that threshold instead at 0.8 (which is probably still rather too permissive a standard), then merely five rival hypotheses would be enough to keep us below it, for this degree of reliability.3 Thus, no degree of poor reliability is sufficient to rule out trusting a source of evidence in inquiry, so long as it is at least modestly above chance. Nor is any high degree of reliability sufficient to establish trustworthiness, so long as it is at least in any practical sense less than utterly certain. One must attend to more than the general reliability of a source of evidence when deciding whether and how to put it to methodological work. These cases draw our attention to important aspects of methodology that a purely reliabilist account of source selection cannot accommodate. The key moral of these two cases is simply that, when we are deciding whether or not to rely on a source of evidence for guiding our inquiries at some stage, it is just not enough to know how reliable it is (again, outside of knowing that it is perfectly reliable or perfectly unreliable or close enough to either for practical purposes). I take it that we seek reliability not just at the level of inputs to our processes of theory selection, but moreover—and more importantly—at the level of the outputs of such processes. We are trying to

16

Advances in Experimental Philosophy and Philosophical Methodology

arrive at a theory that, at a minimum, is highly likely to be true. But these cases teach, first, that further methodological resources can yield reliable theory selection from not-especially-reliable data; and second, that weaknesses in our inferential resources can make even a highly reliable source nonetheless inadequate to our theoretical demands. I will argue that philosophy is currently in something like the latter situation—but can take a lesson from the former situation about using further resources to supplement a methodological weakness. Here is a very useful observation from Weatherson (2003), on the extent to which many areas of philosophy seem perhaps to rely too much on the power of a small number of data points in guiding theory selection: In epistemology, particularly in the theory of knowledge, and in parts of metaphysics, particularly in the theory of causation, it is almost universally assumed that intuition trumps theory. Shope’s The Analysis of Knowledge contains literally dozens of cases where an interesting account of knowledge was jettisoned because it clashed with intuition about a particular case. In the literature on knowledge and lotteries it is not as widely assumed that intuitions about cases are inevitably correct, but this still seems to be the working hypothesis. (Weatherson 2003: 1)

One might reasonably object that this at least a little bit overstates the degree to which current philosophical practices in these areas are vulnerable to one-off counterexamples (and Weatherson’s paper on the whole is sensitive to such concerns). For example, it is reasonably standard to attempt to explain a recalcitrant intuition away, or to say that one’s theory is still the best overall fit, even if it misses a few cases (since, in general, everyone’s theory misses a few cases). But I fear that the standard moves here actually underscore how weak our resources are for handling the possibility of intuitive error. I am not sure that we have ever, in the history of the profession, had a case where there were tough cases for everyone, but someone said that the other side had the better overall trade-off. It is just too easy to impeach another’s theory without being able to make the charges stick sufficiently to convince that one’s own view must be adopted in its stead. We have at present no good resources for debating and evaluating philosophers’ appeals to which theory has the best overall fit, and, when we make such appeals, they are often with highly cherry-picked evidence sets. We don’t have any general consensus as to what



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

17

rules to use for scoring different failures—why missing this intuition should count worse against a theory than missing that one. We give lip service to the strength or centrality of a case, but again we have no means to referee the frequent disagreements about whether a case is really, really central, or how strong an intuition really, really is. We philosophers are thus in a situation like my toy example of a reliablebut-not-trustworthy source above. Across a wide range of domains, reliance on intuitions has perhaps narrowed down our set of live hypotheses to a fairly compact number. Even granted a high level of baseline reliability, intuitions would not have sufficient power of resolution to enable us to decide well between the resulting competitors, and we currently lack the further methodological resources to do better. Since the mere reliability of our intuitions has not proved adequate to our methodological needs, it is time to see whether—and how—we can do better.

2. Thinking about methodology At a very high level of abstraction, we can think of inquiry as involving a space of theories within which to search for the one that is maximally likely to be true, and the primary—but not sole—determinant of our search is the evidence coming in from our sources. At this nosebleedingly-high altitude, we can identify at least four components to the norms governing evidence-driven methodologies.4 None of this is meant to be at all specific to philosophy, and indeed I will illustrate it briefly with some parallels in linguistics; and my intention is that this abstracted picture should speak to many methodologies in the empirical sciences. Also, just to be clear, in identifying sets of norms that are operative in a methodology in a given community is not thereby to endorse those norms as correct. A. Source selection: What sources of evidence are to be taken as credible, and with what sorts of initial weightings? For example, in classical syntax, certain sorts of linguistic judgments by native speakers are given a high degree of evidential weight, such as the acceptability or unacceptability of particular sentences, and not so much on those speaker’s judgments about why any given

18

Advances in Experimental Philosophy and Philosophical Methodology

sentence strikes them as acceptable or not. In philosophy, of course, there is a parallel question regarding intuitions about cases, which then play out further in such debates as whether folk intuitions are to be trusted over those of trained philosophers, or the relevance of how esoteric or far-fetched a case might be, and so on. B. Theory structure: What does the space of competitor theories look like? What forms of theories are to be valorized, and which discouraged? Theories of grammar for a natural language in classical syntax are computationally structured and discrete. One may contrast this with functionalist grammars, which are both highly attuned to frequency of usage and admit a great deal of gradation. Linguists disagreeing across this theoretical divide may still agree as to what the relevant data are to which their theories must answer. In philosophy, we can see debates as to whether and when our theories should take the form of necessary and sufficient conditions, or have modal force; or also just how complicated and baroque a theory of a philosophical domain can be without counting as somehow beyond the pale. C. Modes of inference: What rules of inference, and criteria of theory/modelselection, are taken as canonical? How does one proceed from the deliverances of the sources in A towards winnowing down the set of competitor theories in B? For classical syntax, theories must cleave fairly close to the observed data of linguistic judgments, but there are specific resources for accommodating apparently errant cases as the product of performance errors, computational resource limitations, and the like. Grammars must also be learnable by child speakers on positive data, and should avoid too much language-specific machinery. There seem to be fairly close parallels here in philosophy, with perhaps further work to be done by constraints about logical self-consistency. D. Vindicating metatheory: What reasons does the community offer to itself, and to other communities of inquiry, as to why it makes sense for them to use the methodology as described by A, B, and C? Continuing with our example, here one would appeal to such theoretical commitments as the competence/ performance distinction, or the nativist apparatus for explaining language acquisition, or the modularity of syntactic processing. For philosophers, it can



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

19

matter here a great deal whether we take ourselves to be providing an account of our own concepts or of something robustly extra-mental. With this rough and abstract framework of theory-selection-as-search in place, let me offer the following analysis of the methodology I will term the classical philosophical appeal to intuitions (CPAI): the norms for data, inference, and theory of classical analysis all fit each other very well. Our main data are verdicts about particular cases, as to whether such cases do or do not fall under particular categories, as typically provided by armchair consideration of the cases in a manner distinguishable from recalling those verdicts from previous investigation, or accepting them from another’s testimony, or arriving at them by means of explicit inference from applying one’s theory of the category in question. Whatever our psychological capacity may be that gives us access to such verdicts in such a way, many philosophers seem happy to call it “intuition” even while disagreeing starkly as to just what underwrites that capacity. (One can contrast George Bealer’s 1998 rationalist account in terms of concept mastery; Jennifer Nagel’s 2012 appeal to the notion of epistemological intuitions as implemented in our tacit domain-specific capacity for “folk psychology”; and Williamson’s5 2008 approach in terms of our general capacity for judgment.) For the purposes of analyzing this practice here, that term will suit us fine. Although we are certainly not limited to intuition as a source of evidence in philosophy, and it is generally legitimate to supplement with other sorts of evidence, it is still the most basic sort of evidence that our theories must be responsible to in this methodology. And, when we do advert to other evidence while using this methodology, we often do so in the service of evaluating some reported intuition as legitimate or not. For example, when we test an intuition for perhaps being driven by pragmatics, we are not so much looking to bring a second line of evidence to bear on the matter as to place more scrutiny on our main line of intuitional evidence in the first place. We can thus see that this methodology is committed to a high-but-not-practically-infallible level of reliability for these intuitions. I am thus taking it that certain sorts of anti-intuition metaphilosophers are simply wrong as a matter of description as to what philosophers do.6 But the analysis is consistent with those who want to deny any central evidential role for the intuitions themselves—that is, facts of the form that philosophers

20

Advances in Experimental Philosophy and Philosophical Methodology

might commonly report by saying something like, “I have an intuition that ….”7 The account here is consistent either with taking such psychological facts to constitute our evidence itself, or, instead, to be the source of the case evidence but not itself the weight-bearing part of our evidence. (Such facts will still be among our evidence for these philosophers, since they endorse E=K, and these psychological facts about we do or don’t report as intuited are among what we know.) On the former sort of anti-intuitionism, it doesn’t really make sense that philosophers should bother with debating whether a given reported intuition does or does not come from the right kind of psychological mechanism. On the latter sort, however, it is highly analogous to scientists worrying about whether a given observation is or is not an artefact of their experimental apparatus. As for our theories, there is a strong norm for them to be logically sharpedged—that is, to be exception-intolerant,8 in a manner often secured by way of universal quantifiers and necessity operators. That almost all instances of knowledge are also instances of justified, true belief is taken to be a fairly worthless epistemological generalization if it can only be defended in that hedged and partial form. There is also a norm preferring more compact theories to more expansive ones, in which the latter might get tagged negatively as, say, “baroque,” and especially run the risk of suffering accusations of being ad hoc. Each of these norms makes sense in the light of the other: point-data are putatively armchair-accessible, and can put severe pressure on logically sharp-edged theories. And, by allowing only a fairly modest degree of complication, with strong norms against baroqueness or being ad hoc, and moreover with strong norms in place encouraging rigorous exploration of possible counterexamples, we can take it that a theory that survives such a process of counterexample testing should thereby legitimately have its credibility raised. In our terms here, our norms of what good theories can look like are serving in part to provide a theoretical context in which a small number of cases can properly be determinative for selecting the best theory. Moreover, a vindicating metatheory in terms of something like our mastery of the relevant concepts is further backed by both the sharp edges of the theories, and the plausibility that this methodology can legitimately be practiced from the armchair.9 In turn, it legitimates the use of off-beat and esoteric cases. While verdicts in such cases can go little distance, if any, in



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

21

determining, say, what is typically true of some philosophical category, or true in normal conditions, they are sufficient to threaten the right sorts of claims pitched in a more demanding modality. And all of this dovetails nicely with our desire to be able to use our well-developed logical tools to extract further consequences, perhaps even surprising ones, once we have amassed a trove of such theoretical results. I am trying to draw our attention to some of the ways in which the reliability of intuition is only one part of what is needed to make the method of cases make sense—an integral part, surely, but not enough by itself without a number of other closely integrated pieces of methodological machinery. But when all the pieces of a methodology fit together so tightly we should not be surprised if a need to modify one piece of the methodological puzzle will require alterations in the others as well. My contention is that our growing awareness of just how much noise there is with our data will require serious reconsideration of these other aspects of the method. In particular, we find ourselves with a mismatch between, on the one hand, just how subtle the differences are between our competitor theories, and just how formally stringent they are, and, on the other hand, just how noisy our primary data source is here. There is probably not much we can do purely from the armchair to raise our baseline reliability much higher than it is now— although I would emphasize that that is in part because we have been able to discover at least some potential sources of error, and to modify our armchair practices accordingly. (To take just one example, it is customary to deploy bits of formal notation in philosophical writing to help avoid scope ambiguities, even outside the context of explicit logical proofs.) We don’t have any reason to think that even more armchair reflection on intuition will root out very much more of the noise that remains. So how else can we try to address this revealed disconnect within the methodology of cases? There is some promise within experimental philosophy of using experimental methods and a greater understanding of the psychology of intuitions to amplify our powers of error-detection here. This is the most methodologically conservative approach, in that all of the x-phi work would be aimed at rendering a cleaned up data set to which our traditional modes of inference could apply. The objective is still to provide sharp verdicts as to what is or isn’t a case of knowledge, or free action, and so on, and the other aspects of

22

Advances in Experimental Philosophy and Philosophical Methodology

the methodology as sketched above could remain pretty much as they are. Other than its abandonment of the armchair, the resulting methodology otherwise retains all the core features of CPAI. I pursue the prospects for such an approach elsewhere.10 But here, let me consider what work might be done by making adjustments elsewhere in the methodological matrix.

3. Can we make room for messy theories? Suppose we decide that we need to start acknowledging that our intuitive case-verdict data really is noisy, and seek to accommodate that situation in our theorizing. We would need to revise CPAI in way that no longer relied on a fairly small number of well-targeted counterexamples to put lethal pressure on rival theories. We could thus consider a broader set of structures for our theories—in particular, ones that are exception-tolerant—and see what additional resources they allow us to bring to bear. Before we do so, though, we should ask whether we have good reason to think that philosophical truths must generally have an exception-intolerant form. What other work do we want philosophical theories to do for us, beyond being true? We want them to be “illuminating,” or perhaps “perspicuous”? Surely, but then is there a reason only logically sharp-edged theories can do that? Of course soft-edged theories won’t generate surprising theorems nearly as copiously as sharp-edged ones—a small set of propositions with just a few interlocking universally-quantified pieces can produce a truly astounding amount of structure, as the Peano axioms nicely illustrate. Replace “All numbers have a successor” with “Most numbers have a successor, typically,” and you can’t really get too far in your number theory. But it is not clear why propensity to rich formal treatment should be a requirement on our theories. Theorems are wonderful, both in their own right and where they can legitimately be put to work, but most of human knowledge just does not take that form. And, for that matter, even the formal sciences have their own devices of exception-tolerance to deploy where they may be useful, such as the highly useful mathematical notion of conditions that hold “almost everywhere.” Moreover, the same aspects of logically sharp-edged claims that make them amenable to proofs also seems to make them susceptible to paradox!



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

23

Weaken the premises of most paradoxes to less than strict universality, and the paradox disappears. Moreover, we can still use formal tools where appropriate, including formal models; we will just recognize that such models, in virtue of their particular formal structure, may be making claims that are sharpened past the point of actual philosophical accuracy. This should not count against such models even a little bit, since we already often expect such models to be simplifications of, and abstractions from, their target philosophical truths, and there is nothing wrong with that. One might think that the modal force of philosophical truths is an important feature that they have. But propositions that are themselves exception-tolerant can still hold with various sorts of modal forces. Depending on one’s views about ceteris paribus laws, one might think that most natural laws that our sciences actually traffic in are exception-tolerant. Exception-tolerant propositions can also support counterfactuals: if we were now to be set upon by tigers, then it is highly likely they would have four legs (and so, say, we would not expect to be able to outrun them). Perhaps philosophers are especially interested in very compelling subsets of modally strong claims: claims of identity. If our goal is to establish a claim that starts “knowledge is …,” where that “is” is an “is” of identity, then we would indeed seem to fail that goal and fail it badly should the right-hand side pick out a class that was not at least extensionally equivalent to knowledge. But should we really want to be only, or even primarily, interested in theories that express identities? It’s been a long time since Socrates, after all; and philosophers as otherwise dissimilar as Jerry Fodor and Timothy Williamson have raised very good doubts as to whether many such philosophical truths will be forthcoming in any meaningful way. Even if we decide that such identities are worthy of philosophical pursuit, it does not follow that they are the only kind of theoretical result worth achieving in philosophy. For I wonder whether we are, by refusing to make heavier use of soft-edged claims, leaving important philosophical facts unused on the floor. There are counterexamples to KK, sure, but is it not an important fact about knowledge that, say, KK is true of the vast majority of knowledge claims that we self-consciously deploy, such as in argumentation? Whether or not all justified true beliefs are knowledge, surely there is a mammoth overlap between JTB and K. Is that really not a result worth taking epistemological notice of? These questions are not rhetorical; I

24

Advances in Experimental Philosophy and Philosophical Methodology

don’t think it is obvious what the answers to them are. But what I want to draw our attention to here is that, in particular, it is not obvious that the answer to such questions must always be no. One kind of such structure that already has some currency in current philosophical practice is that of “family resemblances” or “cluster concepts,” where we do not look for more from an analysis than a rather motley set of disjuncts that may have little to no deeper unity. But, outside some recent reemergence in philosophical aesthetics, this approach has not generally proved an attractive framework.11 This perhaps makes sense: in many parts of aesthetics, one expects the concepts to twist and wind around our contingent and all-too-human practices of art creation and appreciation. We do not necessarily expect a category like art or (even more) horror film to have too much depth beyond our historically-configured patterns of attribution. For categories like knowledge or causation, though, philosophers seem in general to expect more organic unity. There are of course exceptions to those general trends,12 and the issues under discussion here may be cause for re-think about the merits of such theories. But let us see what other options can be put upon our methodological table. One candidate that should be considered here is generics. Even if we grant that K=JTB can’t be true as an exceptionless universal, one might nonetheless be struck by the weirdness of Gettier cases on the whole, and wonder whether knowledge is justified true belief might still be a useful piece of epistemological lore when considered as a generic. One very attractive feature of generics is that multiple inequivalent generics can all be true. Tigers have four legs and tigers have stripes are both true generics about tigers, for example. Similarly, knowledge is justified true belief and knowledge is true belief caused by a reliable process are rival claims when understood as necessary and sufficient condition analyses—but they are both surely true if understood as generics.13 Could we perhaps take the whole class of major rival theories of what knowledge is and turn the theories instead into a compilation of valuable generic claims about knowledge? With this one pirouette, we would transform the state of our inquiry to one mainly characterizable in terms of our ignorance as to the One True Theory of knowledge, into a copious stockpile of epistemological discoveries. Maybe this would be an unacceptable shrinking of philosophical ambition. But maybe, instead, the



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

25

feeling that philosophy’s progress is painfully slow is an artifact of a slightly hyperbolic set of expectations as to what philosophy should deliver to us; and those expectations themselves a false hope produced by too narrow a focus on one form that our theories can take. Now, philosophical generics would perhaps be impossible to test well with CPAI, since they are counterexample-tolerant. But the question does immediately arise: how well can we test them at all, even with a suitably modified methodology? Would we need to bring in a notion of proper function, or statistical normalcy, or what? One methodological upshot would be that esoteric sorts of cases would be seen as of little or no evidential value, akin to trying to look at the amputee ward of the zoo when evaluating that generic about quadruped tigers. All in all, it would not be a trivial methodological change to aim at generics instead of strict generalizations; indeed, the main moral of this paper is that methodological changes in general ramify more broadly than we have heretofore considered. Let me put one more theoretical option on the table, drawn more from empirical psychology than from philosophy of language: weighted feature sets, such as prototypes. One interesting fact about the growing set of x-phi studies about Gettier cases is that, in many instances, the Gettier cases come out generally on the “not knowledge” side of the scale, yet often not nearly so far as other sorts of cases, such as falsehoods or lucky explicit guesses. Indeed the overall pattern can be a bit embarrassing for defenders of the classical analyses of knowledge, since, at least sometimes, the folk’s epistemometer needle leans equally far towards “not knowledge” for some paradigm Gettier cases as for cases which are supposed to count as knowledge, such as skeptical pressure cases.14 If we are theorizing knowledge in terms of a weighted feature set instead of strict generalizations, we could allow that Gettierized could be a feature that applies some negative impetus against counting as knowledge, while perhaps accommodating how that might be a notably weaker antiknowledge factor than, say, falsehood of the belief. Such a treatment might also prove a valuable way to include in our theory of knowledge factors that have been proposed as partly determinative of knowledge, but which have proved elusive to detect experimentally, such as stakes effects. But this line of thought brings us to the greatest danger of trying to theorize in terms of weighted feature sets, especially if we will allow in features that

26

Advances in Experimental Philosophy and Philosophical Methodology

are not weighted especially heavily. Namely, we may find ourselves unable to determine which psychologically subtle factors to include as part of the theory of the category, and which to exclude as mere noise. Perhaps the major lesson from “negative program” x-phi is that there are lots of funny little effects on offer out there, lots of ways in which our attributions of philosophically important concepts can be unexpectedly manipulated. We would need a principled way to include what we take to be legitimate factors in our theories (such as, perhaps, Gettierization and stakes) and still exclude oddball ones (like order or font choice). I have argued elsewhere for the possibility of using effect size for help in separating signal from noise in our intuitions, but, on this specific proposal of using weighted feature sets, especially if we are looking to use fairly modestly weighted features, that likely won’t do the trick.

4. What kinds of inferences can select between messy theories? Now, if all we do is broaden the set of competitor theories to include various sorts of exception-tolerant ones, then we will only have made matters worse, with more theories to choose from and without any relevant tools to help us with that choosing. We must consider alterations not just to the forms of our theories but to our modes of inference. In particular, we require modes of inference that can better deploy noisy data. Perhaps our best model of making inferences with noisy data is modeling itself. If we adopt a view of philosophical inquiry as a kind of model-building, then inference should be a matter of model-selection, and we can look to use the tools of model-selection in our philosophizing. One might think we are already kinda sorta doing that, but then we’re doing it kinda sorta not very well. What we lack are decent metrics of goodness of model here. We mostly use binary data, so all we have is whether the model hits or misses at any given point, and we don’t really have any way of making any comparison between several different models that all have different sets of hits with some misses. Philosophical fans of the model-building model often suggest that we can and do appeal to broad, holistic criteria of a good model à la Quine.15 I am friendly to that general idea, and am not looking to raise difficulties about it here in terms of whether, e.g. parsimony is properly understood as a truth-conducive



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

27

property of a theory in philosophy.16 Even setting such concerns aside, however, I want to present two problems with the idea that holistic theoretical virtues can do the necessary work of theory selection. First, in the absence of any sort of further resources to discipline our evaluations of overall holistic quality, I am not sure that we have any right to expect our appeals to such evaluations to bring in as much psychological noise as those same noisy intuitions that the holistic judgments are supposed to help us overcome. Confirmation bias in particular will be a huge threat here, and it is a kind of bias that we know is not overcome well just by being smarter and thinking harder.17 And just ask yourself: when was the last time you saw a participant in a philosophical debate take a look at the holistic character of the state of play and plump for an opponent’s theory as a result? Not that it never happens, but it is perhaps notable just how notable it is when someone does change their flag from one ism to another. We should expect also that various path-dependencies and order effects play an outsized role. I need to offer two caveats in understanding this objection. First, I am not saying that we can never make good use of holistic judgments of theory quality. But my concern is that the noise to signal ratio will likely be poor when we attempt to use such judgments to choose between a small set of elite contender theories. We are in a situation highly similar to my example above of reliable-but-untrustworthy sources. Second, this issue is not quite the same problem as that of the incommensurability of different weightings of the theoretical virtues, as it would arise even between rival theorists who agree about those weightings. However, the existence of those different weightings does exacerbate the problem. In general, the more slack we allow in the rope of cognition, the more tightly our minds will bind themselves to whatever theory they happen to prefer. My second objection against the idea that holistic theoretical virtues can do the required work is less psychological and more methodological. I am concerned about a potentially misleading picture favored by that idea’s proponents in which we imagine applying those virtues after the evidence has done all of its work already. But considerations of simplicity and parsimony actually need to be brought to bear both more rigorously, and much earlier in inquiry, than this picture allows. We are used to thinking of them as very “big picture” sorts of considerations, yet in fact, when we are reasoning from noisy data, in

28

Advances in Experimental Philosophy and Philosophical Methodology

addressing the problem of overfitting they must be used at a stage much closer to the initial analysis of the data itself. Philosophical theory-selection and empirical model-selection are highly similar problems: in both, we have a data stream in which we expect to find both signal and noise, and we are trying to figure out how best to exploit the former without inadvertently building the latter into our theories or models themselves. Under-utilizing the signal is one kind of danger—but clinging too close to the precise contours of our data stream is yet another. This is just the trade-off we have been discussing. But where there are quantitative data and models, quantitative tools can be brought to bear. Our lack of any substantive means for evaluating the fit of models in philosophy contrasts sharply with the range of tools available in the sciences. For example, one standard measure for model selection is the Akaike information criterion. Suppose we have some ranges of models which vary from each other in two key ways. First, they deploy different numbers of parameters. And, second, they vary in their likelihood—intuitively, how much the parameter choices agree with the observations, measured in terms of how probable the observations are, on the assumption that the model is correct. (Note that this is not the same as how probable the model is, on the assumption that the observations are correct.) The AIC is defined as 2k—2ln(L) where k is the number of parameters, and L is a particular measure of likelihood.18 (Don’t even begin to worry about the 2’s, or the natural log in there, for our purposes here!) Speaking very roughly, the idea is that, when considering some range of competitor models, one ought to prefer the one that minimizes AIC, and there are ways of measuring just how much preference to give for any set of competitor AIC values. In general, models are rewarded for closeness of fit to the data, since a closer fit means a greater maximum on the likelihood function, which in turn means a lower AIC (since the 2ln(L) term is subtracted). And models are penalized by the number of parameters they have, largely to hedge against the risk of overfitting the data. There’s nothing special or magical about AIC, and there are a variety of other tools out there. I am using it as an illustration of the general way such tools work: by having a measure of fittingness for competitor models, and a penalty for additional parameters, and a way of putting those together to yield an evaluation of the bang-for-the-buck of any new proposed way of capturing



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

29

more observations by increasing the number of parameters. Here’s the thing: in order to use any tool of this sort, like AIC, we need some way of generating a quantitative measure of likelihood for philosophical theories. This is not something that we have generally looked to do, certainly not outside of x-phi, and not even that widely within x-phi. Let’s turn to Gettier cases again.19 I mentioned above the problem that the effect size of Gettierization is in some places of a comparable size to effects that need to be considered as potential noise effects. There are further reasons to think that the evidence on the whole is pointing towards at least some Gettier cases counting as non-knowledge, but in a way that is rather messier than the received philosophical view of such cases would predict.20 For the sake of the illustration, let us simplify matters drastically and stipulate that we have two rival theories: JTB and JTBW, in which both (i) agree on all of the non-Gettier cases, and (ii) which, as their names suggest, have a clear asymmetric relationship in terms of their parameters, in which JTBW has all three components that JTB has, plus one more, a warrant condition that steers the predicted attribution values in a negative direction for at least some Gettier-type cases. A question we need to be able to answer is: how much of a better fit must JTBW be, in order to license our preferring it over JTB? And I do not see how, at this point, we even know how to begin answering that question in an intellectually respectable way. We measure these trade-offs in entirely hand-waving, it-seems-to-me kinds of ways—which is to say, we don’t really measure them at all. Among the methodological innovations that we need in the face of the fact of intuitional noise, we need tools to address such questions. The holistic criteria cannot do this work for us if we can appeal to them only in an unregimented manner. Yet we may legitimately hope that more richly quantitative data would allow for just the sort of rigorous, quantitative measures that are called for here. (In principle such quantitative data could be possible from the armchair, should we adopt norms of intuition reporting that standardized some sort of scale of strength of intuition or something like that, perhaps even on a model with the fairly simple */?/?? system used in linguistics. But I think we have reason now to think that there would be a lot of disagreement within given communities about a great many cases and just how strongly felt various intuitions are. We should also suspect that these will be susceptible to all sorts of

30

Advances in Experimental Philosophy and Philosophical Methodology

sources of error, such as order effects and confirmation biases, in a way that cannot be managed from the armchair but which can be filtered out using the methods of the social sciences.) This seems to me an important future priority for experimental philosophy. I think an early hope—certainly an early hope I had harbored— was that experimental investigations would, in good time, reveal individual cases to be problematic, or not. Further investigation into the problematic cases would thereupon lead to their being either settled or discarded, or, on, hopefully rare, occasion, relativized. Had things worked out that way, then x-phi could have played an important role in cleaning up philosophy’s data set, but then our more traditional modes of inference might have remained untouched. But while that first, problematizing part has clearly happened, by and large very few cases have been either settled by means of x-phi or shown to be so noisy as to be beyond hope of further research. As I mentioned earlier, some version of the cleaning-up, separating-wheat-fromchaff project is surely still worth pursuing. At the same time, we should pursue other forms of theories, and appropriately altered modes of inference and model selection, in case the noise proves (as it has thus far) tricky to factor out of our evidential base.

5. Conclusion Experimental philosophy is helping teach us just how messy philosophical practice is. I don’t think we made the mess—I think we are revealing a mess that was already there, but which was, as it were, swept under the rug of logically sharp theories. These theories are often elegant, even beautiful, but they turn out to rest only awkwardly atop the lumpy, bumpy floor of our all-too-human philosophical intuitions. One response x-phi can, and should, have to this situation is to clean up that floor—to tidy up and help take out the garbage that cannot be disposed of from the armchair. But I am urging here that we need also to pursue another line of response: learning how we can stand comfortably amid the mess, by means of a better understanding of it. Where we cannot learn how to get rid of the mess, we must learn how to find a secure footing within it. And to do so (and to leave the metaphor of floor



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

31

coverings behind us) may require a more radical reimagining of philosophical practice, especially in its inferences, than even experimental philosophers have heretofore imagined.21

Notes   1 See Huemer (2008) and Talbot (2014) for applications of this principle in the context of our armchair intuitional practices.   2 This is what I call in my (2015) the “error profile” of the instrument.   3 See also Gibson et al. (2013), both for greater mathematical rigor and for an application of similar reasoning in the context of methodology in linguistics.   4 I don’t mean anything too particular by “evidence-driven” here, but I include the restriction to indicate that there may be very different sorts of philosophical methodologies for which this analysis may just not fit at all, e.g. Foucauldian genealogy—without meaning to be at all derogatory towards those methodologies.   5 Though one should note that he is an influential member of the minority that abjures the use of the term “intuitions,” he is still defending the methodological practices that are under discussion here.   6 E.g. Deutsch (2010, 2015); Cappelen (2012); Earlenbaugh and Molyneaux (2009).   7 Such as Williamson, as noted above in n5. Ichikawa (2014) might be another example.   8 Thanks to Wesley Buckwalter for this useful piece of terminology. Also, see Nado (2015) for further examination of the methodological consequences of the exacting demands made by our preferred forms of philosophical theories.   9 Although there has been some important dissent from any sort of conceptualism, no other vindicating metatheory has emerged with anything near the popularity of that view. I believe this is in part because it has proved hard to justify these other elements of the practice without this further commitment. 10 Weinberg (forthcoming-a). 11 E.g. Gaut (2005). 12 See, e.g. Williams (1996). 13 See Lerner and Leslie (2013). 14 Nagel et al. (2013). 15 See, e.g. Paul (2012), who is also very explicit in her defense of metaphysical theorizing as a kind of modeling; Nolan (2015).

32

Advances in Experimental Philosophy and Philosophical Methodology

16 My discussion here is thus on the whole, pace Sober (2009). In general, we should be careful here to distinguish parsimony considerations from simplicity considerations, and here I am only intending to traffic in the latter. 17 Stanovich and West (2007). 18 Burnham and Anderson (2002) is a canonical reference here. 19 For a more extensive treatment of this example, see my (forthcoming-b). 20 See also Blouw et al. (forthcoming) for an attempt to impose some order on this messiness. 21 I am grateful to Jennifer Nado, Michael Johnson, and Ron Mallon for very useful comments on an earlier draft.

References Bealer, G. (1998), “Intuition and the Autonomy of Philosophy,” in M. DePaul and W. Ramsey (eds), Rethinking Intuition, Lanham: Rowman & Littlefield, 201–39. Bem, D. (2011), “Feeling the Future: Experimental Evidence for Anomalous Retroactive Influences on Cognition and Affect,” Journal of Personality and Social Psychology 100 (3): 407. Blouw, P., Buckwalter, W., and Turri, J. (forthcoming), “Gettier Cases: A Taxonomy,” in R. Borges, C. de Almeida and P. Klein (eds), Explaining Knowledge: New Essays on the Gettier Problem, Oxford: Oxford University Press. BonJour, L. (1998), In Defense of Pure Reason, Cambridge: Cambridge University Press. Boyd, K. and Nagel, J. (2014), “The Reliability of Epistemic Intuitions,” in E. Machery and E. O’Neill (eds), Current Controversies in Experimental Philosophy, New York: Routledge, 109–27. Burnham, K. and Anderson, D. (2002), Model Selection and Multimodel Inference: A Practical Information-theoretic Approach, Springer Science & Business Media. Cappelen, H. (2012), Philosophy Without Intuitions, Oxford: Oxford University Press. Deutsch, M. (2010), “Intuitions, Counter-examples, and Experimental Philosophy,” Review of Philosophy and Psychology 1 (3): 447–60. Deutsch, M. (2015), The Myth of the Intuitive: Experimental Philosophy and Philosophical Method, Cambridge, MA: MIT Press. Earlenbaugh, J. and Molyneux, B. (2009), “Intuitions are Inclinations to Believe,” Philosophical studies 145 (1): 89–109. Gaut, B. (2005), “The Cluster Account of Art Defended,” The British Journal of Aesthetics 45 (3): 273–88.



Experimental Philosophy, Noisy Intuitions, and Messy Inferences

33

Gibson, E., Piantadosi, S., and Fedorenko, E. (2013), “Quantitative Methods in Syntax/semantics Research: A Response to Sprouse and Almeida (2013),” Language and Cognitive Processes 28 (3): 229–40. Goldman, A. (2010), “Philosophical Naturalism and Intuitional Methodology,” Proceedings and addresses of the American Philosophical Association, American Philosophical Association: 115–50. Huemer, M. (2008), “Revisionary Intuitionism,” Social Philosophy and Policy 25 (01): 368–92. Ichikawa, J. (2014), “Who Needs Intuitions?,” in A. Booth and D. Rowbottom (eds), Intuitions, Oxford University Press, 232–56. Lerner, A. and Leslie, S. (2013), “Generics, Generalism, and Reflective Equilibrium: Implications for Moral Theorizing from the Study of Language,” Philosophical Perspectives 27: 366–403. Nado, J. (2015), “Intuition, Philosophical Theorising, and the Threat of Scepticism,” in E. Fischer and J. Collins (eds), Experimental Philosophy, Rationalism, and Naturalism: Rethinking Philosophical Method, Abingdon: Routledge. Nagel, J. (2012), “Intuitions and Experiments: A Defense of the Case Method in Epistemology,” Philosophy and Phenomenological Research 85 (3): 495–527. Nagel, J., San Juan, V., and Mar, R. A. (2013), “Lay Denial of Knowledge for Justified True Beliefs,” Cognition 129 (3): 652–61. Nolan, D. (2015), “The A Posteriori Armchair,” Australasian Journal of Philosophy, 93(2): 211–31. Paul, L. (2012), “Metaphysics as Modeling: The Handmaiden’s Tale,” Philosophical Studies 160: 1–29. Sober, E. (2009), “Parsimony Arguments in Science and Philosophy—A Test Case for Naturalism,” In Proceedings and Addresses of the American Philosophical Association, American Philosophical Association: 117–55. Stanovich, K. E. and West, R. F. (2007), “Natural Myside Bias is Independent of Cognitive Ability,” Thinking & Reasoning 13 (3): 225–47. Talbot, B. (2014), “Why So Negative? Evidence Aggregation and Armchair Philosophy,” Synthese 191 (16): 3865–96. Wagenmakers, E., Wetzels, R., Borsboom, D., and van der Maas, H. (2011), “Why Psychologists Must Change the Way They Analyze Their Data: The Case of Psi,” Journal of Personality and Social Psychology 100 (3): 426–32. Weatherson, B. (2003), “What Good Are Counterexamples?,” Philosophical Studies 115 (1): 1–31. Weinberg, J. (2015), “Humans as Instruments: Or, the Inevitability of Experimental Philosophy,” in E. Fischer and J. Collins (eds), Experimental Philosophy,

34

Advances in Experimental Philosophy and Philosophical Methodology

Rationalism, and Naturalism: Rethinking Philosophical Method, Routledge, 171–87. Weinberg, J. (forthcoming-a), “Going Positive by Going Negative: On Keeping X-phi Relevant and Dangerous,” in W. Buckwalter and J. Sytsma (eds), The Blackwell Companion to Experimental Philosophy, Oxford: Wiley-Blackwell. Weinberg, J. (forthcoming-b), “Knowledge, Noise, and Curve-fitting: A Methodological Case For K=JTB?,” in R. Borges, C. de Almeida, and P. Klein (eds), Explaining Knowledge: New Essays on the Gettier Problem, Oxford: Oxford University Press. Williams, M. (1996), Unnatural Doubts, Princeton, NJ: Princeton University Press. Williamson, T. (2008), The Philosophy of Philosophy, Oxford: Blackwell.

3

How to Do Better: Toward Normalizing Experimentation in Epistemology John Turri

1. Introduction Appeals to ordinary thought and talk are frequent in philosophy. For example, Aristotle claimed that a theory of happiness should be evaluated in light of “what is commonly said about it” (Aristotle 350 bce, 1941: 9–11). He judged it improbable that a widely held view would be completely incorrect (1941: 28–9). John Locke claimed that a theory of knowledge should be informed by how we ordinarily talk about knowledge (Locke 1690/1975, book 4.11.3–8). Thomas Reid claimed, “Philosophy has no other root but the principles of Common Sense,” and that in a contest between “Common Sense and Philosophy, the latter will always come off both with dishonour and loss” (Reid 1764/1997: 19). More recently, J. L. Austin advised that “ordinary language” should get “the first word” in philosophical theorizing (Austin 1956: 11). And Wilfrid Sellars argued that identifying the defining features of ordinary thought—“the manifest image”—is “a task of the first importance” for philosophers (Sellars 1963: Ch. 1). Other leading philosophers, including G. E. Moore and Roderick Chisholm, held common sense and ordinary language in similarly high regard (Lemos 2004). Many lines of research in contemporary epistemology are based on assumptions about ordinary thought and talk about knowledge—that is, about alleged central tendencies in commonsense epistemology. In the next section, I review five case studies where epistemologists have seriously mischaracterized commonsense epistemology. In the conclusion, I consider some implications of these findings. Epistemology, as currently practiced, fails in its ambitions

36

Advances in Experimental Philosophy and Philosophical Methodology

and might be alienating people from the field. Consequently, standard practice in the field should change. The good news is that the tools for change are already at the field’s disposal.

2. Case studies 2.1 Knowledge and reliability Contemporary philosophers nearly unanimously agree that knowledge requires reliability (for references, see Turri 2015c). On this view, knowledge must be produced by abilities that “will or would yield mostly true beliefs” (Alston 1995: 8–9). An enormous amount of work has been based on the assumption that knowledge requires reliability (for recent reviews, see Goldman 2008; Becker 2009). But there are theoretical reasons to expect that knowledge does not require reliability (Turri 2015c; Turri 2015d), and the literature contains surprisingly little explicit argumentation for reliabilism about knowledge (hereafter just “reliabilism”). Instead, reliabilists typically appeal to commonsense epistemology to support reliabilism. They claim that reliabilism best fits the patterns in “ordinary, intuitive judgments” about knowledge (Dretske 1981: 92; see also 249: n.8), that reliabilism is just “commonsense” (Sosa 1991: 100), and that our “inclinations” and “intuitions” about knowledge are keyed to reliability (Goldman 1993: 271). This “proto-reliabilist hypothesis,” as we might call it, has testable predictions. First and foremost, people will not count unreliably produced belief as knowledge. For example, if an ability produces only 10 percent or 30 percent true beliefs, then the beliefs it produces will not be judged knowledge. Second, and closely related, vast and explicit differences in reliability will produce large differences in knowledge judgments. It is easy to test these predictions: we just need to elicit judgments about simple cases that vary how reliably beliefs are formed. When this was done, the results were grim for the proto-reliabilist hypothesis (Turri under review a). Commonsense fully embraces the possibility of unreliable knowledge: rates of unreliable knowledge attribution sometimes reach 90 percent. Knowledge judgments are surprisingly insensitive to information about reliability: people



How to Do Better: Toward Normalizing Experimentation in Epistemology

37

attribute knowledge at similar rates whether the agent gets it right 90 percent of the time or 10 percent of the time. For example, in one study, participants were divided into two groups. They all read a story about Alvin, who has been out running errands all day. In the morning, Alvin’s wife told him that he should stop at the dry cleaners. On his way home, Alvin stops at the dry cleaners. Participants in the unreliable condition were also told that Alvin’s memory is very “unreliable”; participants in the reliable condition were told that Alvin’s memory is very “reliable.” That was the only difference. Participants were then asked whether Alvin knows that he should stop at the dry cleaners. In the unreliable condition, participants categorized Alvin as unreliable and 86 percent of them attributed knowledge; in the reliable condition, participants categorized Alvin as reliable and eighty-eight percent of them attributed knowledge (Turri under review a, Experiment 3). Similar results were observed for different cognitive faculties, ways of describing unreliability, and narrative contexts.

2.2 Knowledge and context Epistemic contextualism is the view that the verb “know” is a context sensitive expression. More specifically, according to contextualism, in order for us to truthfully say a person “knows” a proposition, that person must meet the evidential standard set by our context; and, critically, the standard changes across contexts. In support of their view, contextualists claim that it best explains competent speakers’ linguistic behavior in certain situations (Lewis 1996; DeRose 2009; Cohen 2013). That is, contextualism has its “basis in ordinary language,” where we allegedly find “evidence of the very best type” that the standards of “knowledge” are context sensitive (DeRose 2009: 48). Contextualism is motivated on the grounds that it is uniquely positioned to explain “evident facts about usual and … appropriate linguistic behavior” (DeRose 2009: 81). The most famous way of illustrating the idea involves a pair of cases about a man who wants to deposit a check and is deciding whether to wait in a long line at the bank on a Friday afternoon, or come back on Saturday morning when the line would be short (DeRose 2009). But then the question arises: is this bank actually open tomorrow (Saturday) morning? The man visited this

38

Advances in Experimental Philosophy and Philosophical Methodology

bank two Saturdays ago and it was open then, but banks do sometimes change their hours. In the “low stakes” version of the case, nothing serious hinges on whether he deposits the check before the weekend is over, and the man says, “I know that the bank is open tomorrow.” In the “high stakes” version of the case, something very serious hinges on whether he deposits the check before the weekend is over, and the man says, “I don’t know that the bank is open tomorrow.” Contextualists claim competent speakers will judge that the man truthfully says he “knows” in the low stakes version, and that the man truthfully says he “doesn’t know” in the high stakes version. It is an empirical question whether people behave as contextualists predict. Prior work has explored this empirical question, with mixed results. Some results seemed inconsistent with contextualist predictions (e.g. Buckwalter 2010; Feltz and Zarpentine 2010; Buckwalter 2014; see also May et al. 2010), but others were consistent with the predictions (e.g. Hansen and Chemla 2013; see also Alexander et al. 2014). Contextualists have made several methodological objections to previous studies, particularly emphasizing the importance of eliciting metalinguistic judgments about a knowledge statement’s truth value, and of having the agent say “I know” in the low case and “I don’t know” in the high case (DeRose 2009: 49 n. 2; DeRose 2011). It turns out that, when the cases are tested while respecting those methodological strictures, the basic contextualist prediction turns out true: people tend to agree that the agent who says “I know” in the low stakes case speaks truthfully, and they tend to agree that the agent who says “I don’t know” in the high stakes case speaks truthfully (Turri under review b, Experiment 1). Is this result good evidence that “know” is a context-sensitive expression? No, it is not good evidence, because simpler alternative explanations immediately suggest themselves. On the one hand, across the pair of cases, people might draw different inferences about whether the agent satisfies requirements of knowledge that contextualists and non-contextualists alike accept, such as truth, belief, and evidence. For example, people might assume that someone who says “I don’t know” is less likely to have the relevant belief, or to have a true belief, or to have good evidence, than someone who says “I do know.” This could lead people to judge that the agent knows in the one case but not in the other, which in turn would lead people to judge that the



How to Do Better: Toward Normalizing Experimentation in Epistemology

39

one agent truthfully says “I know,” whereas the other agent truthfully says “I don’t know.” Contextualists do not accept these alternative explanations. They claim that it is “natural” to interpret the agent as equally confident in the low and high stakes cases (DeRose 2009: 190–3). The claim that “nothing changes” about the quality of the agent’s evidence across the cases (Lawlor 2013: 81; see also DeRose 2009: 2). And some imply that we can or will “assume” that the relevant proposition is true in each case (DeRose 2009: 2). Such interpretations might seem natural to contextualists, but it is an empirical question whether the cases are actually interpreted that way. And it is not hard to test the matter: we just need to (i) collect judgments about belief, truth, and evidence in the same context as the metalinguistic “knowledge” judgments, and (ii) see which judgments differ between low and high conditions. When this was tested, people judged that the low stakes agent is less likely to have the relevant belief, less likely to have a true belief, and less likely to have good evidence than the high stakes agent (Turri under review b, Experiment 1). Any of or all of these differences could produce the observed pattern of metalinguistic judgments about “knowledge,” even if contextualism about “knows” is false. Similar results were observed for another famous pair of low/high cases proposed by contextualists (Cohen 1999). On the other hand, the low/high cases contain a critical confound. Contextualists note that it is important for the agent in the low stakes case to say “I know,” and for the agent in the high stakes case say “I don’t know.” As a result, the low/high difference is confounded with the “I know/I don’t know” difference. Thus, an alternative, non-contextualist explanation of the finding is that people tend to defer to others’ self-regarding knowledge statements. That is, people might simply assume that others are well positioned to report on whether they know a proposition, which could produce the pattern. This assumption could lead them to defer to the agent’s self-regarding knowledge statement. If this deferral hypothesis is true, then we would expect the observed pattern of behavior even if contextualism is false. Testing the deferral hypothesis is not difficult. The basic idea is to divide people into two groups and have them read the exact same story, except for one small difference: have the agent say “I do know” in one story, and “I don’t know” in the other. Do not vary the stakes or anything else. When people were tested in this way, they exhibited the exact same pattern noted above: they

40

Advances in Experimental Philosophy and Philosophical Methodology

agreed when the person said “I do know,” and they agreed when the person said “I don’t know” (Turri under review b, Experiment 2). Similar results were observed for another famous pair of low/high cases proposed by contextualists (Cohen 1999). The critical point here is that the difference between “I do know” and “I don’t know” suffices, all by itself, to produce the pattern that contextualists predict. In other words, deferral is enough to produce the principal datum used to motivate contextualism. Varying the stakes is superfluous, as is the hypothesis that “know” is context-sensitive.

2.3 Knowledge and closure In its simplest form, the epistemic closure principle states that, necessarily, if you know that P, and you know that Q if P, then you also know that Q. Or, at least, you would then be in a position to easily know Q by simple inference. For instance, if you know that your car is parked in the parking structure, and you know that your car has not been stolen if it is parked in the structure, then you know that your car has not been stolen. The epistemic closure principle is implicated in many debates in contemporary epistemology, including debates about skepticism (Stroud 1984), the semantics of knowledge attributions (DeRose 1995; Cohen 1999), and the relationship between knowledge and practical considerations (Hawthorne 2004). The epistemic closure principle is not as widely or uncritically accepted as, say, knowledge reliabilism, but it is very widely accepted. And, as with knowledge reliabilism, the scarcity and weakness of explicit arguments for the epistemic closure principle is surprising (Turri 2015a: § 1.2). In defense of the principle, many claim that it is a defining feature of commonsense epistemology. Epistemic closure, they claim, is “just a familiar fact about human knowledge, something we all recognize and abide by in our thought and talk about knowing things” (Stroud 1984: 18–19). The principle deserves to be called “axiomatic” (Cohen 1999: 69). Theoretical proposals to abandon closure are “revisionary” and suffer from “deep and wide-ranging” “conflict” with our “intuition” (Hawthorne 2004: 36). It is, we are told, “striking” (Hawthorne 2004: 36) and “startling” (Fumerton 1987) that “some philosophers have gone so far as” to deny the principle and, in the process,



How to Do Better: Toward Normalizing Experimentation in Epistemology

41

“blatantly violate” such a “basic and extremely plausible intuition” (Steup 2005). For a long time, this extravagant characterization of what “we” find intuitively compelling went unchallenged, even by closure’s opponents (e.g. Nozick 1981: 205–6; Dretske 2013: 32). But no serious evidence was ever produced for the claim that closure is a defining feature of commonsense epistemology. And it is a straightforwardly empirical question whether commonsense exhibits such patterns. Testing for a commitment to the epistemic closure principle is not difficult. The basic idea is to have people consider simple scenarios and judge whether the agent knows a proposition and one of its obvious, trivial, implications. For example, consider this simple scenario (inspired by Vogel 1990: § 2): When Maxwell arrives at work in the morning, he always parks in one of two spots: C8 or D8. Half the time he parks in C8, and half the time he parks in D8. Today Maxwell parked in C8. It’s lunchtime at work. Maxwell and his assistant are up in the archives room searching for a particular document. Maxwell says, “I might have left the document in my car.” The assistant asks, “Mr. Maxwell, is your car parked in space C8? It’s not unheard of for cars to be stolen.” Maxwell thinks carefully for a moment and then responds, “No, my car has not been stolen. It is parked in C8.”

Which of the following options best describes Maxwell? 1 He knows that his car is parked in C8. And he knows that his car has not been stolen. 2 He does not know that his car is parked in C8. But he does know that his car has not been stolen. 3 He knows that his car is parked in C8. But he does not know that his car has not been stolen. 4 He does not know that his car is parked in C8. And he does not know that his car has not been stolen. If the epistemic closure principle is a defining feature of commonsense epistemology, then, when people consider which option best describes Maxwell, the intuitively best answer will not be option 3. Indeed, many philosophers claim that statements like 3 are “abominable,” “ridiculous,” and “distinctly repugnant” (DeRose 1995: 27–9; Dretske 2013: 31; Sosa 2004: 41). However,

42

Advances in Experimental Philosophy and Philosophical Methodology

when this case was actually tested, nearly two-thirds of people selected option 3 as the best option (Turri 2015a: Experiment 3). That is, the strong central tendency was to attribute the only combination of knowledge and ignorance inconsistent with the epistemic closure principle. Nearly no one selected option 3 as best in a closely matched control condition. This same basic pattern persisted across different questioning procedures and narrative contexts (Turri 2015a: Experiments 1, 2, 4, 5; Turri 2015b).

2.4 Justification and truth Suppose that Maria is a watch collector who owns over 10,000 watches. She cannot keep track of all her watches by memory alone, so her accountant maintains a detailed inventory of them. Maria knows that the inventory isn’t perfect, but it is extremely accurate. Someone asks Maria, “Does your collection contain a 1990 Rolex?” Maria consults the inventory and it says that she does have one. And this is just another occasion where the inventory is right. Does Maria’s evidence justify her in believing that the collection contains a 1990 Rolex? Now imagine a case exactly like the one just described except that it is Mario who collects watches and maintains an inventory, and this is one of those rare cases where the inventory is wrong. Does Mario’s evidence justify him in believing that the collection contains a 1990 Rolex? Can the fact that the inventory is right in the one case but not the other create a difference in what Maria and Mario are justified in believing? According to the dominant view in contemporary epistemology—the “truth-insensitive” approach to belief evaluation—the answer is a resounding “no.” On this view, Maria and Mario are equally justified because they have “exactly the same reasons for believing exactly the same thing” (Feldman 2003: 29; see also Conee and Feldman 2004). Their evidence and justification are equivalent across the cases, despite the variation in truth. Focusing on more extravagant examples, consider a staple of contemporary epistemology, the notorious thought experiment involving a radically deceived “brain-in-a-vat.” Leading epistemologists tell us that it is “beyond question at an intuitive level” that such victims are “justified” in their beliefs about their surroundings, just as we are (BonJour and Sosa 2003: 185–6). “The intuition” here, we are told, is that you and your victimized counterpart in the super-psychologist’s tank



How to Do Better: Toward Normalizing Experimentation in Epistemology

43

have “equally good evidence” for your respective beliefs (Russell 2001: 38; see also Cohen 1984: 281–2). We have an “overpowering inclination to think” that the two of you have equally good evidence (Fumerton, 2006: 93). At least one philosopher has claimed that truth profoundly affects what an agent is justified in believing (Sutton 2007: 19; see also Littlejohn 2013). According to this view, “there are no false justified beliefs” because justification requires knowledge, and knowledge requires truth (Sutton 2007: 1). This view has been labeled an “extraordinary” and “dissident doctrine” lacking in “intuitive credentials” (Conee 2007). It deviates from “the usual” and “traditional” approach to questions about belief evaluation, according to which there is no connection between justification and truth (Chisholm 1989: 76). But is it so extraordinary that truth can directly affect whether a belief is justified? Is it revolutionary to propose that false beliefs are not justified or should not be held? Some philosophers propose that such views are exactly the opposite of extraordinary or revisionary. The extraordinary and revisionist view, they suggest, is the one cultivated in narrow circles within “the professional philosophical community” (Sutton 2007: vii). So which view is “intuitive” and “natural,” and which is “extraordinary” and “dissident”? Is truth-sensitive epistemology revisionary, or does it accurately reflect our ordinary evaluative practices? It is easy to test the matter: we just need to elicit judgments about simple cases that vary only the truth-value of the relevant proposition. When this was done, the results showed that ordinary evaluations of belief are deeply sensitive to truth. When confronted with cases very similar to the ones about Maria and Mario above, approximately 75 percent of people judged that the true belief was justified, but only approximately 25 percent of people judged that the false belief was justified. The same pattern was observed across different narrative contexts and when using different normative vocabulary to evaluate belief and evidence (e.g. “rational,” “reasonable,” “Maria should believe,” “the responsible thing for Maria to believe is,” “Maria has good evidence to believe,” and “the evidence justifies Maria in believing”) (Turri in press d: Experiment 1; see also Turri in press c: Experiments 1 and 4). Similarly, in a study that asked people to compare the evidence of a normally embodied human and the evidence of his victimized “brain-in-a-vat” twin,

44

Advances in Experimental Philosophy and Philosophical Methodology

nearly eighty percent of people judged that the normal human’s evidence was better (Turri in press d, Experiment 3).

2.5 Knowledge and near-misses For decades epistemologists have claimed that, intuitively, agents in “fake barn” cases lack knowledge (e.g. Goldman 1976; Sosa 1991: 238–9; Plantinga 1993: 33; Pritchard 2005: 161–2; Kvanvig 2008: 274; Pritchard 2014: Ch. 6). In a fake barn case, the agent (“Henry” in the original) perceives an object (a barn in the original) and this perceptual relation remains intact throughout. If that is where the story ended, then we would have no problem saying that the agent knows that the object is present. However, it is then revealed, the agent currently inhabits an environment filled with numerous fakes (“papier-mâché facsimiles of barns” in the original), which look like real ones. Epistemologists tell us that, when this information is revealed, we are “strongly inclined” to deny that the agent knows that the object is present. We are then asked, “How is this change in our assessment to be explained?” (Goldman 1976: 773). Any attempt to explain this behavior is idle if the behavior does not occur. And recent research has shown that in fact the behavior does not occur. Instead, people tend to attribute knowledge in fake-barn cases (Colaco et al. 2014; Turri under review c; Turri under review d). More generally, in cases where an agent perceives something that makes her belief true and this perceptual relation remains intact, people overwhelmingly attribute knowledge, even when lookalikes pose a salient threat to the truth of the agent’s belief (Turri et al. 2014). Indeed, commonsense views these cases similarly to paradigm instances of knowledge, with rates of knowledge attribution often exceeding eighty percent and not differing from closely matched controls.

3. Conclusion Appeals to commonsense and ordinary thought and talk are frequent in philosophy, perhaps nowhere more so than in contemporary epistemology.



How to Do Better: Toward Normalizing Experimentation in Epistemology

45

When an epistemological theory implies serious error in commonsense epistemology, it is counted as a cost of the view. It is not necessarily a prohibitive cost, but it is a cost nonetheless. Similarly, when an epistemological theory respects or vindicates deep patterns in commonsense epistemology, it is viewed as a benefit of the view. It is not necessarily a conclusive benefit, but it is a benefit nonetheless. Standard practice in Anglo–American “analytic” epistemology is to rely on introspection and anecdotal social observation to characterize commonsense epistemology. This method has led to serious mischaracterizations of commonsense and ordinary practice. Above, I reviewed five examples of this: knowledge reliabilism, contextualism, the epistemic closure principle, truthinsensitive theories of justification, and knowledge attributions in “fake barn” cases. In each case, the central tendencies in commonsense epistemology are very different from what epistemologists have assumed. The implications of this are potentially significant. I will discuss two. One implication pertains to how people perceive philosophy. Recently there have been public discussions, in The New York Times (Fish 2011) and The Chronicle of Higher Education (McIntyre 2011), of whether academic philosophy has any value at all. Extremely successful senior philosophers question the value of philosophy practiced from the armchair (Glymour 2011; Thagard 2012). Many factors probably contribute to negative perceptions of contemporary academic philosophy. One factor is that people are suspicious of attempts to answer philosophical questions from the armchair (Buckwalter and Turri under review). Indeed, greater prior exposure to academic philosophy is associated with more suspicion of armchair philosophical inquiry (Buckwalter and Turri under review: Experiment 2). A second factor might be that people are alienated by the verdicts that philosophers treat as obvious or commonsensical. For instance, imagine students taking an introductory epistemology course. They are asked to consider a radically deceived “brain-in-a-vat” undergoing experiences qualitatively similar to those of a normal human. “The intuition here,” they are told, is that the brain-in-a-vat and the normal human are equally justified in all of their beliefs. This intuition contradicts most people’s judgment about the case, but it is treated as a touchstone for constructing and evaluating theories of justification. The same is true for “the intuition” that the agent in a “fake barn”

46

Advances in Experimental Philosophy and Philosophical Methodology

case does not have knowledge. Before long, some students might conclude that they are not good at philosophy (Buckwalter and Stich 2014: § 4). But, I suspect, it is equally likely that students will conclude, along with Thomas Reid, that, if this is how philosophers conduct their business, then philosophy is “justly ridiculous” (Reid 1764/1997: 21). Another implication is methodological. To the extent that epistemologists aim to understand the concepts and categories we care about in our ordinary lives—the things which led many of us to study philosophy in the first place— then standard practice in contemporary epistemology should change. As the case studies reviewed above suggest, it is inadequate for epistemologists to rely only on introspection of their own impressions, the reported impressions of other professionals, and anecdotal social observation. A few simple controlled experiments can quickly cut through mistaken impressions and observations. In short, epistemologists should be quick to rely on controlled experiments to separate wheat from chaff. This is especially true early in a research program, when results can provide formative clues putting researchers on more promising paths. For example, an enormous amount of time and energy has been devoted to knowledge reliabilism and contextualism over the past few decades. In that time, these views went from novelties to standard topics covered in textbooks, handbooks, anthologies, encyclopedias, and survey courses. Epistemologists lengthily debated finer details of how to understand reliability and the context-sensitivity of knowledge judgments, as well as their relationship to other complicated issues in philosophy. But none of these details matters if the assumptions initiating the discussion are false. Moving forward, some basic preliminary experiments can help to avoid similar false starts. Would that a false start last a few weeks rather than a few decades! Philosophers are typically not trained to conduct experiments, so this methodological proposal might seem onerous and impractical. But this concern is addressed by two developments. On the one hand, a critical mass of philosophers today can conduct experiments—the experimental philosophers. The number of philosophers fitting this description grows each year. On the other hand, interdisciplinary collaboration is increasingly common, and there have always been trained scientists interested in philosophical issues and willing to work with philosophers on questions of common concern.



How to Do Better: Toward Normalizing Experimentation in Epistemology

47

Nothing I say here implies the illegitimacy of prescriptive epistemology. Epistemologists need not be limited to mapping the contours of commonsense epistemology. Instead, they might wish to cultivate new and better epistemological concepts and categories. But this cannot sensibly be done in ignorance of commonsense concepts and categories. Indeed, quite the opposite—it is silly to prescribe changes unless you understand what you are prescribing changes to! Such understanding is required to intelligently evaluate whether your proposed changes would improve people’s performance. In this respect, experimentation is a prerequisite to good prescriptive epistemology. Of course, not every epistemological project need be experimental and not every epistemologist need be an experimentalist. That would be boring and dysfunctional. It is enough that the field as a whole includes the relevant experimental expertise and brings it to bear in appropriate ways at appropriate junctures. As happens in other fields, researchers will choose a path that suits their interests and temperament—some will focus on more purely theoretical or prescriptive projects, others will focus on more descriptive or experimental projects, others will focus on applied projects, and others will pursue an eclectic mix. The sooner epistemology evolves in this direction, the sooner it will become a more productive, exciting, and respected discipline. Of course, this will probably not happen unless practitioners understand and embrace the value of doing so. In sum, in light of a string of serious and widespread failures in the field, normalizing experimentation in epistemology is wise. It will help the field avoid unfavorable perceptions, and help practitioners identify genuinely commonsensical starting points.

Acknowledgments For helpful comments and feedback, I thank Wesley Buckwalter, Jennifer Nado, and Angelo Turri. This research was supported by the Social Sciences and Humanities Research Council of Canada and an Early Researcher Award from the Ontario Ministry of Economic Development and Innovation.

48

Advances in Experimental Philosophy and Philosophical Methodology

References Alexander, J., Gonnerman, C., and Waterman, J. (2014), “Salience and Epistemic Egocentrism: An Empirical Study,” in J. R. Beebe (ed.), Advances in Experimental Epistemology, London: Bloomsbury Academic, 97–117. Alston, W. P. (1995), “How To Think About Reliability,” Philosophical Topics 23 (1): 1–29. Aristotle (350 bce/1941), Nichomachean Ethics. W. D. Ross, trans., R. McKeon, ed. Austin, J. L. (1956), “A Plea for Excuses,” Proceedings of the Aristotelian Society 57: 1–30. BonJour, L. and Sosa, E. (2003), Epistemic Justification: Internalism vs. Externalism, Foundations vs. Virtues, Malden, MA: Blackwell. Buckwalter, W. (2010), “Knowledge isn’t Closed on Saturday: A Study in Ordinary Language,” Review of Philosophy and Psychology 1 (3): 395–406. doi:10.1007/ s13164-010-0030 Buckwalter, W. (2014), “The Mystery of Stakes and Error in Ascriber Intuitions,” in J. R. Beebe (ed.), Advances in Experimental Epistemology, London: Bloomsbury Academic, 145–73. Buckwalter, W. and Stich, S. (2014), “Gender and Philosophical Intuition,” in J. Knobe and S. Nichols (eds), Experimental Philosophy (2nd edn), Oxford: Oxford University Press. Buckwalter, W. and Turri, J. (under review), “Perceived Weaknesses of Philosophical Inquiry: A Comparison to Psychology,” Ontario: University of Waterloo. Chisholm, R. (1989), Theory of Knowledge (3rd edn), Englewood Cliffs, NJ: Prentice Hall. Cohen, S. (1984), “Justification and Truth,” Philosophical Studies 46 (3): 279–95. doi:10.1007/BF00372907 Cohen, S. (1999), “Contextualism, Skepticism, and the Structure of Reasons,” Philosophical Perspectives 13: 57–89. Cohen, S. (2013), “Contextualism Defended,” in M. Steup, J. Turri and E. Sosa (eds), Contemporary Debates in Epistemology (2nd edn), Malden, MA: Wiley-Blackwell, 69–75. Colaco, D., Buckwalter, W., Stich, S., and Machery, E. (2014), “Epistemic Intuitions in Fake-barn Thought Experiments,” Episteme 11 (02): 199–212. doi:10.1017/ epi.2014.7 Conee, E. (2007), “Review of Jonathan Sutton, Without Justification,” Notre Dame Philosophical Reviews, 12. Available from http://ndpr.nd.edu/review.cfm?id=11803 [accessed March 31, 2015].



How to Do Better: Toward Normalizing Experimentation in Epistemology

49

Conee, E. and Feldman, R. (2004), Evidentialism: Essays in Epistemology, Oxford: Oxford University Press. DeRose, K. (1995), “Solving the Skeptical Problem,” The Philosophical Review 104 (1): 1–52. DeRose, K. (2009), The Case for Contextualism, Oxford: Oxford University Press. DeRose, K. (2011), “Contextualism, Contrastivism, and X-Phi Surveys,” Philosophical Studies 156 (1): 81–110. doi:10.1007/s11098-011-9799-x Dretske, F. I. (1981), Knowledge and the Flow of Information, Cambridge, MA: MIT Press. Dretske, F. I. (2013), “The Case against Closure,” in M. Steup, J. Turri, and E. Sosa (eds), Contemporary Debates in Epistemology (2nd edn), Malden, MA: Wiley-Blackwell. Feldman, R. (2003), Epistemology, Upper Saddle River, NJ: Prentice Hall. Feltz, A. and Zarpentine, C. (2010), “Do You Know More When It Matters Less?,” Philosophical Psychology 23 (5): 683–706. doi:10.1080/09515089.2010.514572 Fish, S. (2011), “Does Philosophy Matter?,” New York Times. Available from http:// opinionator.blogs.nytimes.com/2011/08/01/does-philosophy-matter [accessed March 31, 2015]. Fumerton, R. (1987), “Nozick’s Epistemology,” in S. Luper-Foy (ed.), The Possibility of Knowledge, Lanham, MD: Rowman & Littlefield. Fumerton, R. (2006), Epistemology, Malden, MA: Blackwell. Glymour, C. (2011), “Manifesto,” Choiceandinference.com. Available from http:// choiceandinference.com/2011/12/23/in-light-of-some-recent-discussion-over-atnew-apps-i-bring-you-clark-glymours-manifesto [accessed March 31, 2015]. Goldman, A. I. (1976), “Discrimination and Perceptual Knowledge,” Journal of Philosophy 73 (20): 771–91. Goldman, A. I. (1993), “Epistemic Folkways and Scientific Epistemology,” Philosophical Issues 3: 271–85. Hansen, N. and Chemla, E. (2013), “Experimenting on Contextualism,” Mind & Language 28 (3): 286–321. doi:10.1111/mila.12019 Hawthorne, J. (2004), Knowledge and Lotteries, Oxford: Oxford University Press. Kvanvig, J. (2008), “Epistemic Luck,” Philosophy and Phenomenological Research 77 (1): 272–81. doi:10.1111/j.1933-1592.2008.00187.x Lawlor, K. (2013), Assurance: An Austinian View of Knowledge and Knowledge Claims, Oxford: Oxford University Press. Lemos, N. (2004), Common Sense: A Contemporary Defense, Cambridge: Cambridge University Press. Lewis, D. (1996), “Elusive Knowledge,” Australasian Journal of Philosophy 74 (4): 549–67.

50

Advances in Experimental Philosophy and Philosophical Methodology

Littlejohn, C. (2013), “The Russellian Retreat,” Proceedings of the Aristotelian Society 113 (3): 293–320. doi:10.1111/j.1467-9264.2013.00356.x Locke, J. (1690/1975), An Essay Concerning Human Understanding, P. H. Nidditch, ed., Oxford: Clarendon Press. May, J., Sinnott-Armstrong, W., Hull, J. G., and Zimmerman, A. (2010), “Practical Interests, Relevant Alternatives, and Knowledge Attributions: An Empirical Study,” Review of Philosophy and Psychology 1 (2): 265–73. doi:10.1007/ s13164-009-0014-3 McIntyre, L. (2011), “Making Philosophy Matter—or Else,” The Chronicle of Higher Education. Available from http://chronicle.com/article/Making-PhilosophyMatter-or/130029 [accessed March 31, 2015]. Nozick, R. (1981), Philosophical Explanations, Cambridge, MA: Harvard University Press. Plantinga, A. (1993), Warrant and Proper Function, Oxford: Oxford University Press. Pritchard, D. (2005), Epistemic Luck, New York: Oxford University Press. Pritchard, D. (2014), What is This Thing Called Knowledge? (3rd edn), New York: Routledge. Reid, T. (1764/1997), An Inquiry Into the Human Mind on the Principles of Common Sense, D. R. Brookes, ed., University Park, PN: Pennsylvania State University Press. Russell, B. (2001), “Epistemic and Moral Duty,” in M. Steup (ed.), Knowledge, Truth, and Duty: Essays on Epistemic Justification, Responsibility, and Virtue, Oxford: Oxford University Press, 34–48. Sellars, W. (1963), Science, Perception and Reality, Atascadero, CA: Ridgeview Publishing Company. Sosa, E. (1991), Knowledge in Perspective, Cambridge: Cambridge University Press. Sosa, E. (2004), “Relevant Alternatives, Contextualism Included,” Philosophical Studies 119 (1): 35–65. Steup, M. (2005), “Epistemology,” Stanford Encyclopedia of Philosophy. Available from http://plato.stanford.edu/entries/epistemology/ [accessed March 31, 2015]. Stroud, B. (1984), The Significance of Philosophical Skepticism, Oxford: Clarendon Press. Sutton, J. (2007), Without Justification, Cambridge, MA: MIT Press. Thagard, P. (2012), “Eleven Dogmas of Analytic Philosophy,” Psychologytoday.com. Available from http://www.psychologytoday.com/blog/hot-thought/201212/ eleven-dogmas-analytic-philosophy [accessed March 31, 2015]. Turri, J. (2015a), “An Open and Shut Case: Epistemic Closure in the Manifest Image,” Philosophers’ Imprint 15 (2): 1–18.



How to Do Better: Toward Normalizing Experimentation in Epistemology

51

Turri, J. (2015b), “Skeptical Appeal: The Source-Content Bias,” Cognitive Science 39 (2). doi:10.1111/cogs.12153 Turri, J. (2015c), “Unreliable Knowledge,” Philosophy and Phenomenological Research, 90(3): 529–45. Turri, J. (2015d), “From Virtue Epistemology to Abilism: Theoretical and Empirical Developments,” in C. B. Miller, M. R. Furr, A. Knobel, and W. Fleeson (eds), Character: New Directions from Philosophy, Psychology, and Theology, Oxford: Oxford University Press. Turri, J. (2015e), “Evidence of Factive Norms of Belief and Decision,” Synthese. Turri, J. (in press), “The Radicalism of Truth-Insensitive Epistemology: Truth’s Profound Effect on the Evaluation of Belief,” Philosophy and Phenomenological Research. Turri, J. (under review-a), “A New Paradigm for Epistemology: From Reliabilism to Abilism,” Ontario: University of Waterloo. Turri, J. (under review-b), “Epistemic Contextualism: An Idle Hypothesis,” Ontario: University of Waterloo. Turri, J. (under review-c), “Knowledge and Assertion in Gettier Cases,” Ontario: University of Waterloo. Turri, J. (under review-d), “Vision, Knowledge, and Assertion,” Ontario: University of Waterloo. Turri, J., Buckwalter, W., and Blouw, P. (2014), “Knowledge and Luck,” Psychonomic Bulletin & Review. doi:10.3758/s13423-014-0683-5 Vogel, J. (1990), “Are there Counterexamples to the Closure Principle?” in M. D. Roth and G. Ross (eds), Philosophical Studies Series (Vol. 48), 13–27, Springer. doi:10.1007/978-94-009-1942-6_2

4

Thought Experiments, Mental Modeling, and Experimental Philosophy Joshua Alexander

There has been a lot of debate about whether philosophical intuitions play a significant role in contemporary philosophy, so much so that this question has come to dominate discussions of philosophical methodology. According to the now orthodox view, philosophical intuitions not only form part of our “standard justificatory procedure” they are also part of what makes philosophical methodology unique (see, for example, Bealer 1998; Levin 2005; Goldman 2007). Herman Cappelen has recently challenged the orthodox view, arguing that there is no agreement about what philosophical intuitions are supposed to be and, more importantly, that philosophers simply do not need them anyway (Cappelen 2013). Max Deutsch has raised similar worries over the course of a number of recent articles, including his contribution to this volume, and joined Cappelen in using these worries to undermine the significance of experimental philosophy (Deutsch 2009, 2010). They are both right and wrong. They are right that philosophers have not been univocal about what philosophical intuitions are supposed to be, something that Timothy Williamson has also called to our attention, writing colorfully that their “main function is not to answer questions about the nature of evidence on offer but to fudge them, but appearing to provide answers without really doing so” (2007: 220). But they are wrong that judgments about cases play no significant role in philosophical practice, and therefore wrong about the philosophical significance of experimental philosophy, something I have argued for elsewhere (Alexander 2010, 2012). Rather than rehash these arguments, I would like to try a different approach here, one that moves past talking about philosophical intuitions

54

Advances in Experimental Philosophy and Philosophical Methodology

altogether, and instead involves thinking more carefully about how they are generated, whatever they might be. This approach involves thinking carefully about thought experiments, how they work, and why they sometimes go wrong. Thought experiments are our stock-in-trade, the way that we support or test the kinds of modal claims that are central to philosophical discourse and debate; and what I would like to do here is to examine what we can learn about how thought experiments are supposed to work in contemporary philosophy by looking at how thought experiments are supposed to work in science, focusing on two views about thought experiments that emerge from the philosophy of science—what are sometimes called the “argument view” and the “mental model view.” The key insights of these two views are that thought experiments are arguments, and that conducting thought experiments allows us to mobilize certain cognitive resources and bring those cognitive resources to bear on important questions. These insights are present in Williamson’s recent observation that thought experiments “turn out to be fairly straightforward modal arguments, typically valid ones, with counterfactual conditionals and possibility claims as premises,” and his conviction that we “assess the premises by the same offline methods as for other claims of this sort” (2009: 433). And, understanding these insights will help highlight the role that philosophical intuitions play in contemporary philosophy, underscore the importance of experimental philosophy, and even provide new insights into ways that thought experiments might fail.

1. The argument view John Norton has been the leading proponent of the argument view, according to which thought experiments are nothing more than colorful arguments (1991, 1996, 2004a, 2004b; see also Sorensen 1992 and Häggqvist 1996). Norton’s principal aim is to give a faithful account of how thought experiments are used in physics, and so he supports his view by reconstructing formal arguments that can be found in some of Einstein’s more famous thought experiments, as well as other famous thought experiments from the history of physics. He reduces the cases used in these thought experiments to sets of propositions, adds to these sets certain background assumptions,



Thought Experiments, Mental Modeling, and Experimental Philosophy

55

and demonstrates that these premises and assumptions lead to the conclusions they are meant to support through recognized sorts of inductive or deductive inference. On the basis of these examples, he claims, “the workings and achievements of any thought experiment can be revealed and captured fully in an explicit argument which employs the same resources” (1996: 339). To see how this is supposed to work, let’s take a look at one of his examples. One of the most famous thought experiments in physics involves Galileo’s attempt to refute the Aristotelian claim that heavier objects fall faster than lighter ones, or more generally that objects fall at speeds proportional to their weights. Galileo’s refutation comes in a dialogue between Salviati and Simplicio, and includes the following passage spoken by Salviati: Tell me, Simplicio, whether you assume that for every heavy falling body there is a speed determined by nature such that this cannot be increased or diminished except by using force or opposing some impediment to it… [Simplicio agrees.] Then if we had two moveables whose natural speeds were unequal, it is evident that were we to connect the slower to the faster, the latter would be partly retarded by the slower, and this would be partly speeded up by the faster … But if this is so, and if it is also true that a large stone is moved with eight degrees of speed, for example, and a smaller one with four [degrees], then joining both together, their composite will be moved with a speed less than eight degrees. But the two stones joined together make a larger stone than the first one which was moved with eight degrees of speed; therefore this great stone is moved less swiftly than the lesser one. But this is contrary to your assumption. So you see how, from the supposition that the heavier body is moved more swiftly than the less heavy, I conclude that the heavier move less swiftly. (Galileo 1954: 66–7)

Norton contends that this thought experiment is rather obviously a reductio argument, and provides something like following reconstruction: Suppose that objects fall at speeds proportional to their weights. This means that a heavier stone will fall faster than a lighter stone. It also means that connecting the two stones would cause the lighter stone to move faster and the faster stone to move slower, and this would mean that the composite stone will fall slower than the heavier stone would fall alone. But the composite is heavier than the heavier stone, and so should fall faster than the heavier stone would fall alone. But this is a contradiction, which means that our initial assumption is false: objects do not fall at speeds proportional to their weights.

56

Advances in Experimental Philosophy and Philosophical Methodology

The thrust of the reconstruction is this: thought experiments can be reconstructed as arguments, and what justification we have for believing their conclusions comes from the evidence these reconstructed arguments provide those conclusions. Timothy Williamson has advanced a similar view about the role that thought experiments play in contemporary philosophy, writing, “on our account, a thought experiment … embodies a straightforward valid modal argument for a modal conclusion” (2007: 187). Again, an example will help us to see how this is supposed to work. Quite possibly the most famous thought experiment in contemporary philosophy involves Edmund Gettier’s attempt to show that a person’s justified true belief need not count as knowledge. Gettier uses two cases involving Smith who has deduced a true belief that q from a justified false belief that p and, on that basis, formed a justified true belief that q. According to Gettier, despite now having a justified true belief that q, Smith does not know that q. Here is one of the relevant cases: Suppose that Smith and Jones have applied for a certain job. And suppose that Smith has strong evidence for the following conjunctive proposition: (d) Jones is the man who will get the job, and Jones has ten coins in his pocket. Smith’s evidence for (d) might be that the president of the company assured him that Jones would in the end be selected, and that he, Smith, had counted the coins in Jones’s pocket ten minutes ago. Proposition (d) entails: (e) The man who will get the job has ten coins in his pocket. Let us suppose that Smith sees the entailment from (d) to (e), and accepts (e) on the grounds of (d), for which he has strong evidence. In this case, Smith is clearly justified in believing that (e) is true. But imagine, further, that unknown to Smith, he himself, not Jones, will get the job. And, also, unknown to Smith, he himself has ten coins in his pocket. Proposition (e) is then true, though proposition (d), from which Smith inferred (e), is false. In our example, then, all of the following are true: (i) (e) is true, (ii) Smith believes that (e) is true, and (iii) Smith is justified in believing that (e) is true. But it is equally clear that Smith does not KNOW that (e) is true; for (e) is true in virtue of the number of coins in Smith’s pocket, while Smith does not know how many coins are in Smith’s pocket, and bases his belief in (e) on a count of the coins in Jones’s pocket, whom he falsely beliefs to be the man who will get the job. (Gettier 1963: 122)

Williamson contends that this thought experiment is a modal argument, and provides something like the following reconstruction:



Thought Experiments, Mental Modeling, and Experimental Philosophy

57

(1) It is possible for someone to stand in relation to some proposition p just as the protagonist of Gettier’s cases to stand to the relevant propositions. (2) If someone were to stand in relation to some proposition p just as the protagonists of Gettier’s cases stand to the relevant propositions, then anyone who stood in relation to p just as the protagonists of Gettier’s cases stand to the relevant propositions would have a justified belief that p that is not knowledge. (3) Therefore, it is possible for someone to have a justified true belief that is not knowledge.

The idea that thought experiments can be reconstructed as arguments is a bold thesis, and philosophers of science have raised a variety of concerns about this view. One worry is that Norton simply has not done enough work to substantiate either his reconstruction thesis or his overall narrative eliminativism (Cooper 2005). He has shown that some of the more famous thought experiments in the history of physics can be reconstructed as arguments, but this is not enough to demonstrate that all thought experiments can be reconstructed in this way, a point made more salient by the fact that Tamar Gendler has even challenged the idea that the Galilean experiment discussed above can be reconstructed as an argument (Gendler 1998). A second worry is that Norton cannot account for cases where people disagree about the results of a thought experiment; the most natural way of understanding this kind of disagreement is that even when people reconstruct the thought experiment as different arguments, they are still talking about the same thought experiment (Bishop 1999). A third worry is that it is not clear what constrains how we reconstruct thought experiments, and, if we cannot fix the premises in a fairly straightforward manner, then it will be hard to assess whether the argument is good or bad, something that makes it hard to see why thought experiments would ever be preferable to non-colorful traditional arguments (Brown 2011). And a fourth worry is that thought experiments and arguments seem different from a phenomenological point of view; when we perform a thought experiment it does not seem to us that we are evaluating premises, running through a series of inferential steps, or drawing conclusions (Cooper 2005). Most of these worries can be set aside rather quickly. It is probably best to think of the reconstruction thesis as a generic claim rather than a universal generalization, something that dampens somewhat the worry that Norton

58

Advances in Experimental Philosophy and Philosophical Methodology

and others defend the argument view on the basis of a small set of cases; but, in any event, they have surely done enough to make the argument view a live option worthy of continued examination. And it really is no objection to the argument view that people can draw different arguments from the same fictional narrative. As we will see, even rich fictional narratives are open to different interpretations; but, while narrative incompleteness is one way that thought experiments sometimes fail, this hardly counts against the view that they are arguments. Brown’s worry can be met in similar fashion. The fact that it can be hard to draw arguments from fictional narratives, and sometimes even harder to determine whether their premises are true or false, does not mean that thought experiments are not arguments; if anything, it simply points out that they are sometimes hard to use, something that perhaps should not be all that surprising. Set against these first three worries, the phenomenological worry seems rather more significant since it suggests that there is something involved in performing a thought experiment that is not captured by the suggestion that they can be reconstructed as arguments. The target here is not really the reconstruction thesis at all, but rather Norton’s eliminativism about the narrative content of thought experiments. According to Norton, since thought experiments can be reconstructed as arguments, the narrative details can be eliminated without compromising our ability to arrive at the conclusion; they play no significant epistemic or cognitive role, and merely provide the argument with colorful but inessential detail. This just seems wrong. As Tamar Gendler writes, “contemplation of an imaginary particular may have cognitive and motivational effects that differ from those evoked by an abstract description of an otherwise similar state of affairs” (2007: 68). According to this view, sometimes called the “mental model view,” thought experiments allow us to mobilize cognitive resources that would not otherwise be available, or at least that are not mobilized when we are asked to evaluate a formal argument. The basic idea is that thought experiments cannot be reconstructed solely as arguments without losing something important because of the way that they mobilize cognitive resources, even cognitive resources that might be available for use elsewhere. This worry is more substantial, and understanding it means looking at the mental model view of thought experiments.



Thought Experiments, Mental Modeling, and Experimental Philosophy

59

2. The mental model view Nancy Nersessian (1993, 2002, 2007) has been one of the leading proponents of the mental model view, according to which thought experiments involve the manipulation of mental models in our imaginations (see also Mišcevic 1992, 2007). Nersessian draws extensively on work from cognitive science on the construction and manipulation of mental models, especially JohnsonLaird’s work on the use of mental models in the comprehension of fictional narratives. The basic idea is that the narratives used in thought experiments prompt readers to construct mental models of the situations described in those narratives. These mental models are mental representations stored in short or long-term memory that differ from both propositional and pictorial representations, and thinking about the kinds of fictional narratives used in thought experiments involves operations performed directly on these mental representations rather than on, say, linguistic representations used to construct them. The features of the mental models are drawn from a highly organized sets of background beliefs that are abstracted from experience presented in associative chains or chunks; and manipulating these models involves mobilizing a number of cognitive resources, including background knowledge, individual expertise, practical know-how, and so on. Nersessian offers several reasons for resisting the move to look past the narrative content of thought experiments, perhaps the most compelling of which is that it is simply hard to square Norton’s eliminativism about the narrative content of thought experiments with the way thought experiments are almost always presented in both scientific and philosophical literature; if Norton is right that the narrative content of thought experiments can be eliminated without compromising our ability to arrive at the conclusion, then why present them in narrative form in the first place? Nersessian suggests that the answer is that thought experiments invite readers to do something that formal arguments do not; they invite us to “follow a sequence of events or processes as one would in the real world,” and “can reveal something in our experience that we did not see the import of before” (Nersessian 1993: 296–7). To see this, let us return to two examples discussed in the previous section. We can agree that it is possible to reconstruct these two thought experiments as arguments,

60

Advances in Experimental Philosophy and Philosophical Methodology

something that Norton and Williamson ably show us how to do, and still wonder whether purely formal presentations of the arguments would help us see, for example, that “connecting the two stones would cause the lighter stone to move faster and the faster stone to move slower, and this would mean that the composite stone will fall slower than the heavier stone would fall alone,” or that “it is possible for someone to stand in relation to some proposition p just as the protagonist of Gettier’s cases to stand to the relevant propositions.” The narrative content of the thought experiments allows us to see these things by prompting us to mobilize cognitive resources that would not otherwise be available, or at least that are not mobilized when we are asked to evaluate only the formal arguments themselves, and this is why thought experiments cannot be reconstructed solely as arguments without losing something important. There is controversy about the details of Nersessian’s account, something that should probably come as no surprise given how little agreement there is about the nature of mental content. Still, the natural link between mental models and the imagination is something that gets picked up on by philosophers interested in the role that thought experiments play in contemporary philosophy. Williamson talks repeatedly about the role that imagination plays in conducting thought experiments, writing that most philosophers allow “that a judicious act of the imagination can refute a previously well-supported theory,” and, “thought experiments do constitute arguments, but the imagination plays an irreducible role in warranting the premises” (2007: 188). Moreover, writing against a kind of pernicious philosophical exceptionalism, Williamson writes that one of his main goals is “to subsume the epistemology of thought experiments under the epistemology of counterfactual conditionals and metaphysical modality … and thereby to reveal it as an application of quite ordinary ways of thinking, not as something peculiarly philosophical” (2007: 180). This claim against exceptionalism echoes a central theme of Nersessian’s work: While thought experimenting is a truly creative part of scientific practice, the basic ability to construct and execute a thought experiment is not exceptional. The practice is a highly refined extension of a common form of reasoning. It is rooted in our abilities to anticipate, imagine, visualize, and re-experience from memory. That is, it belongs to a species of thinking by means of which we grasp alternatives, make predictions, and draw conclusions about potential real-world situations we are not participating in at that time. (1993: 292)



Thought Experiments, Mental Modeling, and Experimental Philosophy

61

While the role of imagination in conducting thought experiments is a central part of Williamson’s recent work on the philosophy of philosophy, perhaps no one embraces the role of imagination in philosophical thought experiments more than Tamar Gendler, who focuses on the role that thought experiments play in helping to make philosophical theories cognitively accessible. According to Gendler, thought experiments provide readers “with a powerful frame through which the target material … can be reconceptualized” (2007: 85). Put another way, thought experiments make cognitively accessible things that might otherwise not be. Gendler provides two vivid examples of this way of thinking about philosophical thought experiments. Both examples involve thought experiments intended to help readers see what is wrong with “first-person exceptionalism,” the view that we are deserving of unequal moral consideration. Gendler’s first example is the Biblical story of David and Bathsheba. According to this story, David, the King of Israel, had taken Bathsheba from her husband, Uriah, and arranged to have Uriah killed. God, angry with David, sends Nathan with following story: There were two men in a certain city, the one rich and the other poor. The rich man had very many flocks and herds; but the poor man had nothing but one little ewe lamb, which he had bought. And he brought it up, and it grew up with him and with his children; it used to eat of his morsel, and drink from his cup, and lie in his bosom, and it was like a daughter to him. Now there came a traveler to the rich man, and he was unwilling to take one of his own flock or herd to prepare for the wayfarer who had come to him, but he took the poor man’s lamb and prepared it for the man who had come to him. (2 Sam. 12)

Nathan’s story makes vivid what is wrong with first-person exceptionalism, and allows David to understand something about his own actions; by inviting David to think about why the rich man’s actions are morally impermissible, the narrative allows him to understand that Uriah deserved equal moral consideration. Gendler puts the point nicely: By framing the story so that David is not in a position to exhibit first-person bias with respect to what turns out to be his own actions, Nathan has enabled David to acknowledge a moral commitment that he holds in principle, but has failed to apply in this particular case. There is no ambiguity here about which

62

Advances in Experimental Philosophy and Philosophical Methodology commitment, on reflection, David endorses: The story he has been told is fully effective; it reshapes his cognitive frame, and brings him to view his own previous actions in its light. (2007: 82)

Gendler’s second example is Judith Thomson’s famous Violinist Case: You wake up in the morning and find yourself back to back in bed with an unconscious violinist. A famous unconscious violinist. He has been found to have a fatal kidney ailment, and the Society of Music Lovers has canvassed all the available medical records and found that you alone have the right blood type to help. They have therefore kidnapped you, and last night the violinist’s circulatory system was plugged into yours, so that your kidneys can be used to extract poisons from his blood as well as your own. The director of the hospital now tells you, “Look, we’re sorry the Society of Music Lovers did this to you— we would never have permitted it if we had known. But still, they did it, and the violinist is now plugged into you. To unplug you would be to kill him. But never mind, it’s only for nine months. By then he will have recovered from his ailment, and can safely be unplugged from you.” Is it morally incumbent on you to accede to this situation? No doubt it would be very nice of you if you did, a great kindness. But do you have to accede to it? What if it were not nine months, but nine years? Or longer still? What if the director of the hospital says, “Tough luck. I agree. But now you’ve got to stay in bed, with the violinist plugged into you, for the rest of your life. Because remember this. All persons have a right to life, and violinists are persons. Granted you have a right to decide what happens in and to your body, but a person’s right to life outweighs your right to decide what happens in and to your body. So you cannot ever be unplugged from him.” I imagine you would regard this as outrageous. (Thomson 1971: 48–9)

Just as Nathan’s story allows David to reframe his own actions and represent the moral implications of his own actions in a way he might not have been able to otherwise, Thomson’s case allows the reader to reframe attitudes about the relationship between mother and fetus and represent the moral implications of abortion in a way the reader might not have be able to otherwise. Both cases, then, provide a picture of how thought experiments involve imagination, and suggest that the narrative details of philosophical thought experiments cannot be eliminated without compromising our ability to arrive at the conclusions they are intended to support, even if those thought experiments could be reconstructed as arguments.



Thought Experiments, Mental Modeling, and Experimental Philosophy

63

3. Thought experiments and experimental philosophy Let’s suppose that all of this is right. Philosophical thought experiments can be reconstructed as arguments, but their narrative details cannot be eliminated without compromising our ability to arrive at the conclusions that they are intended to support. By putting philosophical cognition front and center, this way of thinking about philosophical practice actually underscores the significance of experimental philosophy because, whatever else it might be, or be taken to involve, experimental philosophy starts as the empirical study of how we think about philosophical issues. So we can agree with philosophers like Williamson, who council us to move past talking about philosophical intuitions while at the same time redoubling our interest in experimental philosophy; it is studying how our minds work when we think about philosophical issues that can help reveal what people actually think about the kinds of fictional narratives used in philosophical thought experiments, can help us to better understand what kinds of things influence how we think about these kinds of fictional narratives, and can help us use thought experiments more effectively. In short, this way of thinking about the role of thought experiments in philosophical practice makes experimental philosophy more important rather than less. It also underscores the cooperative nature of the relationship between experimental philosophy and the method of cases, and helps to guard against the tendency to exaggerate the differences between experimental and more traditional approaches to philosophical questions. Contrary to popular caricatures of the movement, which tend to portray experimental philosophy as a revolutionary break from more traditional ways of thinking about philosophy, most experimental philosophers are actually interested in quite ordinary philosophical questions, but think that we need to sharpen the tools that we use to answer those questions, including how we use thought experiments. Recent work on the nature of phenomenal consciousness provides a nice example. This work focuses on our folk psychological understanding of what is required in order for something to count as capable of subjective experience, and the empirical study of how people think about the kinds of fictional narratives used in this discussion has played a significant role in shaping

64

Advances in Experimental Philosophy and Philosophical Methodology

this discussion. Many of these fictional narratives involve the attribution of experiential states to both humans and nonhumans, and an interesting pattern emerges when we look carefully at the way that different people think about these fictional narratives, a pattern that reveals a striking difference between our folk conception of subjective experience and traditional philosophical conceptions of phenomenal consciousness. It seems that our folk conception of subjective experience marks a difference between sensation and perception that is absent from traditional philosophical conceptions of phenomenal consciousness, and that ordinary mental state attributions are sensitive to whether or not those states involve pleasant or unpleasant sensations. These observations have sparked a great deal of controversy, but there is no denying how they have helped move forward philosophical discussion of the nature of phenomenal consciousness (for discussion, see Fiala et al. 2014 and Sytsma 2014). Of course, even cooperative relationships can be complicated, and the relationship between experimental philosophy and how philosophers use thought experiments is complicated by the fact that what we think about fictional narratives is not always what we are intended to think about them, and by the fact that there is a whole lot we do not yet know about what shapes how we think about them, something that I have argued elsewhere raises important methodological questions about what role thought experiments can and should play in philosophical practice (see, e.g. Alexander 2012). But even here the news is not all bad, and there is reason to hope that learning more about philosophical cognition and what kinds of things shape how we think about the fictional narratives used in philosophical thought experiments may help us use them more effectively; after all, one way to see how things work is to think carefully about what causes them not to work. So perhaps it is worth mentioning here one of the ways that studying philosophical cognition might help to improve the way that we use thought experiments. Perhaps the most controversial claim in experimental philosophy is the claim that different groups of people think differently about the kinds of fictional narratives used in philosophical thought experiments. And it is easy to see why this claim would generate so much critical attention; after all, part of what seems to underwrite the use of thought experiments in philosophy is the idea that the fictional narratives used in these thought experiments help



Thought Experiments, Mental Modeling, and Experimental Philosophy

65

everyone to see the same thing about the kinds of modal claims that drive philosophical discussion and debate. Some philosophers have responded by challenging evidence used to support the claim that different groups of people think differently about these narratives (see, e.g. Nagel et al. 2013); others have responded by appealing to specific difficulties that arise from the nature of fictional narratives and how we engage with them—for example, that information presented in fictional narratives is not always cognitively accessible to readers, and that sometimes readers draw on information not presented in the fictional narratives themselves when forming judgments about what is being said. Tamar Gendler (2000, 2006) has already done great work bringing to light the first kind of difficulty, which she calls “imaginative resistance”; so let’s look briefly here at the second kind of difficulty, which we might call “narrative incompleteness.” Narrative incompleteness is a common feature in fiction, where many details about fictional worlds are left out of the presentation of those worlds, and for good reason; bringing them up would serve only to distract the reader from those details to which the author wants the reader to attend. There is latitude about how complete a narrative need be, of course, and room for both Hemingway and Joyce at the literary table; but the fact remains that writing fiction involves directing the reader’s attention to some details, while letting others go unmentioned, and this means that readers will be free to fill in certain, but certainly not all, of these details for themselves. Ernest Sosa (2009) has used this feature of fictional narratives and how we engage with them to challenge the idea that what experimental philosophers have found is evidence of genuine philosophical disagreement, arguing that what might appear to be evidence that people disagree about what to think about some fictional narrative might actually be evidence that people are just thinking about different fictional narratives. It is important to see that this observation about the nature of fictional narratives and how we engage with them cuts both ways, and so it is not clear that, or at least how, it will help someone interested in using fictional narratives to help people see things about the kinds of modal claims that drive philosophical discussion and debate. But it is also important to see that this observation underscores the relevance of experimental philosophy, since experimental philosophy is precisely what would be needed to help us determine what information people are responding to when they are being asked to respond to the kinds

66

Advances in Experimental Philosophy and Philosophical Methodology

of fictional vignettes commonly used in philosophical thought experiments. This is just one way that studying how our minds work when we think about philosophical issues is philosophically relevant; by revealing what people actually think about the kinds of fictional narratives used in philosophical thought experiments, we learn what kinds of things influence how we think about these kinds of fictional narratives, and knowing this can help us learn to use thought experiments more effectively.

4. Conclusion Timothy Williamson and others have urged philosophers to move past talking about philosophical intuitions since there is no agreement about what philosophical intuitions are supposed to be. I agree, and have suggested that we reframe recent discussions about philosophical methodology in terms of thought experiments, how they are supposed to work, and why they sometimes go wrong. There is certainly more to say, but I hope to have done enough here to show that the reframing of discussions of philosophical methodology only serves to underscore the significance of experimental philosophy. If Williamson is right that philosophical thought experiments can be reconstructed as arguments, but that their narrative details cannot be eliminated without compromising our ability to arrive at the conclusions that they are intended to support, then the empirical study of philosophical cognition is more important than ever, and we should embrace experimental philosophy as the best way to refine, and (sometimes) reform, the way we use thought experiments in contemporary philosophy.

Acknowledgments I would like to thank Bill Bechtel, Ingo Brigandt, Karim Bschir, Sara Green, Ron Mallon, Jennifer Nado, John Norton, Maria Serban, and Jonathan Weinberg for helpful comments on earlier drafts of this chapter. Support for work on this chapter was provided by the Center for Philosophy of Science at the University of Pittsburgh.



Thought Experiments, Mental Modeling, and Experimental Philosophy

67

References Alexander, J. (2010), “Is Experimental Philosophy Philosophically Significant?” Philosophical Psychology 23: 377–89. Alexander, J. (2012), Experimental Philosophy: An Introduction, Malden, MA: Polity Press. Bealer, G. (1998), “Intuition and the Autonomy of Philosophy,” in M. DePaul and W. Ramsey (eds), Rethinking Intuition, Lanham, MD: Rowman and Littlefield, 201–40. Bishop, M. (1999), “Why Thought Experiments are not Arguments,” Philosophy of Science 66: 534–41. Brown, J. (2011), The Laboratory of the Mind: Thought Experiments in the Natural Sciences, Second Edition, New York: Routledge. Cappelen, H. (2013), Philosophy Without Intuitions, Oxford: Oxford University Press. Cooper, R. (2005), “Thought Experiments,” Metaphilosophy 36: 328–47. Deutsch, M. (2009), “Experimental Philosophy and the Theory of Reference,” Mind & Language 24: 445–66. Deutsch, M. (2010), “Intuitions, Counter-Examples, and Experimental Philosophy,” Review of Philosophy and Psychology 1: 447–60. Fiala, B., Arico, A., and Nichols, S. (2014), “You, Robot,” in E. Machery and E. O’Neill (eds), Current Controversies in Experimental Philosophy, New York: Routledge 31–47. Gendler, T. (1998), “Galileo and the Indispensability of Scientific Thought Experiment,” British Journal for the Philosophy of Science 49: 397–424. Gendler, T. (2000), “The Puzzle of Imaginative Resistance,” Journal of Philosophy 97: 55–81. Gendler, T. (2004), “Thought Experiments Rethought—And Reperceived,” Philosophy of Science 71: 1152–63. Gendler, T. (2006), “Imaginative Resistance Revisited,” In S. Nichols (ed.), The Architecture of the Imagination, Oxford: Oxford University Press, 149–73. Gendler, T. (2007), “Philosophical Thought Experiments, Intuitions, and Cognitive Equilibrium,” Midwest Studies in Philosophy XXXI: 68–89. Goldman, A. (2007), “Philosophical Intuitions: Their Target, Their Source, and Their Epistemic Status,” Grazier Philosophische Studien 74: 1–26. Häggqvist, S. (1996), Thought Experiments in Philosophy, Stockholm: Almqvist and Wiksell International. Levin, J. (2005), “The Evidential Status of Philosophical Intuition,” Philosophical Studies 121: 193–224.

68

Advances in Experimental Philosophy and Philosophical Methodology

Mišcevic, N. (1992), “Mental Models and Thought Experiments,” International Studies in the Philosophy of Science 6: 215–26. Mišcevic, N. (2007), “Modeling Intuitions and Thought Experiments,” Croatian Journal of Philosophy 7: 181–214. Nagel, J., San Juan, V. and Mar, R. (2013), “Lay Denial of Knowledge for Justified True Beliefs,” Cognition 129: 652–61. Nersessian, N. (1993), “In the Theoretician’s Laboratory: Thought Experiments as Mental Modeling,” in D. Hull, M. Forbes, and K. Okruhlik (eds), Proceedings from the Philosophy of Science Association, 291–301. East Lansing, MI: Philosophy of Science Association. Nersessian, N. (2002), “The Cognitive Basis of Model-Based Reasoning in Science,” in P. Carruthers, S. Stich, and M. Siegal (eds), The Cognitive Basis of Science, Cambridge: Cambridge University Press, 133–53. Nersessian, N. (2007), “Thought Experiments as Mental Modeling,” Croatian Journal of Philosophy 7: 125–61. Norton, J. (1991), “Thought Experiments in Einstein’s Work,” in T. Horowitz and G. Massey (eds), Thought Experiments in Science and Philosophy, Lanham: Rowman and Littlefield Publishers, 129–48. Norton, J. (1996), “Are Thought Experiments Just What You Thought?” Canadian Journal of Philosophy 26: 333–66. Norton, J. (2004a), “Why Thought Experiments Do Not Transcend Empiricism,” in C. Hitchcock (ed.), Contemporary Debates in the Philosophy of Science, Madden, MA: Blackwell. Norton, J. (2004b), “On Thought Experiments: Is There More to the Argument?” Proceedings of the 2002 Biennial Meeting of the Philosophy of Science Association, Philosophy of Science 71: 1139–51. Sorensen, R. (1992), Thought Experiments, Oxford: Oxford University Press. Sosa, E. (2009), “In Defense of the Use of Intuitions in Philosophy,” in M. Bishop and D. Murphy (eds), Stich and His Critics, Malden, MA: Wiley-Blackwell, 101–12. Sytsma, J. (2014), “The Robots of the Dawn of Experimental Philosophy of Mind,” in E. Machery and E. O’Neill (eds), Current Controversies in Experimental Philosophy, New York: Routledge, 48–64. Williamson, T. (2007), The Philosophy of Philosophy, Oxford: Blackwell. Williamson, T. (2009), “Précis of The Philosophy of Philosophy,” Philosophical Studies 145: 431–34.

5

Gettier’s Method Max Deutsch

Every analytic philosopher knows that Edmund Gettier’s 1963 paper, “Is Justified True Belief Knowledge?” had, and continues to have, an enormous influence on work in epistemology. It led to the demise of the Justified True Belief (JTB) theory of knowledge and the now five-decades-plus of attempts to solve the problem of finding a replacement theory. Recently, however, the paper has acquired new significance: it looms large in contemporary discussions of method in analytic philosophy. For example, many experimental philosophers have pointed to Gettier’s paper as the locus classicus of the “method of cases,” a method that appeals to hypothetical cases and in which “intuitions” about these cases are taken as evidence for or against general philosophical theories. Not that experimental philosophers are solely responsible for the focus on Gettier’s paper and the method of cases—many more traditional analytic philosophers, long before experimental philosophy broke on to the scene, held up Gettier 1963 as clear example of an application of the method of cases. And many analytic philosophers—not just experimental philosophers—take this method to represent a core practice, something by which analytic philosophy is defined. In an earlier work (Deutsch 2010), I argued that this is a mistake. Gettier does not employ the method of cases, and intuitions play no role in his refutation of the JTB theory. I also argued that this is likely true of analytic philosophy, generally speaking. Given the standard construal of the method of cases, the method must be very rarely employed. Certainly, given that standard construal, it is by no means a core practice. I will rehearse later the

70

Advances in Experimental Philosophy and Philosophical Methodology

important elements of these earlier arguments of mine, but their main idea is roughly this: Gettier argues for the judgments he makes about his famous cases and makes no explicit appeal anywhere to intuitions or the intuitiveness of any principle or proposition. Hence, Gettier’s attack on the JTB theory does not involve an application of the method of cases. This idea and the arguments based on it seem to have convinced only a few. Even philosophers who, like me, are opposed to certain projects in experimental philosophy disagree with me about the method of cases. In fact, if anything, the metaphilosophical view that analytic philosophy employs the method of cases is even more entrenched, and certainly more often explicitly endorsed, than ever before.1 There are certainly plenty of philosophical methodologists who hold up Gettier and the argument in his 1963 paper as the example of the application of the method. It is high time, then, to revisit the issue. Is the argument in Gettier (1963) an example, indeed, a paradigm, of one employing the method of cases? I will argue that, on the usual understanding of what the method of cases involves, the answer is no. I will not, in this paper, attempt to argue for the extension of this conclusion to other arguments in analytic philosophy, but surely there is something to the idea that Gettier’s argument is paradigmatic of a certain kind of analytic philosophizing. If there is no application of the method of cases in Gettier (1963), that is at least a very strong hint that the method is much less common than is ordinarily supposed. My thinking about Gettier’s paper has evolved in certain ways in response to the metaphilosophical work of others in which Gettier’s paper is taken as representative of analytic philosophical methods. So, beyond reviewing and supplementing the reasons for holding that Gettier does not apply or endorse the method of cases (on the usual understanding of the method), I will delve into several related issues, including: (a) the reception of Gettier’s paper in the 1960s and the rise of “Gettierology;” (b) the question of exactly where intuitions are supposed to play an evidential role in Gettier’s argument; (c) the issue of whether some of the reasoning in Gettier 1963 is purely abductive; (d) the significance of the distinction between producing and consuming a philosophical case or thought experiment; and (e) the significance (or lack thereof) of textbook renderings of Gettier’s argument against the JTB theory. My opinions about these issues all hinge on one central claim: the argument



Gettier’s Method

71

in Gettier (1963) is not an application of the method of cases.2 Accordingly, I turn first to defending this claim.

1. Arguments, not intuitions I will argue in this section that, since Gettier does not appeal to intuitions in his argument against the JTB theory, the argument does not involve an application of the method of cases. Gettier’s argument does, famously, involve hypothetical cases and the assertion of what Gettier takes to be true in these cases, but intuitions about these cases, or what’s intuitively true with respect to them, play no role whatsoever. On the usual construal of the method of cases, intuitions are an essential component; the method, on the usual construal, is one whereby a principle or theory is rejected if intuitions about a hypothetical case conflict with the principle or theory. It is this, I claim, that Gettier is not doing; in rejecting the JTB theory he is not appealing to intuitions about his cases. Here, from a recent paper by Anna-Sara Malmgren, is an example, of what I am calling the “usual construal” of the method of cases: In brief outline, it has the following structure: the hypothesis or theory that is under evaluation states or entails some modal claim (typically a necessary bi-conditional or one-way implication) and in a thought experiment we check that modal claim against our intuitive verdict on an imaginary problem case. If the claim conflicts with our intuitive verdict, this is treated as strong evidence against the theory—indeed the theory may be abandoned as a result. We say that we found a counterexample to it.3 (Malmgren 2011: 264)

It is difficult to interpret a construal such as this without understanding it as implying that the intuitiveness of “verdicts” on imaginary cases matters a great deal: the construal claims that conflict with intuitive verdicts is what is “treated as strong evidence against the theory.” It does not say simply that the facts about, or truths with respect to, imaginary cases are treated as strong evidence against the theory (although it is of course compatible with this simpler claim); rather, it is when the verdicts about imaginary cases are intuitive verdicts that they are treated as strong evidence against the theory. This is important, since, on my view, if one takes “intuitive” and cognates out of the usual construal of

72

Advances in Experimental Philosophy and Philosophical Methodology

the method of cases, one gets a perfectly accurate description of what Gettier and other analytic philosophers—and, indeed, inquirers of all kinds—often do: they test theories against cases. This involves determining whether what is true with respect to this or that case is compatible with the theory. Theorists of all kinds test their theories in this way, and “the method of cases” would be a perfectly acceptable shorthand for this sort of activity, if not for the association with intuitions and intuitiveness, and if not for the suggestion that the method is peculiar to philosophy. My complaint with what I have been calling the “usual construal” of the method of cases is precisely that it takes intuitions and intuitiveness to be a central and essential component of the method.4 Where, exactly, are intuitions supposed to come in? Relative to Gettier’s original two cases, which I will here call the “10 Coins Case” and the “Brown in Barcelona Case” (and whose full presentations, just as they appear in Gettier 1963, can be found, for readers wishing to refresh their memories, in the appendix to this paper, “Gettier’s Original Cases”), intuitions are alleged to come in with respect to the to the judgment that the agent in each case does not know that a certain proposition is true. The agent in each case is “Smith.” Relative to the 10 Coins Case, Gettier judges, and invites his readers to judge, that: (G1) Smith does not know that the man who will get the job has ten coins in his pocket.

Relative to the Brown in Barcelona Case, Gettier judges, and invites his readers to judge, that: (G2) Smith does not know that either Jones owns a Ford or Brown is in Barcelona.5

In setting up the cases, Gettier argues that Smith truly and justifiably believes, in the 10 Coins Case, that the man who will get the job has ten coins in his pocket, and, in the Brown in Barcelona Case, that either Jones owns a Ford or Brown is in Barcelona. However, if (G1) and (G2) are true in his cases, as Gettier maintains, it follows that there can be instances of justified true belief that are not instances of knowledge and, hence, that the JTB theory is false.6 The question is: what justifies, or provides evidence for, (G1) and (G2)?7 According to many who have reflected on the matter, Gettier means us to



Gettier’s Method

73

accept that (G1) and (G2) are true in his cases, on the basis of intuition.8 Here is a smattering of quotations that express this common view: At one time many people accepted the doctrine that knowledge is justified true belief. But now we have good evidence to the contrary, namely our intuitions that situations like those described in the Gettier literature are possible and that the relevant people in those situations would not know the things at issue. (Bealer 1996: 122; emphasis added) Examples of this practice in epistemology [of appealing to intuitions as evidence] abound. Most famous of these examples is Edmund Gettier’s use of two thought-experiments to generate intuitions intended to prosecute the claim that a person knows that p just in case that person’s true belief is justified. (Alexander and Weinberg 2007: 56–7; emphasis added) Many philosophers [and Neta makes it clear later that he himself counts as one of the many] assume that our intuitions—our spontaneous, seemingly non-inferential and non-perceptual inclinations to judge—as to whether or not a particular property is instantiated in a hypothetical case can provide us with evidence that bears on some philosophical hypothesis about the nature of that property. For instance, our intuition to the effect that the property of knowledge is not instantiated in Gettier cases provides us with evidence against justified, true belief accounts of knowledge. (Neta 2012: 329; emphasis added)

This common view, however, is wrong. In the first place, Gettier nowhere uses intuition language in his presentation of the cases. He does not, for example, explicitly say that (G1) or (G2) is intuitive, nor that we are meant to accept them on the basis of their intuitiveness. Of course, by itself, this doesn’t imply that Gettier is not appealing to intuition in support of (G1) and (G2); such an appeal might be implicit. But the force of this “might” is rather weak. Gettier might, after all, intend us to accept the claims on the basis of tealeaf readings or the position of the stars. Surely, more than the bare possibility that Gettier is appealing to intuitions in support of (G1) and (G2) is needed if we are to say that this is, in fact, what Gettier is up to.9 In the second place, Gettier leaves no doubt as to what, in his view, justifies (G1) and (G2): He gives an explicit argument for them. Here is the argument in a nutshell: In both cases, though Smith believes truly and with justification, his beliefs are merely luckily true. Gettier argues that merely luckily true beliefs are not knowledge and hence that (G1) and (G2) are true. I will return

74

Advances in Experimental Philosophy and Philosophical Methodology

to this argument soon, and will provide textual evidence that this argument is explicit in Gettier’s presentations, but first I want to pause briefly to draw what seems to me to be an obvious methodological moral from the fact that Gettier offers an explicit argument in defense of (G1) and (G2). In general, if one gives an explicit argument for p and does no more by way of indicating why p ought to be accepted, then one can be taken as having provided what one takes to be the justification or evidence for p. This is especially true in argument-centric disciplines such as philosophy. In such disciplines, it is standard practice to “show your cards” when it comes to evidence. Indeed, this is the main point of, for example, writing a philosophical paper defending p; one aims to say explicitly what the evidence for p consists in. So, given that Gettier offers an explicit argument for (G1) and (G2), and given that he does no more by way of explicitly convincing us that they are true than by presenting us with this argument, we can fairly confidently conclude that Gettier does not mean us to accept (G1) and (G2) on the basis of their intuitiveness. If he had meant us to do that, he would have said so. This, I think, suffices as a reply to those moderates regarding the method of cases (e.g. Bengson 2014; Chalmers 2014; Devitt forthcoming; Weinberg 2014; Mallon this volume; and Nado and Johnson this volume) who would say (and have said) that the method does not exclude argument-based evidence for judgments about hypothetical cases. Their idea is that offering an argument for p does not foreclose offering intuitions, perhaps implicitly, as additional, complementary evidence for p. Hence, even if Gettier argues for (G1) and (G2), for example, their intuitiveness might still be relevant and meant to qualify as additional evidence for their truth. My reply is that, yes, that’s possible, but there is no evidence to suggest it’s actual, and, moreover, given the general point about the function of explicit argumentation in argumentcentric disciplines such as philosophy, there is reason to suppose that it is non-actual.10 The more usual view of how the method of cases is employed in Gettier (1963), however, is that its use there involves a brute appeal to intuition. There is an argument against the JTB theory in Gettier, according to this usual view, but there’s no argument for the premises of this argument, among which are (G1) and (G2). These, according to the usual view, are meant to be supported



Gettier’s Method

75

by intuition alone. Since, as I say, there is an explicit argument for (G1) and (G2) in Gettier 1963, this usual view has always been very puzzling to me. It is hard for me to understand how anyone who has actually read Gettier’s short paper could fail to notice it. I have said that the argument is that (G1) and (G2) are true because Smith’s beliefs, in both the 10 Coins and Brown in Barcelona cases, although true and justified, are true by dint of luck, a kind of luck that prevents them from qualifying as instances of knowledge. How and where does this argument appear in Gettier’s presentations of his cases? In the 10 Coins Case, luck is involved in the following way. Although Smith justifiably and truly believes that the man who will get the job has ten coins in his pocket, this belief as Gettier himself puts it, “is true in virtue of the number of coins Smith’s pocket, while Smith does not know how many coins are in Smith’s pocket ….” So, luck crops up here first: Smith’s belief is true in virtue of a fact that Smith does not know. Furthermore, again as Gettier himself puts it, Smith “bases his belief [that the man who will get the job has ten coins in his pocket] on a count of the coins in Jones’s pocket, whom he falsely believes to be the man who will get the job.” So, if Smith gets it right, as he does, this is mere coincidence, since the belief is based on something irrelevant to the belief ’s truth. Gettier makes it clear that it is the presence of this amount and kind of luck that qualifies as the reason for asserting (G1). Gettier even explicitly says that that (G1) is true “for” the reason just canvassed. In the Brown in Barcelona Case, luck is involved in a similar way. Although Smith justifiably and truly believes that either Jones owns a Ford or Brown is in Barcelona, the belief is true in virtue of the fact that Brown is in Barcelona—in the case, Jones does not own a Ford. Yet, as Gettier puts it, Smith is “totally ignorant” of Brown’s whereabouts. So, again, the belief is true in virtue of something that Smith does not know, a proposition that, as Gettier stresses, “by the sheerest coincidence, and entirely unknown to Smith” just happens to be true. In the Brown in Barcelona case, the coincidental truth of Smith’s belief that either Jones owns a Ford or Brown is in Barcelona is emphasized by comparing it to other beliefs that Smith forms, ones involving different places (Boston and Brest-Litovsk), all of which, like Barcelona, are chosen by Smith “quite at random.” And again, just as in the 10 Coins Case, we find Gettier explicitly concluding, via inference, that (G2) is true. He explicitly says that if

76

Advances in Experimental Philosophy and Philosophical Methodology

Jones does not own a Ford but Smith’s belief is, in the ways just described, only luckily true, then (G2) follows.11 Gettier offers an explicit, anti-luck argument for (G1) and (G2). So the view that it is “just intuitions” that conflict with the JTB theory, or, even less plausibly, that Gettier himself means us to understand that it is just intuitions that reveal the falsity of the JTB theory, is clearly a mistake. For reasons I gave above, intuitions are very likely not meant by Gettier to be even part of the evidential basis for accepting (G1) and (G2). These claims of mine have been resisted in metaphilosophical work that discusses Gettier 1963. I have already mentioned the fact that many characterize Gettier’s method as an application of the method of cases. To characterize it this way is to very strongly suggest, if not strictly to imply, that there’s no argument for (G1) or (G2) in Gettier’s paper. It is also, of course, to imply that intuitions are crucially evidentially involved. Some go so far as to explicitly deny that Gettier 1963 contains an explicit argument for (G1) and (G2). For example, in a recent paper, Jennifer Nagel denies this, writing: Gettier does not present an explicit argument showing exactly why Smith’s judgment does not amount to knowledge: he does not offer any positive analysis of knowledge of his own, nor does he specify any necessary conditions on knowledge which are lacking in this case. But despite neither explaining nor perhaps even knowing exactly why he feels the way he does about this case, Gettier seems convinced that his intuitions will be felt by others and by himself on other occasions. (Nagel 2012: 503)

There is much to complain about in this passage but I will here limit myself to two points. First, a positive analysis of knowledge is obviously not required for having an explicit argument or explanation for the truth of (G1) and (G2), as Nagel herself recognizes when she suggests that such an argument or explanation might proceed via the specification of “necessary conditions on knowledge” that are lacking in the 10 Coins and Brown in Barcelona cases. Second, Gettier does explicitly argue for (G1) and (G2), as I have been at pains to show, and this argument seems perfectly plausibly characterized as one that proceeds via the specification of a necessary condition on knowledge (requiring that justified true beliefs not be merely luckily true) that goes unmet in Gettier’s cases.



Gettier’s Method

77

Does Gettier give an exact explanation of the truth of (G1) and (G2)? Here I’m willing to concede that the explanation is perhaps not perfectly exact. After all, Gettier does not attempt to explain just what the kind of knowledgepreventing luck that is present in both of his cases comes to. There’s no theory or definition of the relevant sort of luck; indeed, there’s no general characterization whatsoever. However, if we widen our focus to include not just Gettier’s method but also the methods of the so-called “Gettierology” literature his paper spawned, we see, I think, many attempts at precisely this. In part to substantiate this claim, and in part to correct a common but mistaken attitude towards this literature, I turn in the next section to reviewing a couple of examples of post-1963 attempts to analyze knowledge.

2. Gettierology It is well known that Gettier 1963 spawned an enormous literature attempting to solve the “Gettier problem,” the problem of adding to, or replacing, the JTB theory in order to correctly capture the necessary and sufficient conditions on knowledge. This literature is fairly widely regarded as a failure; many philosophers believe that the holy grail of a complete analysis of knowledge has not been found, and some take this to suggest that there is no such analysis to be had.12 It is also widely regarded as an example of conceptual analysis and the method of cases gone “off the rails.” Many philosophers confess embarrassment at the cycles of increasingly baroque analyses and ever more nuanced (but still supposedly intuitive) counterexamples that crop up in this literature. In fact, the Gettierology literature is sometimes held up as the example of What’s Wrong with Analytic Philosophy, at least among those who think there must be something wrong with it. It is rare, then, to hear this literature described in positive terms.13 In my view, these attitudes are overly pessimistic. A more positive view of post-1963 work on the Gettier problem is this: much of this work has been illuminating and helpful in characterizing, or at least in attempting to characterize, the kind or kinds of knowledge-preventing luck that is present in Gettier cases. Moreover, and very strikingly, given the alleged prominence

78

Advances in Experimental Philosophy and Philosophical Methodology

of the method of cases in post-1963 work on the Gettier problem, this work has proceeded in an entirely intuition-free way. Take, for example, the very first (and not quite post-1963) attempted solution to the Gettier problem, the solution that has come to be known as the “no false lemmas” solution, first bruited by Michael Clark (1963). Clark claimed that Gettier’s Smith character fails to know, in both of Gettier’s cases, because, in each case, he (Smith) infers his beliefs from falsehoods. Clark thus proposed that adding a condition to the traditional analysis explicitly requiring that beliefs adding up to knowledge not be inferred from falsehoods would rescue the traditional analysis. As we now know, Clark was wrong about this. There are cases of justified true belief, not based on any false lemma, that are not cases of knowledge.14 But this, by itself, should not lead us to write off Clark’s work as a failure. After all, pre-Clark 1963, it was not known that there were such cases. That is, Clark’s mistake led to a genuine philosophical discovery, namely the discovery that believing truly, with justification, and based on no false lemma, does not suffice for knowing. Beyond the fecundity of Clark’s mistake about what counts as a sufficient condition on knowledge, there is his equally important proposal that not inferring a belief from a falsehood is necessary for that belief to qualify as knowledge. This proposal, though also now known to be wrong,15 is important because of its independence from the sufficiency claim and because of its considerable initial plausibility. Clark, although wrong about what suffices for knowing, certainly seemed, at least at first, to be on to something regarding what is necessary for knowing. In both of Gettier’s original cases, it does seem as though Smith’s justified true beliefs fail to be knowledge because he infers these beliefs from falsehoods. Clark’s proposed necessary condition can be seen as an attempt to provide precisely what Nagel says is missing in Gettier’s paper, namely an exact explanation of why “Gettiered” justified true believers fail to know. Where do intuitions come in in Clark’s proposals? Nowhere. Clark is explicit that he accepts Gettier’s arguments for judgments similar to (G1) and (G2). For example, discussing a variation of the Brown in Barcelona case, he writes: Yet Brown’s wild guess can hardly be regarded as providing Smith with knowledge merely because it happens to be right. In this case, then, the grounds



Gettier’s Method

79

on which Smith believes (1) are true, but the grounds on which he accepts these grounds, viz. that Brown knows them, are false. (Clark 1963: 448)

This is an argument that Smith, despite justifiably and truly believing it, does not know (1). ([1] is the proposition that Jones owns a Ford. But this and the other details of the case are immaterial to the points I will make about it.) It is not the presentation of an intuition; it’s an argument for the truth of the judgment that Smith fails to know (1). Furthermore, it is an argument that echoes Gettier’s invocation of knowledge-preventing luck. Smith fails to know because his belief is based on a “wild guess” that just “happens to be right.” This argument is then supplemented with Clark’s account of why Smith’s belief is luckily true in a way that prevents knowledge: Smith’s belief is only luckily true, and hence not knowledge, because it is based, ultimately, on a false lemma (or “ground” in Clark’s own terminology). “Intuition” and cognates are nowhere to be found in Clark’s paper. There are, however, arguments for the “Gettier judgments” (i.e. judgments to the effect that some justified true believer fails to know) that Clark discusses. In addition, one can extract from Clark’s paper something close to a theory of what makes a belief merely luckily true: luckily true beliefs are those true beliefs based on false lemmas or grounds. Other examples of Gettierology are similarly rich, in that they offer insights into possible anti-luck necessary conditions on knowledge, and are similarly intuition-free. Take the “causal theory of knowledge” proposed by Alvin Goldman (1967). Just as in Clark (1963) and Gettier (1963), “intuition” appears nowhere in Goldman (1967). And, as is true of its precursors, there are anti-luck arguments for the Gettier judgments that Goldman makes throughout the paper. Goldman disagrees with Clark about what the relevant sort of luck consists in, but they agree that it is the presence of luck that accounts for the lack of knowledge in Gettier cases. Here is Goldman’s presentation of his strategy in Goldman (1967) (Goldman is here discussing the Brown in Barcelona Case; p is the proposition that either Jones owns a Ford or Brown is in Barcelona and q is the proposition that Jones owns a Ford): Michael Clark, for example, points to the fact that q is false and suggests this as the reason why Smith cannot be said to know p. … I shall make another

80

Advances in Experimental Philosophy and Philosophical Methodology hypothesis to account for the fact that Smith cannot be said to know p, and I shall generalize this into a new analysis of “S knows p.” Notice that what makes p true is the fact that Brown is in Barcelona, but that this fact has nothing to do with Smith’s believing that p. That is, there is no causal connection between the fact that Brown is in Barcelona and Smith’s believing p. … Thus one thing that seems to be missing in this example is a causal connection between the fact that makes p true (or simply: the fact that p) and Smith’s belief of p. The requirement of such a causal connection is what I wish to add to the traditional analysis. (Goldman 1967: 358; emphasis in original)

Goldman’s discussion of Gettier cases, like Clark’s, never presents judgments about them as resting on intuition. For example, here he is discussing a counterexample (due to Saunders and Champawat 1964) to the alleged sufficiency of Clark’s conditions. For my purposes, the details of the counterexample do not matter (though I should explain that the “Patterns” Goldman mentions are supposed to be types of causal connection between facts and beliefs). What matters is the clear presence of an argument for the relevant Gettier judgment: But, as it happened, it is purely a coincidence that Jones owns a Ford today as well as yesterday. Thus, Smith’s belief of p is not connected with p by Pattern 2, nor is there any Pattern 1 connection between them. Hence, Smith does not know p. (Goldman 1967: 367)

Note that it is not an intuition but rather the presence of luck—“pure coincidence”—that Goldman takes to be evidence for the conclusion that Smith does not know. Then, an explanation, different from Clark’s, is given: the luck in this case is due to lack of appropriate causal connections, not the presence of false lemmas or grounds. So, here’s the picture of early work on the Gettier problem that one should come away with after examining examples such as Clark (1963) and Goldman (1967). There are, first, no explicit appeals to intuitions as evidence in this work. Instead, early Gettierologists appear to accept Gettier’s arguments for (G1) and (G2) and to argue on similar grounds for the further Gettier judgments they discuss. This strikes me as something approaching a case-study demonstration that the method of cases, under its usual “intuition”-involving description, is simply not employed in this work. Second, early Gettierology



Gettier’s Method

81

is usefully viewed as the beginnings of what is now known as “anti-luck epistemology.” Like latter-day anti-luck epistemologists, the early forebears believed that it is the presence of luck in Gettier cases that argues for our Gettier judgments. Disagreements between anti-luck epistemologists are disagreements over how to accurately, and fully generally, characterize this kind of luck. Early instances of anti-luck epistemology ambitiously sought a full analysis of knowledge. Perhaps—perhaps—this ambition has not been, and perhaps— perhaps—cannot be, realized. Even so, the ambition strikes me as having been extremely philosophically fruitful. A variety of new necessary conditions, from “no false lemmas” and “appropriate causal connections” on up to the more recent “sensitivity” and “safety” conditions, have been proposed, and some of these proposals might even be correct. None of them is predicated on the intuitiveness of this or that counterexample. Instead, all of them trace their method back to Gettier’s, and to his arguments, not intuitions, in favor of (G1) and (G2).

3. Intuitions elsewhere? Berit Brogaard (2014) has recently argued that even if it is true (as she takes it to be) that Gettier offers an argument for (G1) and (G2), there is still a case for supposing that Gettier’s method relies on intuition.16 After all, this argument has a premise. What is the intended evidence for it? Brogaard reasons that it is unlikely that Gettier meant to treat this premise as an assumption, and unlikely, too, that he took it to have no justification at all. So, she concludes, it is likely that he implicitly meant it to be justified by an intuition (which she takes to be a Bealerian “intellectual seeming”).17 As I understand Brogaard’s position, she is not defending the view that Gettier’s method is an application of the method of cases. To defend that view, she would need to insist that Gettier treats intuitions as evidence for (G1) and (G2). But she is clear that this is not her view. Intuitions, according to Brogaard, come in at an earlier stage in Gettier’s reasoning. Still, my insistence that Gettier does not apply the method of cases would lose much of its bite if the correct description of Gettier’s method did aver that intuitions are

82

Advances in Experimental Philosophy and Philosophical Methodology

involved, but just at an earlier stage in the overall argumentative structure of Gettier (1963). But Brogaard’s position is not especially plausible. It is true that Gettier offers no explicit argument, or reason, for the premise of his argument for (G1) and (G2), which I take to be roughly the premise that luckily true beliefs do not qualify as knowledge. And it is true also that it is unlikely that Gettier meant this premise to be treated as an assumption or such that it has no justification whatsoever. But there are at least two alternatives to the view that this makes it likely that Gettier meant an implicit appeal to intuition to justify the premise that luckily true beliefs do not qualify as knowledge. The first alternative, and the one that I favor, is that Gettier simply did not consider the question of what might justify this premise of his argument for (G1) and (G2), and hence did not intend any justification at all for it. But that’s not tantamount to treating it as an assumption, or supposing it to be unjustified or unjustifiable. In all likelihood, if asked, Gettier would say that this premise is, or could be, justified, but probably the question never occurred to him. Gettier very likely simply took the premise as the starting point in his reasoning about the JTB theory. Every argument in any text, philosophical or otherwise, takes at least one premise for granted or as a starting point. That is not an important epistemological or methodological truth; it is rather a reflection of practical, human limitations on time, space, and effort. As the saying goes, “you gotta start somewhere,” and it’s just as true of philosophical reasoning as it is of anything else. The propositions that are taken as philosophical starting points fall into widely heterogeneous kinds. Some are regarded as a priori, some empirical. Some are taken to be analytic, while some are not. Some are very general in character, while others are more specific. Some constitute a supposed piece of “common sense,” while others are rather recherché conclusions from mathematics, science, or some other area of inquiry. It is a mistake to suppose that all philosophical starting points are regarded as intuitive—some are and some most definitely are not. These facts about philosophical starting points reveal that Gettier having taken his anti-luck premise as a starting point is very poor evidence for supposing him to have taken it to be justified by way of intuition.



Gettier’s Method

83

The second alternative is that Gettier implicitly meant his anti-luck premise to be evidentially supported by further arguments. Often enough, one begins one’s arguments with premises for which one could offer argumentative support, but one does not explicitly give the this support, because time, or even just what one wants to focus on and emphasize, does not allow it. It is not as though the anti-luck premise of Gettier’s argument for (G1) and (G2) is such that arguments for it are impossible or inconceivable. Indeed, much of the anti-luck epistemological literature can be regarded as providing various reasons for supposing Gettier’s anti-luck premise to be true. Perhaps Gettier had an argument up his sleeve for the claim that luckily true beliefs don’t add up to knowledge. This is no less likely, at any rate, than that he meant it to be justified by intuition.18

4. The order of explanation I have claimed that Gettier argues for (G1) and (G2) via a premise barring luckily true belief from adding up to knowledge. An alternative view is that the order of explanation goes the other way: Gettier is instead arguing for the anti-luck condition via abduction from (G1) and (G2). He is arguing, that is, that the anti-luck condition is the best explanation of the truth of (G1) and (G2). How might this be a problem for my claim that Gettier does not apply the method of cases, or appeal to intuitions as evidence, in arguing against the JTB theory? It is a problem for that claim, it seems to me, only if Gettier intended the relationship between (G1) and (G2) and the anti-luck condition to be fully abductive. For, in that case, there would remain the question of the intended justification for (G1) and (G2), a question some might claim should be answered by taking Gettier to be implicitly relying on intuitions as evidence.19 On the other hand, if he didn’t intend the relationship to be fully abductive—if, in other words, he meant the order of explanation to go both ways, from (G1) and (G2) via abduction to the anti-luck condition, but from the anti-luck condition via deduction to (G1) and (G2) as well20—then there is no bar to supposing that Gettier’s method is entirely intuition-free. For, in that case, there would not remain any question of the intended justification for

84

Advances in Experimental Philosophy and Philosophical Methodology

(G1) and (G2)—that justification would simply be that the anti-luck condition is true and that it implies (G1) and (G2). That Gettier perhaps also meant us to suppose that we derive some justification for the anti-luck condition by seeing that it best explains the truth of (G1) and (G2) is no grounds for taking Gettier to be making any implicit appeal to intuition in support of, or even in partial support of, (G1) and (G2). So, did Gettier intend us to take the relationship between (G1) and (G2) and the anti-luck condition to be fully abductive? In my view, this is extremely unlikely. I have already argued at some length that Gettier’s explicit appeal to the anti-luck condition is an appeal to a premise in an argument for (G1) and (G2). The language he uses in presenting his cases strongly suggests this. His talk of “sheer coincidence” and of certain propositions characterizing his cases merely “happening” to be true certainly sounds as though it is the citation of reasons in favor of (G1) and (G2). I think it is possible that Gettier also intended his presentation of the cases and his assertion of (G1) and (G2) to count abductively in favor of the anti-luck condition, but that, as I’ve said, is consistent with there being no application of the method of cases nor any appeal to intuition in his argument. So long as Gettier does not treat the anti-luck condition solely as the conclusion of an abductive argument, there is no reason to think that intuitions enter the picture at all. Another reason for holding it unlikely that Gettier intended the anti-luck condition only as a conclusion and not as premise is that (G1) and (G2), were, when Gettier first asserted them, not established, or widely agreed on, truths. Given the widespread acceptance of the JTB theory, their negations were very much live options relative to the debate over the correct analysis of knowledge. Typical cases of abduction proceed from premises that are taken as granted. But (G1) and (G2) would not have been taken as granted in 1963; they were controversial claims, the truth of which implied the falsity of a longstanding orthodox theory of knowledge. A last reason I’ll discuss for rejecting an account of Gettier’s reasoning involving the anti-luck condition as fully abductive turns on the distinction between producing and consuming a philosophical case or thought experiment. However, since, as I believe, the issues raised by drawing this distinction are of central methodological significance, I devote a section of their own to them.



Gettier’s Method

85

5. The producer-consumer distinction What, exactly, does it mean to say that Gettier treats intuitions as evidence for (G1) and (G2)? Is he himself supposed to have found these propositions intuitive and to have then taken their intuitiveness-for-him as evidence for their truth? That strikes me as an odd and unlikely account of how Gettier arrived at the view that (G1) and (G2) are true. Even if one objects to the details of the picture I have been painting, whereby Gettier inferred (G1) and (G2) from an explicitly proposed necessary condition on knowledge barring luckily true beliefs, it would be difficult, I think, to reject the general outline of my picture. Gettier inferred (G1) and (G2) from something. As the producer or inventor of the 10 Coins and Brown in Barcelona cases, he did not spontaneously and noninferentially judge that (G1) and (G2) are true relative to them. How would these propositions have presented themselves to his consciousness in the first place? They didn’t materialize unbidden, as if by magic, ready to test for intuitiveness. Instead, in settling on the details of his cases, Gettier engaged in careful and creative imaginative and stipulative activity. Then he inferred, from the facts of the cases thus imagined and stipulated, that (G1) and (G2) are true. If intuitions are evidence for (G1) and (G2), they are not Gettier’s intuitions, for the simple reason that Gettier almost certainly did not come to believe (G1) and (G2) on the basis of intuition. Arriving at that belief, for Gettier, did not involve the passive sort of cognizing characteristic of intuiting that something is so. It must instead have involved an active “thinking through” the details of the cases, and reasoning that his conclusions about them were true. However, as consumers of Gettier’s cases, you and I might experience intuitions that have (G1) and (G2) as their contents. Did Gettier treat our intuitions, yours and mine, as evidence? Presumably, Gettier came to believe, and perhaps know, that his cases constituted counterexamples to the JTB theory before he presented them to the rest of the world. If so, then how could your, or my, intuitive reactions to the cases matter to the method he employed in arriving at his view? It is possible that he tried to assure himself that if people were presented with his cases then they would have the relevant intuitions. Then again, it is possible that he didn’t tarry over the question in the

86

Advances in Experimental Philosophy and Philosophical Methodology

slightest, or perhaps pessimistically believed that longstanding commitment to the JTB theory would skew intuitions in the wrong direction. None of these speculations strikes me as relevant to the question of what the evidence for (G1) and (G2) is, or what Gettier took it to be. Imagine that Gettier wrote up his cases but never published them and so gave no one the opportunity to have this or that intuition with respect to them. Would that be a situation in which Gettier had no evidence, or considerably less evidence, for (G1) and (G2)? No. It would be a situation in which belief in the falsity of the JTB theory would still be fully evidentially supported and—for Gettier, if no one else—fully justified. How we consumers of Gettier’s cases do, or would, react to them psychologically is, therefore, methodologically irrelevant. These reflections on the producer-consumer distinction as applied to Gettier’s cases strongly suggest that the fully abductive model of the reasoning in Gettier 1963, discussed in the previous section, cannot be right. For they show, I think, that there is no sense in which (G1) and (G2) were taken by Gettier as given or as granted. In particular, there’s no sense in which (G1) and (G2) were taken by Gettier as intuitively given or granted. So, the question of what best explains their truth could have arisen for Gettier only after he had supplied himself with reasons—straightforwardly inferential reasons—for thinking them to be true. The producer-consumer distinction applies to all philosophical thought experiments. I think that if we are interested in the methodology of philosophical thought experimentation we should focus our investigations on the producer’s side of the producer-consumer divide, for it is that focus that will lead to insight into how thought experimentation can yield genuine philosophical knowledge. A good thought experiment, such as Gettier’s, can lead to real discoveries, but it leads there, when it does, because of the effort that its producer/inventor puts into developing it. How consumers of a thought experiment consume it is independent, it seems to me, of whether it yields genuine knowledge. Some consumers of a philosophical thought experiment will have intuitions related to it. If I am right about the methodological irrelevance of consumers’ psychological reactions or attitudes to a thought experiment, generally speaking, then consumers’ intuitions, in particular, are irrelevant. The fairly widespread belief that they are relevant can be chalked up to focusing, inappropriately and misleadingly, on the consumer’s side of the producer-consumer divide.



Gettier’s Method

87

6. Textbook renditions Gettier’s argument against the JTB theory is justly famous and rightly taken to be a modern classic of epistemology. Renditions of it therefore show up in almost every competent textbook introduction to the field. If one is interested in discovering whether intuitions are appealed to as evidence against the JTB theory in Gettier-style arguments, should one examine not only Gettier’s original paper, but these textbook renditions as well? The issue, in a more general form, comes up in Herman Cappelen’s Philosophy Without Intuitions (2012). Cappelen canvasses a wide range of philosophical arguments and finds no evidential appeals to intuitions in any of them. But he looks only at the original texts in which these arguments were first presented. He has been criticized for this by David Chalmers (2014), who suggests, in a commentary on Cappelen’s book, that this pursues the question of the role of intuitions in philosophy in a blinkered way, since it ignores the role that intuitions might play in textbook representations of the relevant arguments. In registering some “preliminary worries” about Cappelen’s strategy, Chalmers writes: Second and more important, why the original texts? Cappelen stresses a number of times that these texts often differ in surprising ways from the way these arguments are described by later philosophers and by textbooks. But insofar as there is a difference here it is presumably the arguments as represented in later texts and textbooks that are most influential and most central to the practice of philosophy. My suspicion is that an examination of textbook presentations would not be nearly so friendly to Cappelen’s case against [the claim that intuitions are treated as evidence in philosophy].21 (Chalmers 2014: 539)

Is Chalmers right? A problem for his view is that textbook presentations of philosophical arguments might be not just different from the original presentations but wrong. That is, they can misrepresent the arguments they seek to summarize or explain. This is what Cappelen stresses when he warns his readers to avoid textbook and other secondary literature versions of the arguments he canvasses. His worry is that these might seriously misrepresent the arguments in question. not just be different from them. If a textbook describes an important philosophical argument as appealing to a kind of evidence to which the argument does not, in fact, appeal, then

88

Advances in Experimental Philosophy and Philosophical Methodology

the textbook description is simply mistaken; it says the argument relies on evidence that it does not rely on. In such a case, it would be bizarre to say that the textbook’s mistake might nevertheless show, or contribute to showing, that that kind of evidence is regularly appealed to in philosophy. Imagine that many philosophy textbooks mysteriously say that philosophical arguments rely on astrological evidence. Obviously, the textbooks’ saying so does not somehow make it true that the arguments really do rely on astrological evidence. And no one would seriously suggest that the textbooks’ mistaken ascription of a method of appealing to astrological evidence is itself an application of that very method and, hence, that philosophy, insofar as it is practiced in the writing of philosophy textbooks, really does rely on astrological evidence. What we can glean from textbooks about philosophical methods depends on the textbooks’ accuracy. If textbooks present this or that argument in philosophy as relying on intuitions, but those arguments do not in fact rely on intuitions, then the textbooks offer no support at all for the view that intuitions play an important role in the methods of philosophy. But how do we tell whether a textbook presentation of a philosophical argument is accurate? By consulting the original text of course—what else? So, Cappelen is right, it seems to me, to insist that, as philosophical methodologists, we should look mainly to the original presentations of philosophical arguments, for these original presentations are really our only direct evidence about the methods philosophers employ. For the most part, philosophy textbooks’ assertions are assertions of descriptive metaphilosophy, and, as such, can be just as inaccurate as other examples of descriptive metaphilosophy. Cappelen says, and I wholeheartedly agree, that experimental philosophers, for example, predicate their critique of philosophy on a misdescription of philosophy’s methods. Unfortunately, nothing prevents philosophy textbook writers from making exactly the same mistake. Of course, later representations of an argument, whether accurate or not, can be very influential, and Chalmers is surely right to say that later texts and textbooks might be responsible for the way in which the philosophical community as a whole views and understands an argument. That is close to undeniable, I think, but I also think that, as a piece of philosophical sociology, the point does not really speak to any central methodological issue. All sorts of factors influence the ebb and flow of philosophical belief in the philosophical



Gettier’s Method

89

community, and the exposure that certain representations of an argument get is certainly one of these. But it is still true, despite this, that if the representations are misrepresentations, even widely accepted misrepresentations, then they will not reveal anything about the true workings of philosophy. I suspect that, by now, the reader can guess what I am going to say about the methodological relevance of textbook renditions of, specifically, Gettier’s argument against the JTB theory. Since, as I have argued, Gettier does not treat intuitions as evidence in his argument, any textbook rendition that says otherwise misrepresents the way in which Gettier argues against the JTB theory, and should be taken no more seriously as an account of Gettier’s method than a rendition that claimed that Gettier makes an implicit appeal to astrological evidence. Here is an interesting twist, however: While I gave examples in section 1 of descriptions of Gettier’s argument that do claim that it depends on evidential appeals to intuitions, and while the raison d’être of this paper is to stem the tide of these sorts of mischaracterizations, consulting textbook renditions of Gettier’s argument reveals far less of this kind of mischaracterization than recent metaphilosophical discussions of Gettier’s method would lead one to expect. In fact, only one of the eight textbooks I consulted (including a “handbook” and a “companion”) characterized Gettier’s argument as relying on intuitions.22

7. Conclusion The argument in Gettier 1963 is often regarded as a paradigmatic example of a philosophical argument that applies the method of cases, a method held to essentially involve evidential appeals to intuitions. I have argued that regarding it as such is a mistake. I have also argued that there is no support for the view that Gettier treats intuitions as evidence, either from examining examples from the Gettierology literature that Gettier’s paper spawned, or the from the thought that intuitions might appear “earlier” in Gettier’s reasoning than with respect to judgments about his hypothetical cases themselves. Likewise, I have claimed that the presence of abductive reasoning in Gettier’s argument, and possible textbook renditions of the argument that represent it as applying

90

Advances in Experimental Philosophy and Philosophical Methodology

the method of cases, do not make any more likely the view that it really does involve an application of that method. I have also urged that cognizance of the producer-consumer distinction, as it applies to cases and thought experiments, sheds considerable light on how we ought to think of Gettier’s method. My hope is that another long paper on one of the shortest papers in analytic philosophy’s history will correct several serious misconceptions about Gettier 1963, and set future methodological investigations on a straighter path.23

Appendix: Gettier’s original cases The 10 Coins Case Suppose that Smith and Jones have applied for a certain job. And suppose that Smith has strong evidence for the following conjunctive proposition: (d) Jones is the man who will get the job, and Jones has ten coins in his pocket.

Smith’s evidence for (d) might be that the president of the company assured him that Jones would in the end be selected, and that he, Smith, had counted the coins in Jones’s pocket ten minutes ago. Proposition (d) entails: (e) The man who will get the job has ten coins in his pocket.

Let us suppose that Smith sees the entailment from (d) to (e), and accepts (e) on the grounds of (d), for which he has strong evidence. In this case, Smith is clearly justified in believing that (e) is true. But imagine, further, that, unknown to Smith, he himself, not Jones, will get the job. And, also, unknown to Smith, he himself has ten coins in his pocket. Proposition (e) is then true, though proposition (d), from which Smith inferred (e), is false. In our example, then, all of the following are true: (i) (e) is true, (ii) Smith believes that (e) is true, and (iii) Smith is justified in believing that (e) is true. But it is equally clear that Smith does not know that (e) is true; for (e) is true in virtue of the number of coins in Smith’s pocket, while Smith does not know how many coins are in Smith’s pocket, and bases his belief in (e) on a count of the coins in Jones’s pocket, whom he falsely believes to be the man who will get the job (Gettier 1963: 122).



Gettier’s Method

91

The Brown in Barcelona Case Let us suppose that Smith has strong evidence for the following proposition: (f) Jones owns a Ford.

Smith’s evidence might be that Jones has at all times in the past within Smith’s memory owned a car, and always a Ford, and that Jones has just offered Smith a ride while driving a Ford. Let us imagine, now, that Smith has another friend, Brown, of whose whereabouts he is totally ignorant. Smith selects three place names quite at random and constructs the following three propositions: (g) Either Jones owns a Ford, or Brown is in Boston. (h) Either Jones owns a Ford, or Brown is in Barcelona. (i) Either Jones owns a Ford, or Brown is in Brest-Litovsk.

Each of these propositions is entailed by (f). Imagine that Smith realizes the entailment of each of these propositions he has constructed by (f), and proceeds to accept (g), (h), and (i) on the basis of (f). Smith has correctly inferred (g), (h), and (i) from a proposition for which he has strong evidence. Smith is therefore completely justified in believing each of these three propositions, Smith, of course, has no idea where Brown is. But imagine now that two further conditions hold. First Jones does not own a Ford, but is at present driving a rented car. And secondly, by the sheerest coincidence, and entirely unknown to Smith, the place mentioned in proposition (h) happens really to be the place where Brown is. If these two conditions hold, then Smith does not know that (h) is true, even though (i) (h) is true, (ii) Smith does believe that (h) is true, and (iii) Smith is justified in believing that (h) is true (Gettier 1963: 122–3).

Notes   1 A recent example is Anna-Sara Malmgren’s (2011) articulation and defense of the method of cases.   2 For the most part, I object in this paper to the idea that Gettier 1963 involves an application of the method of cases. But I do this by objecting to the idea that

92

Advances in Experimental Philosophy and Philosophical Methodology

intuitions specifically about Gettier’s hypothetical cases themselves are appealed to in his argument against the JTB theory. An argument might fail to involve an application of the method of cases and yet involve intuitions in some other way. With one exception having to do with whether intuitions are involved at an earlier stage of Gettier’s reasoning than the stage at which he arrives at his judgments about his hypothetical cases, I ignore this possibility. Thanks to Jenny Nado for asking me to be clearer about my target.   3 Construals of the method of cases differ from philosopher to philosopher. Here, I focus on construals that describe it as a method meant to reveal evidence against a principle or theory. I ignore the issue of whether and how such construals cast the method as one that can provide positive evidence in favor of a principle or theory. In other words, I am interested in those construals that cast the method of cases as a way of revealing, or at least justifying belief in the existence of, counterexamples to a theory. There is also an issue of the strength of the evidence that a conflicting “intuitive verdict” provides against a given theory. Malmgren’s construal takes conflicting intuitive verdicts to be treated as “strong” evidence against a theory, for example. My view is that, at least as a description of Gettier’s method, intuitive verdicts are not treated as any evidence whatsoever. Hence, even a very weak version of the method of cases, one that says that intuitive verdicts are treated only as some small amount of “prima facie” evidence against a theory, for example, is false of Gettier’s method. Or so I will argue.   4 I do not have a theory of the “nature” of intuitions. But I do not think I need one. It is clear that anyone who thinks that Gettier is applying the method of cases in his argument against the JTB theory thinks that Gettier treats intuitions or intuitive verdicts about hypothetical cases as evidence against the JTB theory. I think, and will argue that, this is wrong, regardless of what intuitions are, and regardless of what anyone takes them to be. For example, it is irrelevant to my concerns whether one’s conception of intuitions takes them to be “thick” or “thin.” (See Weinberg and Alexander 2014.) Let them be as thick or thin as you like. Still, Gettier does not treat them as evidence for his judgments about his cases.   5 The Brown in Barcelona Case, in its original presentation (see this chapter’s appendix), involves Smith failing to know three distinct propositions: that Jones owns a Ford or Brown is in Boston, that Jones own a Ford or Brown is in Barcelona, and that Jones owns a Ford or Brown is in Brest-Litovsk. However, only the second of these is true in Gettier’s story, and only it, therefore, makes the anti-JTB point. The three different propositions about the three different places



Gettier’s Method

93

serve the purpose of emphasizing the randomness by which Smith selects the places and place-names in formulating the relevant propositions. I will return to this randomness and its significance soon.   6 This simple, informal account of how Gettier takes (G1) and (G2) to lead rather directly to the anti-JTB conclusion is not endorsed by most of the recent work on the “logical form” of Gettier’s argument. Examples include Williamson 2007, Ichikawa and Jarvis 2009, Malmgren 2011, and Pust 2014. Most of this work is based on the belief, mistaken in my view, that there is a “content problem,” as Malmgren puts it, with Gettier’s crucial premise, the premise, that is, that is meant to conflict with the JTB theory. (Malmgren 2011: 271) The content problem is the (alleged) problem that, since the name, “Smith,” in Gettier’s stories, is a non-referring term, it is incapable of figuring in statements of the literal truth. For example, given that “Smith” fails to refer, (G1) and (G2) are not even true, let alone such that their truth conflicts with the JTB theory. On my view, this is a pseudo-problem. Gettier never intended us to assert or believe claims such as (G1) and (G2). Rather, as he explicitly indicates, by prefacing the presentations of his cases with “suppose that,” these propositions are to be treated as the contents of suppositions. It is by seeing that that these suppositions follow from other suppositions (in conjunction with a certain epistemic principle that I will discuss later, and which Gettier takes to be literally true), whose contents are clearly metaphysically possible, that we are led to conclude that these suppositions too are metaphysically possible. But from that—namely the claim that (G1) and (G2) are possible, and consistent with the claim that Smith, in each case, justifiably and truly believes—the anti-JTB conclusion follows immediately. Needless to say, this idea needs more fleshing out, which I hope to give it in future work. For now it will have to suffice to note that, on my view, much of the recent work on the logical form of Gettier’s argument is based on a mistake concerning Gettier’s method.   7 Most of the recent metaphilosophical work on Gettier’s method takes the relevant question here to be what justifies the judgment that Smith has “non-knowledge justified true belief.” I think, rather, that Gettier comes close to stipulating that Smith has a justified true belief in each case.   8 It should be borne in mind that Gettier takes (G1) and (G2) to be true in the cases, rather than true simpliciter. The cases are purely hypothetical, after all. In my later discussion, I will speak loosely of “the truth of (G1) and (G2)” and Gettier “arguing for the truth of (G1) and (G2).” But Gettier does not take (G1) and (G2) to be literally true, and the argument he gives is an argument that they are true, granted earlier suppositions, and in the cases. See note 6, above, for related clarifications.

94

Advances in Experimental Philosophy and Philosophical Methodology

  9 Joshua Alexander suggests to me that it is not the mere possibility of such an appeal that leads some to say Gettier is, at least in part, appealing to intuitions in favor of (G1) and (G2). Rather, it is a principle of charity—something along the lines of: “Gettier needs something to get his anti-JTB argument off the ground, so, probably, it is intuitions.” Three points in reply: 1. I do not think this is very charitable, especially not coming from experimental philosophers, who are prone to “charitably” ascribe the method, and then beat up on those to whom they have ascribed it. 2. Even if genuine charity genuinely motivates some such ascriptions, it’s entirely unnecessary, since Gettier provides an argument for (G1) and (G2). 3. I think it would be unnecessary anyway, since every argument has a first premise, a starting point. No one thinks that all arguments require charitable ascriptions of this or that justificatory source for their first premises. So why think Gettier is especially in need of charity here? 10 Also, cases in which one has both an argument for p and an intuition that p are, arguably, cases in which one should not appeal to the intuition as additional evidence. For, if the argument causes the intuition, then it seems as though the intuition adds no more justification to belief in p than that provided by the argument. Many intuitions in philosophy are caused by accompanying arguments, including many that are intuitions about cases or thought experiments. Hence, these intuitions should not be treated as evidence. I think intuitions about Gettier’s cases are caused by the explicit argument he gives for their contents (and to which I will turn, shortly, in the main text). If I’m right, then there are grounds for saying that intuitions about Gettier’s cases should not be treated as additional evidence for their truth. That is not quite the conclusion that they are not treated as such, but it points in that direction. It is plausible to suppose that smart philosophers like Gettier do as they ought to, methodologically speaking. I develop this idea and several other “philosophers shouldn’t appeal to intuitions, so they don’t”-style arguments in an as yet unpublished paper, “The Methodological Irrelevance of Intellectual Seemings.” 11 The word “luck” does not appear in Gettier’s presentations. By what right, then, do I claim that an anti-luck argument for (G1) and (G2) is explicit in them? First, “luck”-like terminology is used—for example, “sheerest coincidence.” More importantly, in both cases, Gettier stresses the fact that Smith’s beliefs are true for a reason that is independent of the reason on which he “bases” the belief. He also stresses that Smith fails to know, indeed, is “entirely ignorant” of, the bases on which the beliefs are true. These features of the cases, which Gettier takes pains to stress, are features that suffice, it seems to me, to make Smith’s beliefs merely luckily true. So the anti-luck argument for (G1) and (G2)



12 13

14

15

16 17

18

19

20

Gettier’s Method

95

is no reconstruction or post hoc explanation on my part; it is explicitly there in Gettier’s text. Thanks to Jenny Nado and Joshua Alexander for making me say more about this issue. Timothy Williamson (2002) is the most prominent recent proponent of the view that knowledge cannot be analyzed. There are exceptions to this pessimism about Gettierology. William Lycan (2006) has written a highly entertaining and interesting defense of the view that such pessimism is unwarranted. The examples taken to demonstrate the insufficiency of Clark’s conditions are usually perceptual cases (such as Chisholm’s 1966 sheep-in-the-field case) that appear not to involve any inference at all, and so no inference from a false lemma. More interestingly, there are plenty of examples of inferences from only true lemmas to justified, true, yet non-knowledge, belief. Lycan (2006) provides a nice, quick summary of many of these. (See Lycan 2006: 144.) Cases that show this are cases in which a believer has more than one line of justification for his or her true belief, only some of which are “Gettiered.” See Saunders and Champawat (1966) and Lehrer (1965) for examples. This is also the view taken earlier by Joshua Alexander in Alexander (2012). This simplifies Brogaard’s reasoning somewhat but gets at the essence of it, I think. Officially, Brogaard’s test for implicit appeal to intuition in support of a claim, p, is that p not be “explicitly inferred from other premises, argued for in previous publications, or explicitly treated as an assumption.” Also, this is only one of two such tests she presents in her paper, which is a contribution to a symposium on Herman Cappelen’s (2012) Philosophy without Intuitions. Of course, there is a difference between what Gettier takes to be his evidence for the anti-luck premise and what, if anything, is in fact the evidence for it. In general, one might be wrong about what justifies one’s beliefs or fail to have any articulated or reflective account of why one’s beliefs are true. So I grant that it is possible that intuition is in some sense “relied on” at some stage in Gettier’s overall argument against the JTB theory, even if Gettier never takes intuitions as evidence for anything. But why think that this possibility has any distinctively methodological (as opposed to epistemological) significance? Thanks to Jenny Nado for asking that I consider this issue. Though perhaps this is too concessive. As Cappelen (2012) points out, that a claim is taken as a premise in an abductive argument says nothing at all about how, or even whether, the claim is meant to be justified. Or he might have meant the order of explanation to go just one way, from the anti-luck condition non-abductively to the truth of (G1) and (G2).

96

Advances in Experimental Philosophy and Philosophical Methodology

21 This is a surprising comment coming from a philosopher whose “zombie argument” against materialism is routinely misrepresented as “just an intuition”! 22 My experimental philosophical study included An Introduction to the Theory of Knowledge (2007), by Noah Lemos; Epistemology (2006), by Richard Fumerton; Epistemology (2003), by Richard Feldman; Epistemology: A Contemporary Introduction to the Theory of Knowledge (2011, 3rd edn), by Robert Audi; What is this thing called Knowledge? (2013), by Duncan Pritchard; The Oxford Handbook of Epistemology (2002), edited by Paul Moser; The Blackwell Companion to Epistemology (2010, 2nd edn) edited by Jonathan Dancy, Ernest Sosa, and Matthias Steup; and Epistemology: Classic Problems and Contemporary Responses, by Laurence Bonjour. Only the last of these makes use of “intuition” and “intuitively” in characterizing Gettier’s argument. Perhaps it can be fairly said to cast the argument as making evidential appeals to intuitions. Fumerton’s book describes the JTB theory’s verdict about Gettier’s cases as “counterintuitive,” but otherwise paints the picture as one whereby Gettier argues for (G1) and (G2) based on considerations to do with epistemic luck. Indeed, this is a common theme throughout many of the presentations, none of the rest of which mentions intuitions at all, let alone assigns some important evidential role to them. 23 Thanks to Joshua Alexander and Jennifer Nado for incisive criticism of an earlier draft. Thanks also to Dan Marshall for help with something I was struggling to understand while writing this paper. Participants at the Huaqiao University Philosophy of Mind Conference in June 2015 offered valuable feedback on a talk based on this paper. I thank them for this. The research in this paper was supported in part by a General Research Fund grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. HKU743910H).

References Alexander, J. (2012), Experimental Philosophy: An Introduction. Cambridge: Polity Press. Alexander, J. and Weinberg, J. (2007), “Analytic Epistemology and Experimental Philosophy,” Philosophy Compass 2 (1): 56–80. Bealer, G. (1996), “A Priori Knowledge and the Scope of Philosophy,” Philosophical Studies 81 (2–3): 121–42. Bengson, J. (2014), “How Philosophers Use Intuition and ‘Intuition’,” Philosophical Studies 171 (3): 555–76.



Gettier’s Method

97

Brogaard, B. (2014), “Intuitions as Intellectual Seemings.” Analytic Philosophy 55 (4): 382–93. Cappelen, H. (2012), Philosophy without Intuitions, Oxford: Oxford University Press. Chalmers, D. (2014), “Intuitions in Philosophy: A Minimal Defense,” Philosophical Studies 171: 535–44. Chisholm, R. (1966), The Theory of Knowledge, Upper Saddle River: Prentice Hall. Clark, M. (1963), “Knowledge and Grounds,” Analysis 24: 46–8. Deutsch, M. (2010), “Intuitions, Counter-examples, and Experimental Philosophy,” Review of Philosophy and Psychology 1 (3): 447–60. Devitt, M. (forthcoming), “Relying on Intuitions: Where Cappelen and Deutsch Go Wrong,” Inquiry. Gettier, E. (1963), “Is Justified True Belief Knowledge?” Analysis 23: 121–3. Goldman, A. (1967), “A Causal Theory of Knowing,” Journal of Philosophy 64: 357–72. Ichikawa, J. and Jarvis, B. (2009), “Thought-experiment Intuitions and Truth in Fiction,” Philosophical Studies 142 (2): 221–46. Lehrer, K. (1965), “Knowledge, Truth, and Evidence,” Analysis 25: 168–75. Lycan, W. (2006), “On the Gettier Problem Problem,” in S. Herrington (ed.), Epistemology Futures, Oxford: Oxford University Press, 148–68. Mallon, R. (this volume), “Intuitive Diversity and Disagreement.” Malmgren, A. (2011), “Rationalism and the Content of Intuitive Judgements,” Mind 120 (478): 263–327. Nado, J. and Johnson, M. (this volume), “Theories of Reference: A Case Study for the Role of Intuition in Philosophy.” Nagel, J. (2012), “Intuitions and Experiments: A Defense of the Case Method in Epistemology,” Philosophy and Phenomenological Research 85 (3): 495–527. Neta, R. (2012), “Knowing from the Armchair that our Intuitions are Reliable,” Monist 95 (2): 329–51. Pust, J., “Intuition,” E. Zalta (ed.), The Stanford Encyclopedia of Philosophy. Available from http://plato.stanford.edu/archives/fall2014/entries/intuition/ [accessed June 1, 2015]. Saunders, J. and Champawat, N. (1964), “Mr. Clark’s Definition of “Knowledge,” Analysis 25: 8–9. Weinberg, J. (2014), “Cappelen Between a Rock and a Hard Place,” Philosophical Studies 171 (3): 545–53. Weinberg, J. and Alexander, J. (2014), “Sticking with Intuitions Through Thick and Thin,” in A. Booth and D. Rowbottom (eds), Intuitions, Oxford: Oxford University Press, 187–212. Williamson, T. (2002), Knowledge and its Limits, Oxford: Oxford University Press. Williamson, T. (2007), The Philosophy of Philosophy, New York: John Wiley & Sons.

6

Intuitive Diversity and Disagreement1 Ron Mallon

One important strand of experimental philosophy, so called “negative” experimental philosophy, was born in part as a response to a philosophical method that seems to constrain philosophical theory by an appeal to one’s spontaneous or intuitive judgments about actual and possible cases. On the most extreme account of this method, philosophers begin, aseat in some armchair, consulting their a priori intuitions (and those of their friends) about sometimes mundane and sometimes outlandish cases for insight into abstract, general truths about matters of philosophical interest. Call this picture the platonic armchair. While there are now numerous critiques of experimental philosophy, this paper is concerned with a specific subset of them: a subset that allows that the platonic armchair would be bad method but suggests that this method is not widely employed in the profession of philosophy (Goldman 2010; Williamson 2007, 2013; Cappelen 2012; Deutsch 2009, 2010). These theorists, in effect, charge experimental philosophers, and sometimes others, with misinterpreting or misunderstanding the method exhibited in paradigm instances of philosophy, and they offer alternative interpretations of the method employed in philosophers’ appeals to cases. In this chapter, I consider one core experimental philosophical challenge, the argument from diversity of intuition, and I use it to argue that experimental philosophical work poses a challenge to philosophical method not only on the platonic armchair understanding of that method but on a number of alternative interpretations. Along the way, I consider and reject the increasingly influential suggestion that philosophers have

100

Advances in Experimental Philosophy and Philosophical Methodology

not actually appealed to intuitions at all. The upshot of my argument will be that at least one version of the challenge posed by experimental philosophy persists in the face of a range of different construals of philosophical method.

1. What are intuitions? As a first pass, by “intuitions” I mean the sorts of propositional attitudes that feature as responses to cases in philosophical discussions. Since they are generally understood to be some sort of propositional attitude, they are individuated (in part) by their content. But there is some disagreement about how to understand the attitudes involved. Some philosophers understand intuitions as mere seemings that p while others understand them as judgments or beliefs that p. One plausible possibility is that when other things are equal, seemings-that-p give rise to believings-that-p. On this possibility, because both have one represent p as true, both raise epistemic questions as to the justification for p. As Timothy Williamson has rightly pointed out, there is no broad consensus as to how to answer this question, no “accepted explanation of the hoped-for correlation between our having an intuition that P and its being true that P” (Williamson 2007: 215). In fact, as we see below, not all uses of intuition need rely on this correlation. In addition, we can add that most philosophers have treated “intuition” (as we do here) as picking out noninferential mental states: states that are not the immediate product of a conscious chain of reasoning, but, rather, simply spontaneously arise in one’s mind in response to some circumstance or case. Negative experimental philosophy has been directed at challenging the use of the contents of such noninferentially arrived at attitudes as premises in philosophical arguments, and the argument from diversity offers one such challenge. Here, I use “intuition” to pick out the sorts of seemings or judgments involved in our target cases, but also, more loosely, for the behavioral manifestations of those judgments produced in response to philosophical thought experiments and experimental surveys.2



Intuitive Diversity and Disagreement

101

2. Diversity in philosophical intuitions and disagreement Some of the very first papers in experimental philosophy drew our attention to apparent cultural diversity in intuitive judgments about cases based closely on vignettes figuring in classical philosophical discussion. Inspired by evidence of cross-cultural psychological diversity from the psychologist Richard Nisbett, philosophers Jonathan Weinberg, Shaun Nichols, and Stephen Stich (2001) sought and found evidence of systematic differences in response to classic epistemological thought experiments between cultural East Asians and Westerners among New Jersey undergraduates, and also variation by socioeconomic status. Following on from this work, Edouard Machery, Shaun Nichols, Stephen Stich, and myself (2004) looked for, and found, evidence of diversity in judgments of the reference of proper names among Hong Kong and American undergraduates. More recent work has suggested other sorts of diversity in responses to philosophical thought experiments. For instance, Adam Feltz and Edward Cokely (2008) have found that judgments about moral objectivism vary with possession of the personality trait openness to experience, and responses to questions regarding free will and moral responsibility vary with extroversion (Cokely and Feltz 2009). It would be hard to emphasize enough that these empirical results are preliminary, and each of them could be further supported, but also overturned by further, more careful experiments. For instance, some investigators have been unable to replicate Weinberg et al.’s initial results (Nagel 2012; Nagel et al. 2013; Seyedsayamdost 2015); and others have explained these results as an unintended artifact of the experimental materials or situation (e.g. Nagel 2012; Cullen 2010; Turri 2013). Similarly, Justin Sytsma and Jonathan Livengood (2011) failed to replicate Machery et al. (2004), although subsequent work by them and others has resulted in additional evidence for the original hypothesis of diversity of intuitions about reference (Sytsma et al. 2015; Machery et al. 2010; Seyedsayamdost 2015). Noting these complications, for now I set them aside. My present focus is not the status of the empirical debate but the significance of such diversity in responses to vignettes for philosophical methodology. Thus, for present

102

Advances in Experimental Philosophy and Philosophical Methodology

purposes, I simply assume that these studies show systematic diversity in intuitive judgments about philosophical vignettes by the folk, and ask about the significance of such reports. One way of drawing out the significance of intuitive diversity lies in appreciating the implications of systematic disagreement in judgments among apparent “epistemic peers.” When confronted with genuine disagreement with others over whom we take ourselves to enjoy no epistemic advantage (differing only in, say, culture or personality trait), we ought to reduce our certainty, or perhaps even withdraw belief altogether in the matter of disagreement. Experimental philosophy studies are plausibly interpreted as concerning classes of epistemic peers since the subject populations vary by demographic factors that we have no prior reason to suspect confer epistemic authority. This being so, they are cases in which, if other things are equal, the disagreement of others should lead one to doubt the truth of one’s own judgment, provided one falls into one of the classes, and exhibits the class-typical judgment. It could turn out that cultural East Asians or Westerners have, for cultural reasons, special insight into some philosophical matters, or that, for whatever reason, extraverts or introverts have systematically confused thinking about human freedom. Some evidence of cultural differences, for instance, suggests that East Asians are more susceptible to hindsight bias but less prone to the “fundamental attribution error,” while Western subjects are prone to complementary “hindsight” biases in the opposite direction (Nisbett 2003: 123ff., 130ff.). This phenomenon is not surprising, since culturally transmitted epistemic strategies may encode various shortcuts, and the costs and benefits of those shortcuts (and thus those strategies) will manifest differently in different cases. Still, in advance of any evidence for holding a specific view about who is in error, we are left with the suspicion that it may be us. It is tempting to object here, as many have, that experimental philosophical surveys do not reveal disagreement among epistemic peers of philosophers, since such surveys typically examine the judgments of nonphilosophers. Thus, a substantial literature has developed on the extent to which philosophical expertise is possible, and the extent to which it plays a role in philosophical method concerning the relevant cases (e.g. Weinberg et al. 2010; Machery 2012). For the most part, I leave this debate aside in what follows. The critiques of experimental philosophy that I target here do not suggest that



Intuitive Diversity and Disagreement

103

philosophers rely on expert intuition of cases, but rather that they do not rely on Platonic armchair intuition at all. Even if we see the central issue as one of disagreement among epistemic peers, exactly when and to what extent we ought to respond to such disagreement is a substantial question of contemporary epistemology. At one end of the spectrum, conciliationists suggest that we are required to reduce our credence, and perhaps even suspend our beliefs, when faced with disagreement. And, at the other end, steadfast theorists argue that we are not so required.3 The relevance of experimental philosophy’s argument from disagreement thus requires, at a minimum, the plausible view that intuitive disagreement of the sort experimentalists have uncovered requires one to lower confidence, at least a bit, in the correctness of one’s own judgment. I take it that this view is one that even many steadfast theorists would accept.4 To this we can add that various positions on this conciliationist-steadfastness spectrum may also take into account the sort of evidence for the judgment in question. For instance, steadfast theorists sometimes emphasize cases involving carefully reasoned and reviewed responses, suggesting that these may merit more steadfastness than a snap judgment. I return to this point below. For now, I note that treating the argument from diversity of intuitions as an argument from disagreement makes sense of what many of its proponents say and offers reasons for denying that intuition can play the role that the platonic armchair picture assigns to it. But, as we noted at the outset, some critics of experimental philosophy are also critics of the Platonic armchair picture. How does it fare against these critics?

3. The mentalist move to conceptual analysis One reconstruction of philosophical practice that does not depend upon the Platonic armchair is itself naturalist in orientation. On this view, we ought to understand intuitions primarily as revealing facts about human concepts, conceptions, or other psychological mechanisms that produce intuitions that p (Goldman and Pust 1998; Jackson 1998; Goldman 2010). Many philosophers are attracted to such a mentalist position because of broadly

104

Advances in Experimental Philosophy and Philosophical Methodology

naturalistic worries about how philosophers could, from their armchairs, have access to broad and general facts about properties in the world such as knowledge, freedom, and so forth. Goldman and Pust (1998), for instance, write of the idea that intuitions might give knowledge of abstract, universal properties: Is there any reason to suppose that intuitions could be reliable indicators of a universal’s positive and negative instances (even under favorable circumstances)? The problem is the apparent “distance” or “remoteness” between intuitions, which are dated mental states, and a nonphysical, extra-mental, extra-temporal entity. How could the former be reliable indicators of the latter? (Goldman and Pust 1998: 185)

This passage concerns knowledge of universals, properties of the sort that might be associated with the Platonic armchair picture with which we started. But Goldman and Pust go on to argue that such distance obtains with regard to a range of extra-mental interpretations of the objects of intuitive knowledge: all exhibit such “distance” or “remoteness” from our armchairs. Once we allow that substantive knowledge of the world is not trivial to get, then we must demand some explanation for how we come to have it. In contrast, the idea that a possible target of intuitive knowledge is the concepts, theories, or other psychological features that produce them is eminently plausible: Although we do not currently know the precise causal route that connects concept structures with their conscious manifestations, it is extremely plausible, from any reasonable cognitive-science perspective, that there should be such a causal route. (Goldman and Pust 1998: 188)

Frank Jackson emphasizes a similar approach in his defense of conceptual analysis: Intuitions about how various cases, including various merely possible cases, are correctly described in terms of free action, determinism, and belief are precisely what reveal our ordinary conceptions of free action, determinism, and belief, or, as it is often put nowadays, our folk theory of them. (1998: 31)

This mentalist approach thus abandons the Platonic armchair’s commitment to knowledge of abstract, general truths, but also abandons more mundane



Intuitive Diversity and Disagreement

105

extra-mentalist interpretations of the armchair, by reinterpreting the objects of intuitive knowledge in a naturalistically respectable, psychological way. Advocates of this approach emphasize that a justification for taking intuitions that p as revealing facts about the mechanism that produced such intuitions seems easy to come by: it is just an instance of the more general practice of inferring the nature of a psychological mechanism from the thought and behavior it produces. A second advantage of this approach, but not one emphasized by advocates, is that once we take the object of knowledge to be the theory, concept, representation, or other mechanism that produces our intuitive response to a case, we can avoid commitment to the claim that intuitions that p correlate with the truth of p.5 Intuitions that p can be evidence for mental mechanisms whether or not p is true or false. Consider, for instance, one of Gettier’s famous cases (1963). Here, we read the case and judge that: J. Smith does not know.

J, together with the stipulation S that, S. Smith does have justified true belief.

Entails the metaphysical conclusion that: MC. Justified true belief is not sufficient for knowledge.

This argument is deductively valid, but it is only sound if J is true, and thus J’s epistemic status becomes central. However, once we allow that the mental event of thinking J also allows us to judge: P: I believe that J.6

We can then treat this reflective judgment as evidence for a psychological claim about the psychological mechanism or capacity that gave rise to J. For instance, we might infer: PJ. According to our concept or conception of knowledge, Smith does not know.

And from that infer:

106

Advances in Experimental Philosophy and Philosophical Methodology

PC. According to our concept or conception of knowledge, justified true belief is not sufficient for knowledge.

In contrast to the metaphysical argument leading to MC above, this argument does not require that J be true. Despite these appealing features, many philosophers reject the mentalist interpretation of philosophical practice. For one thing, as Williamson argues, much philosophical practice seems more ambitious: it seems to interpret intuitions that p as evidence for the truth of the content p even when p concerns extra-mental matters of fact. Mentalists would, then, likely need to reform or reconstruct much existing philosophical practice, and mentalism could not be a defense of existing method. For another thing, it is not always clear how to separate the evidence that is relevant to the concept or mechanism from that which is mere noise (Mallon 2007; Alexander et al. 2010). But, even if we allow that a mentalist interpretation of the appeal to philosophical intuition is correct, the argument from disagreement undercuts the armchair. This is because philosophers (including Goldman, Pust, and Jackson) are not particularly interested in a particular individual’s mental concepts or mechanisms, but concepts or mechanisms that are shared among many people. Here’s how Jackson puts it: If I say that what I mean—never mind what others mean—by a free action is one such that the agent would have done otherwise if he or she had chosen to, then the existence of free actions so conceived will be secured, and so will the compatibility of free action with determinism. … I have turned interesting philosophical debates into easy exercises in deductions from stipulative definitions together with accepted facts. … What we are seeking to address is whether free action according to our ordinary conception, or something suitably close to our ordinary conception, exists and is compatible with determinism. (1998: 31)

It falls out of this conception that, despite the plausible considerations that may favor mentalism over extra-mentalist interpretations of armchair philosophy, mentalists are also primarily interested in facts that seem distant from what can be observed simply on the basis of reflection upon one’s own beliefs or mind. They are interested in shared or common concepts, theories or other mental mechanisms that produce judgments, and whether these are



Intuitive Diversity and Disagreement

107

shared or common is itself a fact that may not be discernible from the armchair. This distance undercuts the presumed epistemic advantage that the mentalist enjoys over the extra-mentalist in virtue of the plausible connection between intuitions and the mechanisms that produce them (Mallon forthcoming). Nonetheless, it would be wrong to fall into skepticism about the success of the mentalist turn from these considerations alone. Brian Talbot (2014) argues convincingly (for reasons similar to those that generate the Condorcet Jury Theorem) that, given a range of plausible assumptions about the reliability of one’s own and others’ intuitions, aggregating even the moderately reliable intuitions of oneself and others can give rise to very reliable judgments about the subject matter (see Goldman 2010 for similar suggestions along these lines). In a different vein, Jennifer Nagel (2012) has offered evidence that, in some cases involving language, the confidence of intuiters is a good indicator of the judgments that other subjects in the pool will make (even where the intuitions were not themselves true). And Billy Dunaway, Anna Edmonds, and David Manley have offered evidence that philosophers are pretty good at predicting folk beliefs that experimental philosophers have claimed to be surprising (2013). While these considerations offer reason to think that we can achieve some knowledge of the intuitions of folk like ourselves (and perhaps in turn the mechanisms that generate the intuitions), they offer little succor against the prospect of systematically differing intuitions among people who are, in some systematic respect, not like us. Indeed, they run the risk of treating such disagreement as noise, rather than as the indicator of systematic diversity that it is. Once we recognize the existence of such systematic diversity, abandoning the platonic armchair for the psychologist’s or sociologist’s armchair will not be enough. Such diversity in intuitive judgments may be an indicator that multiple concepts or mechanisms are in play, and this may indicate that our philosophical inquiry is parochial in a troubling way. As Stich (1988, 1990, 2009) has emphasized, once we allow that the concept of knowledge, or free will, or reference that “folk like us” employ is only one of many actual (and infinitely possible) related concepts in the area, the conversation quickly turns away from such analysis to much the much broader philosophical project of figuring out why this one rather than that one?7 Still, many philosophers who reject the platonic armchair reject the turn to mentalism as well; they reject the idea that the primary target of philosophical

108

Advances in Experimental Philosophy and Philosophical Methodology

method is some better understanding of the mental entities that produce those judgments. They hold instead that philosophers are interested in various extra-mental, philosophically relevant properties: They are interested in knowledge itself (and moral responsibility, reference, and freedom of the will), not just our concepts or theories knowledge, moral responsibility, reference, and freedom of the will). I turn now to these.

4. Move to a posteriori faculty of understanding An alternative account of where experimental philosophers have gone wrong is by assuming that the method of cases involves some a priori faculty by which we apprehend truths. While some endorse this position (e.g. Bealer 1998), others insist that our spontaneous philosophical intuitions are produced by a broader competence (perhaps even all the information the subject has at her disposal) that is not unique to philosophical method. Michael Devitt has characterized intuitions as “empirical theory-laden central processor responses to phenomena” (2006: 103ff.). And Timothy Williamson has influentially claimed that “so-called intuitions are simply judgments (or dispositions to judgment); neither their content nor the cognitive basis on which they are made need be distinctively philosophical” (2007: 3). On this view, such judgments can have a range of sources, and they do not form any distinctive psychological kind. Crucially, however, the experimental philosophical critique need not depend on attacking a distinctively Platonic armchair, or on any eccentric psychological construal of the relevant mental states. As Joshua Alexander and Jonathan Weinberg note in responding to Williamson: the results of experimental philosophers are not themselves framed in terms of intuitions, but in terms of the counterfactual judgments of various subjects under various circumstances. Although the results are often glossed in terms of intuitions to follow standard philosophical usage, inspection of the experimental materials reveals little talk of intuitions and mostly the direct evaluation of claims. … the challenge reveals that at the present time philosophers may just not know what their evidence really is. And the true extent of their evidence is not, we think, something that they will be able to learn from their armchairs. (2007: 72)



Intuitive Diversity and Disagreement

109

Alexander and Weinberg emphasize that the problem that experiments reveal with spontaneous judgments about cases does not depend upon their being in some special way a priori or having some special psychological basis. Rather, it arises on any view that assumes them to be noninferential judgments that serve as evidence. We can see how this argument plays out in cases of disagreement. If experimental work shows systematic variability in our intuitive judgments about J along some dimension (e.g. culture, personality traits) that we take not to confer epistemic superiority, such variability ought to lower our own credence in our own judgments about J. This effect of disagreement does not seem to depend on substantive claims about the sources of psychological intuition.

5. Do philosophers actually rely on intuitions? But Max Deutsch (2010) and Herman Cappelen (2012) will not be satisfied with this, because they deny not only that intuitions are some sort of a priori cognition of abstract truths but also that the sorts of noninferential judgments that get called “intuitions” in philosophy (and critiqued by experimental philosophers) play the role of evidence at all. For both Deutsch and Cappelen, since noninferential judgments do not play the role of evidence in arguments, refuting them is beside the point. Cappelen writes: If it were true that philosophers relied on such spontaneous responses as foundational evidence, then it might be very interesting to learn that they vary widely with philosophically irrelevant factors. The problem for x-phi is that this model of case judgments is false. (2014: 273)

This, as both Deutsch (2010) and Cappelen (2012) allow, is a surprising claim, for many metaphilosophical discussions suggest otherwise. But Deutsch and Cappelen suggest that such discussions are in error about the method of philosophy. To show this, they go to considerable trouble to present philosophical case studies, and show that the “intuitive” judgments are actually the subject of considerable philosophical argument. In selecting such cases, there is, of course, the danger of “cherry picking” cases that lack evidential use of

110

Advances in Experimental Philosophy and Philosophical Methodology

intuitions. Deutsch, however, chooses cases—from Edmund Gettier (1963) on knowledge, and Saul Kripke (1980) on reference—that have been central to the experimental philosophical challenge from intuitional diversity. He then argues that “Gettier’s and Kripke’s arguments…do not implicitly assume that intuitions are evidence. Intuitions play no evidential role in these arguments” (Deutsch 2010: 452). I turn to consider these two cases here. Consider first Kripke’s famous Gödel case (which served as the basis for some two cases in Machery et al. 2004): Let’s take a simple case. In the case of Gödel that’s practically the only thing many people have heard about him—that he discovered the incompleteness of arithmetic. Does it follow that whoever discovered the incompleteness of arithmetic is the referent of ‘Gödel’? Suppose that Gödel was not in fact the author of this theorem. A man named ‘Schmidt’, whose body was found in Vienna under mysterious circumstances many years ago, actually did the work in question. His friend Gödel somehow got hold of the manuscript and it was thereafter attributed to Gödel. On the view in question, then, when our ordinary man uses the name ‘Gödel’, he really means to refer to Schmidt, because Schmidt is the unique person satisfying the description, ‘the man who discovered the incompleteness of arithmetic’. Of course you might try changing it to ‘the man who published the discovery of the incompleteness of arithmetic’. By changing the story a little further one can make even this formulation false. Anyway, most people might not even know whether the thing was published or got around by word of mouth. Let’s stick to ‘the man who discovered the incompleteness of arithmetic’.

From this elaborate case, Kripke notes that the descriptivist account of proper names seems to entail that the name “Gödel” picks out Schmidt, the man who actually did the work, and that this seems wrong: So, since the man who discovered the incompleteness of arithmetic is in fact Schmidt, we, when we talk about ‘Gödel’, are in fact always referring to Schmidt. But it seems to me that we are not. We simply are not. (1980: 83–4)

Here, Kripke’s “it seems to me we are not. We simply are not,” makes this case seem like a case where we are being invited to share a spontaneous judgment, and where that judgment serves as one piece of evidence for Kripke’s overall conclusion about the inadequacy of descriptivism. The judgment is that “we are not” using names in the way descriptivism predicts.



Intuitive Diversity and Disagreement

111

Deutsch is initially impressed by the fact that Kripke (and also Gettier) do not explicitly appeal to what is “intuitive” in their presentation of the cases he considers. But, as Deutsch acknowledges in a footnote (p. 451, fn. 2), it is false to say that Kripke nowhere speaks of intuitiveness or its evidential force. In fact, Kripke writes in Naming and Necessity: Of course, some philosophers think that something’s having intuitive content is very inconclusive evidence in favor of it. I think it is very heavy evidence in favor of anything, myself. I really don’t know, in a way, what more conclusive evidence one can have about anything, ultimately speaking. (1980: 42)

Kripke here apparently commits to the very general claim: for all x, x’s having intuitive content is very heavy evidence in favor of x. Surprisingly, Deutsch claims that when we consider the context of the passage, it’s clear that “Kripke is not endorsing an intuitions-as-evidence method for philosophy in general, or for evaluating counterexamples in particular” (p. 451, fn. 2). Deutsch’s interpretation depends upon Kripke’s use of “anything” involving some implicit, bounded quantification that excludes philosophical claims and evaluations of counterexamples from the things that x ranges over. This, in a work of philosophical claims and evaluations of counterexamples, seems implausible. Deutsch goes on to argue that Kripke’s and Gettier’s judgments do not, for either Kripke or Gettier, stand by themselves as justification for acceptance of philosophical theories, but rather find their place in a rich set of reasons to endorse their conclusions (cf. Devitt 2010). To see this, consider Deutsch’s other analysis, which considers Gettier’s first, “10 coins-case.” That case is as follows: Suppose that Smith and Jones have applied for a certain job. And suppose that Smith has strong evidence for the following conjunctive proposition: d. Jones is the man who will get the job, and Jones has ten coins in his pocket.

Smith’s evidence for (d) might be that the president of the company assured him that Jones would in the end be selected, and that he, Smith, had counted the coins in Jones’s pocket 10 minutes ago. Proposition (d) entails: e. The man who will get the job has ten coins in his pocket.

Let us suppose that Smith sees the entailment from (d) to (e), and accepts (e) on the grounds of (d), for which he has strong evidence. In this case, Smith is clearly justified in believing that (e) is true.

112

Advances in Experimental Philosophy and Philosophical Methodology

But imagine, further, that unknown to Smith, he himself, not Jones, will get the job. And, also, unknown to Smith, he himself has ten coins in his pocket. Proposition (e) is then true, though proposition (d), from which Smith inferred (e), is false. In our example, then, all of the following are true: (i) (e) is true, (ii) Smith believes that (e) is true, and (iii) Smith is justified in believing that (e) is true. The conclusion of this case proceeds as follows: But it is equally clear that Smith does not know that (e) is true; for (e) is true in virtue of the number of coins in Smith’s pocket, while Smith does not know how many coins are in Smith’s pocket, and bases his belief in (e) on a count of the coins in Jones’s pocket, whom he falsely believes to be the man who will get the job. (Deutsch 2010: 449–50; Gettier 1963)

Deutsch emphasizes that Gettier does not simply assert that: J. Smith does not know

but also proceeds to give additional reasons that seem to favor the truth of J. This reason-giving, Deutsch suggests, is evidence that Gettier took J to need arguing for, which is evidence that the spontaneous occurrence of J was not treated as independently justifying. Deutsch is relying on something like the following principle: R. If a person gives reasons for a judgment that p, then they do not take the judgment that p to be a source of noninferential justification.

Both Deutsch and Cappelen (2012) seem to infer from the fact that an author gives reasons for a judgment p to the conclusion that the author’s spontaneously judging p plays no evidential role. But this is not a valid inference, for R is false. A person might both hold a spontaneous intuition to be a source of evidence for a philosophical theory and also provide additional reasons for believing that intuition to be true. It follows that, even if case judgments are subject to additional reason giving, it does not follow that they have no evidential role.8 Consider that, on a number of prominent metaphilosophies, we ought to expect that judgments about cases provide only defeasible evidence in philosophical argument. For instance, metaphilosophies that have “reflective equilibrium” at their heart may involve adjusting theory or our considered



Intuitive Diversity and Disagreement

113

judgments about specific cases, “duly pruned and adjusted” (Rawls 1970: 18). Metaphilosophies that involve collecting intuitive “platitudes” in order to construct a Ramsey-sentence that allows us to understand what a property must be like to realize the sentence can also allow that some of the collected platitudes will have to be abandoned or modified (Lewis 1970, 1972; Jackson 1998). These examples make the general point: an intuitive judgment can constrain without having the status of what Cappelen calls “Rock” on which “Intuitive judgments justify, but they need no justification” (p. 112). Rather, one might offer additional justification for an intuition by way of indicating that the judgment in question is not to be abandoned on the basis of other theoretical considerations or competing intuitions, or, at least, not abandoned too lightly. On any metaphilosophy that treats intuitions as defeasible evidence, additional arguments for the truth of an intuition can be expected. Of course, one might still ask for evidence of why we should endorse the view that a particular spontaneous judgment is taken to be evidence. A better indication that Gettier used intuitions as evidence comes from his other, second case (which was the basis for one of Weinberg et al.’s studies 2001). That case concludes with a simple and bald assertion of the conclusion. Here is the case: Let us suppose that Smith has strong evidence for the following proposition: (f) Jones owns a Ford.

Smith’s evidence might be that Jones has at all times in the past within Smith’s memory owned a car, and always a Ford, and that Jones has just offered Smith a ride while driving a Ford. Let us imagine, now, that Smith has another friend, Brown, of whose whereabouts he is totally ignorant. Smith selects three place names quite at random and constructs the following three propositions: (g) Either Jones owns a Ford, or Brown is in Boston. (h) Either Jones owns a Ford, or Brown is in Barcelona. (i) Either Jones owns a Ford, or Brown is in Brest-Litovsk.

Each of these propositions is entailed by (f). Imagine that Smith realizes the entailment of each of these propositions he has constructed by (0, and proceeds to accept (g), (h), and (i) on the basis of (f). Smith has correctly inferred (g), (h), and (i) from a proposition for which he has strong evidence.

114

Advances in Experimental Philosophy and Philosophical Methodology

Smith is therefore completely justified in believing each of these three propositions. Smith, of course, has no idea where Brown is. But imagine now that two further conditions hold. First, Jones does not own a Ford, but is at present driving a rented car. And secondly, by the sheerest coincidence, and entirely unknown to Smith, the place mentioned in proposition (h) happens really to be the place where Brown is. And Gettier then simply concludes: If these two conditions hold, then Smith does not KNOW that (h) is true, even though (i) (h) is true, (ii) Smith does believe that (h) is true, and (iii) Smith is justified in believing that (h) is true. (1963: 122–3)

And that is the end of the second argument. As in Kripke’s judgment about “Gödel,” the reader is supposed to consider the case and simply see that Gettier’s conclusion follows, and from that conclude that the various attempts to offer a JTB analysis of knowledge fails. Looking back to the first example, the 10 coins-case, we might also ask why the additional considerations are not (or not also) manifestations of intuition? Gettier’s reasons there are: for (e) is true in virtue of the number of coins in Smith’s pocket, while Smith does not know how many coins are in Smith’s pocket, and bases his belief in (e) on a count of the coins in Jones’s pocket, whom he falsely believes to be the man who will get the job. (Deutsch 2010: 449–50; Gettier 1963)

But these seem to count as reasons for J only given certain spontaneous judgments about the character of knowledge: for example, that (e) has the wrong source to be knowledge. That is, these additional considerations seem designed to get the reader to see or apprehend why the judgment has to be the case, rather than to offer reasons for the judgment. Might Deutsch insist that Gettier infer the conclusion from the argument? I do not think this is plausible. If we construe the judgment that Smith does not know as the conclusion of an argument, it is an invalid argument. To make it valid, we must supplement it with some very substantial assumption about the nature of knowledge. It seems odd if it is intended as an argument that such a key assumption would be left out. This line of argument might seem too demanding. After all, the thought goes, all arguments start with some premises that cannot themselves be argued



Intuitive Diversity and Disagreement

115

for. Perhaps that is all that Gettier is doing here? I return in the next section to the thought that intuitions simply play the role of starting assumptions. For now, consideration of the Kripke and Gettier texts, I suggest, offers a good reason to think that spontaneous judgment does play an evidential role in philosophizing in the very cases used to show intuitional diversity. In fact, Deutsch seems to allow that the judgments in these cases are intuitive, and that judgments like J might be a manifestation of direct knowledge about the fact in question. But he is unperturbed by this admission, because he goes on to argue that even if the argument does appeal to an intuition, it need not: “the important point is that philosophers need not appeal to intuitions as evidence, regardless of whether they sometimes do” (Deutsch 2010: 456). We can concede that this might be the important point for Deutsch, who wants to conclude that “variability in philosophical intuitions poses no threat to philosophy” (p. 448). But this point offers no rejoinder at all to experimental philosophers demanding methodological revision of existing philosophical practice. Experimental philosophers critique actual rather than possible practice. The important point is simply that experimental philosophers are right that traditional philosophers sometimes do appeal to intuitions as evidence, not whether they might have forgone such appeals. Here I have argued for the view that philosophers really do treat spontaneous intuitions as evidence, and, to show that, we have followed Deutsch to the cases. Both Kripke’s and Gettier’s work suggests a role for intuitive judgment as a kind of evidence, suggesting that experimental philosophers do have a critical target, one in which epistemic disagreement should lead to a reduction in one’s certainty about the truth of one’s own intuitions. The argument from disagreement is, then, straightforwardly relevant.

6. Disagreement, argument, and “intuition”-talk as a rhetorical device I now turn to consider what role disagreement might play if we accept the idea that no philosophers are actually appealing to spontaneous judgments as evidence. Here, we note that whether judgments about cases stand as independent sources of justification—or, as Deutsch and Cappelen suggest, are

116

Advances in Experimental Philosophy and Philosophical Methodology

merely propositions whose rational warrant depends upon the further (actual or implied) arguments that are given for them—is relevant to the strength of the argument for conciliation from disagreement on the assumption that the character of our reasons for believing something is relevant to how we ought to react to disagreement. Concessively, we can say that, insofar as Deutsch and Cappelen are correct that apparent “intuitive” responses to cases are in actual philosophical method intended as reasons only conditional upon their being the product of other explicit arguments, it plausibly favors more steadfast responses to disagreement about them (e.g. Kelly 2005). One’s explicit reasons, in such cases, may rationally demand some substantial degree of steadfastness. This idea plausibly favors the view that we philosophers may stand with our judgments that p even in the face of systematic diversity among our apparent epistemic peers, provided that we have good reasons. Less concessively, however, we can note that, to show the irrelevance of such disagreement to philosophy, Cappelen and Deutsch need the very strong thesis that the disagreements about p revealed by experimental philosophy surveys ought to have no effect at all on our credences with regard to our judgments that p. Anything weaker than this would make experimental philosophical evidence relevant to the extent that conciliation was in order. This is a sort of floor of relevance for experimental philosophical work. However, disagreement matters in another way, for it affects what we can reasonably assume as beginning premises or as common ground. Return to the thought expressed above that the analysis of philosophical arguments in which unargued premises are treated as freestanding “intuitive” sources of evidence is too demanding. Every argument must begin somewhere, and such beginning premises are inevitably assumed, rather than argued for, within the context of a particular discussion. According to this line of thought, it is a mistake to interpret any such assumptions as being treated as freestanding sources of justification. Rather, they are simply provisional suppositions. A number of philosophers have asserted that this is just the way that “intuition” is used in philosophical practice. Cian Dorr has influentially put this view thus: Often, saying ‘Intuitively, P’ is no more than a device for committing oneself to P while signaling that one is not going to provide any further arguments



Intuitive Diversity and Disagreement

117

for this claim. In this use, ‘intuitively …’ is more or less interchangeable with ‘it seems to me that …’. There is a pure and chilly way of writing philosophy in which premises and conclusions are baldly asserted. But it’s hard to write like this without seeming to bully one’s readers; one can make things a bit gentler and more human by occasionally inserting qualifiers like ‘it seems that’. It would be absurd to accuse someone who frequently gave in to this stylistic temptation of following a bankrupt methodology that presupposes the erroneous claim that things generally are as they seem. But the sprinkling of ‘intuitively’s and ‘counterintuitive’s around a typical paper in metaphysics is in most cases not significantly different from this. It may be bad style, but it is not bad methodology, or any methodology at all, unless arguing from premises to conclusions counts as a methodology. (2010)

If this is right, “intuition”-talk simply indicates an interim assumption. Such assumptions would then be rightly subject to further demands for justification, but wrongly critiqued as treating intuitions as evidence. Notice also that this reconstruction can be combined with the Cappelen, Deutsch and Williamson views sketched in previous sections, attractively affording the possibility of combining naturalistic skepticism about the evidential value of intuitions with a nonreformist defense of current philosophical methodology. I think there is good reason for skepticism about whether this is really a plausible reading of many “it is intuitive” claims.9 Our language affords us numerous ways of marking assumptions that are not easily mistaken for asserting them. For instance, one can say, “here I simply assume p for the sake of argument.” In any case, here I simply assume Dorr is correct for the sake of argument. Even then, it still looks like the argument from disagreement has critical bite for reasons that emerged above in our discussion of mentalism. Clearly, it can be perfectly reasonable for philosophers to suppose some proposition p and consider what is entailed by it. (“Suppose humans could regenerate lost body parts in minutes. Would it then be morally permissible to cut off someone’s arm as a prank?.”) But starting places for arguments often reflect not only the assumption that one thinks a proposition is interesting or true but often, also, the assumption that the proposition is at least plausible, or perhaps even shared with one’s audience. Suppose it turns out that, when we do an experimental philosophical study, there is systematic disagreement about p. The argument from epistemic disagreement suggests that this gives us some reason to think that p is not

118

Advances in Experimental Philosophy and Philosophical Methodology

true, and we ought to reduce our credence in it. But diversity regarding p also looks relevant to the permissibility of beginning a discussion with the assertion that p, of treating p as common ground, or even as interim common ground. Because evidence for diversity of judgment gives evidence against a presupposition of shared ground, it undercuts the assertion of p as a starting place. Where experimental philosophical work has already revealed diversity in judgments about p, subsequent passing over of known disagreement without comment can be a form of disregard. But it’s also the case that such diversity suggests that past presuppositions of reasonable starting positions may have been mistaken. If this is right, then evidence of diversity is evidence regarding the appropriateness of the assumptions with which we begin. Thus even if conceding that “intuition”-talk may simply mark an assumption, the revelation of philosophical disagreement is philosophically relevant, certainly to subject matters that have been experimentally examined, but also to those relevantly similar. Such experimental evidence can show that simply assuming a proposition without explanation is rhetorically inappropriate.

7. Disagreements and future philosophy This chapter has concerned critics of experimental philosophy who are, like experimental philosophers themselves, no fans of the platonic armchair picture of one philosophical methodology. I have argued that diversity of philosophical judgments remains a serious problem for such critics, both because such critics have wrongly construed existing practice and because diversity of judgment calls for reform of this “armchair” philosophical method even when we conceive of this method in different ways. The difference among these critics and experimental philosophers is a difference in how we understand the character of existing and past philosophical practice. But it can also be seen as a difference over the direction and character of future philosophy. In contrast to so called “positive experimental philosophers” who see a role for experimental methods in improving philosophical appeals to intuition, and others who see philosophical and psychological concerns as more or less continuous, the philosophers I have



Intuitive Diversity and Disagreement

119

considered here articulate a vision of a post-intuition, future philosophical practice that remains largely in the armchair. Among experimental philosophers, the character of future philosophy commands no clear consensus. So, for instance, while I have argued against Deutsch that it is implausible to read Kripke and Gettier as eschewing evidential appeal to intuitions, I have offered no arguments against his claim that armchair philosophy need not involve such appeal to intuitions. Some experimentalists, we might expect, will be happy to follow Deutsch back to the armchair. Others will continue with increasingly sophisticated surveys and other experimental techniques. Which way is it to be: the armchair or the laboratory? Here, we need make no final decision, for methodologies pay their own way by the number and interest of the results they produce. If what I have argued above is right, we do not yet know what post-intuition armchair philosophy has to offer us. And at the same time, experimental philosophy is still young, and what it might do as the range of its explananda expands and the variety of its methods multiplies remains to be seen.

Notes 1 My thanks go to Max Deutsch, Michael Johnson, Jenny Nado and Jonathan Weinberg for many helpful comments on an earlier draft of this paper. 2 Though see Kauppinen 2007: 107; Cullen 2010. 3 See Christensen 2009 for an introduction. 4 It falls far short of, for example, Adam Elga’s (2007) ‘equal weight view’ that requires we give disagreeing views equal weight with our own. 5 Strictly speaking, Goldman and Pust’s (1998) account of intuition allows this point, for it recognizes the distinction between an intuition like E. x is an instance of knowledge and a judgment like E’s occurrence, in favorable circumstances is evidence that my concept of knowledge is such and so. But their allowance for this is an ad hoc adjustment to their idea of a basic evidential source, a status that a mental state has if it provides evidence of the truth of its content to which they add “(or a closely related content).” My point is that it is an advantage of the extra-mentalist approach that it allows us to see how intuitions could be philosophically important without assuming they provide

120

6 7

8 9

Advances in Experimental Philosophy and Philosophical Methodology

evidence for their content, avoiding the need for spelling out how giving evidence for a “closely related” content is relevant to the truth of p. Perhaps this occurs via introspection or reflection. Or perhaps it occurs simply by appending “I believe that” to J (cf. Shoemaker 1988). Nichols and Ulatowski (2007) have suggested on empirical grounds that “intentionally” expresses multiple, partially overlapping concepts even within U.S. culture, making this issue more pressing. Weinberg 2014 raises a similar objection to Cappelen. I am grateful to Jonathan Weinberg for discussion on this point.

References Alexander, J., R. Mallon, and J. Weinberg (2010), “Accentuate the Negative,” Review of Philosophy and Psychology 1 (2): 297–314. Bealer, G. (1998), “Intuition and the Autonomy of Philosophy,” in M. DePaul and W. Ramsey (eds), Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry, Lanham, MD: Rowman and Littlefield, 201–39. Cappelen, H. (2012), Philosophy without Intuitions, Oxford: Oxford University Press. Cappelen, H. (2014), “X-Phi without Intuitions?,” in A. Booth and D. Rowbottom, Intuitions, Oxford: Oxford University Press, 269–86. Christensen, D. (2009), “Disagreement as Evidence: The Epistemology of Controversy,” Philosophy Compass 4 (5): 756–67. Cokely, E. T. and A. Feltz (2009), “Adaptive Variation in Judgment and Philosophical Intuition,” Consciousness and Cognition 18 (1): 356–8. Cullen, S. (2010), “Survey-Driven Romanticism,” Review of Philosophy and Psychology 1 (2): 275–96. Deutsch, M. (2009), “Experimental Philosophy and the Theory of Reference,” Mind and Language 24 (4): 445–66. Deutsch, M. (2010), “Intuitions, Counter-examples, and Experimental Philosophy,” Review of Philosophy and Psychology 3: 447–60. Devitt, M. (2006), Ignorance of Language, Oxford: Clarendon. Devitt, M. (2010), “Experimental Semantics,” Philosophy and Phenomenological Research LXXXII (2). Dorr, C. (2010), “Review of Every Thing Must Go: Metaphysics Naturalized by James Ladyman and Don Ross, with David Spurrett and John Collier, Oxford OUP, 2007,” Notre Dame Philosophical Reviews.



Intuitive Diversity and Disagreement

121

Dunaway, B., Edmonds, A., and Manley, D. (2013), “The Folk Probably do Think What you Think They Think,” Australasian Journal of Philosophy 91 (3): 421–41. Elga, A. (2007), “Reflection and Disagreement,” Noûs 41 (3): 478–502. Feltz, A. and Cokely, E. (2008), “The Fragmented Folk: More Evidence of Stable Individual Differences in Oral Judgments and Folk Intuitions,” Proceedings of the 30th Annual Conference of the Cognitive Science Society. B. C. Love, K. McRae and V. M. Sloutsky, Cognitive Science Society. Gettier, E. (1963), “Is Justified True Belief Knowledge,” Analysis 23 (6): 121–3. Goldman, A. (2010), “Philosophical Naturalism and Intuitional Methodology (Romnell Lecture),” Proceedings and Addresses of the American Philosophical Association 84. Goldman, A. I. and Pust, J. (1998), “Philosophical Theory and Intuitional Evidence,” in M. DePaul and W. Ramsey (eds), Rethinking Intuition: The Psychology of Intuition and its Role in Philosophical Inquiry, Lanham: Rowman and Littlefield. Jackson, F. (1998), From Metaphysics to Ethics: A Defense of Conceptual Analysis, Oxford: Oxford University Press. Kauppinen, A. (2007), “The Rise and Fall of Experimental Philosophy,” Philosophical Explorations 10 (2): 95–118. Kelly, T. (2005), “The Epistemic Significance of Disagreement,” Oxford Studies in Epistemology, Volume 1, J. Hawthorne and T. Gendler, eds, Oxford: Oxford University Press, 167–96. Kripke, S. A. (1980), Naming and Necessity, Cambridge, MA: Harvard University Press. Lewis, D. (1970), “How to Define Theoretical Terms,” Journal of Philosophy 67: 426–46. Lewis, D. (1972), “Psychophysical and Theoretical Identifications,” Australasian Journal of Philosophy 50: 249–58. Machery, E. (2012), “Expertise and Intuitions about Reference,” Theoria 27 (1): 37–54. Machery, E., M. Deutsch, R. Mallon, S. Nichols, J. Sytsma, and S. P. Stich (2010), “Semantic Intuitions: Reply to Lam,” Cognition 117 (3): 361–6. Machery, E., R. Mallon, S. Nichols, and S. Stich (2004), “Semantics, Cross-Cultural Style,” Cognition 92: B1–B12. Mallon, R. (2007), “Reviving Rawls Inside and Out,” in W. Sinnott-Armstrong (ed.), Moral Psychology, Volume 2: The Cognitive Science of Morality: Intuition and Diversity, Cambridge, MA: MIT Press. 145–55. Mallon, R. (forthcoming), “Experimental Philosophy,” in H. Cappelen, J. Hawthorne

122

Advances in Experimental Philosophy and Philosophical Methodology

and T. S. Gendler (eds), Oxford Handbook of Philosophical Methodology, Oxford: Oxford University Press. Mallon, R., E. Machery, S. Nichols, and S. P. Stich (2009), “Against Arguments from Reference,” Philosophy and Phenomenological Research 79 (2): 332–56. Nagel, J. (2012), “Intuitions and Experiments: A Defense of the Case Method in Epistemology,” Philosophy and Phenomenological Research 85 (3): 495–527. Nagel, J., V. San Juan, and R. A. Mar (2013), “Lay Denial of Knowledge for Justified True Beliefs,” Cognition 129: 652–61. Nichols, S., S. P. Stich, and J. Weinberg (2004), “Metaskepticism: Meditations in Ethno-Epistemology,” in S. Luper (ed), The Skeptics, Burlington, VT: Ashgate. Nichols, S. and J. Ulatowski (2007), “Intuitions and Individual Differences: The Knobe Effect Revisited,” Mind and Language 22 (4): 346–65. Nisbett, R. E. (2003), The Geography of Thought: How Asians and Westerners Think Differently—and Why, London and Yarmouth, ME: Nicholas Brealey Publishing. Rawls, J. (1970), A Theory of Justice, Cambridge, MA: Harvard University Press. Seyedsayamdost, H. (2015), “On Normativity And Epistemic Intuitions: Failure Of Replication,” Episteme 12 (1): 95–116. Shoemaker, S. (1988), “On Knowing One’s Own Mind,” Philosophical Perspectives 2: 183–209. Sosa, E. (2007), “Experimental Philosophy and Philosophical Intuition,” Philosophical Studies 132 (1): 99–107. Stich, S. (1988), “Reflective Equilibrium, Analytic Epistemology and the Problem of Cognitive Diversity,” Synthese 74 (3): 391–413. Stich, S. (1990), The Fragmentation of Reason: A Preface to a Pragmatic Theory of Cognitive Evaluation, Cambridge, MA: MIT Press. Stich, S. (1996), Deconstructing the Mind, New York: Oxford University Press. Stich, S. (2009), “Reply to Sosa,” in M. Bishop and D. Murphy (eds), Stich and His Critics, Malden, MA: Wiley-Blackwell. Stich, S. (2013), “Do Different Groups Have Different Epistemic Intuitions? A Reply to Jennifer Nagel,” Philosophy and Phenomenological Research 87 (1): 151–78. Stich, S. and Weinberg, J. (2001), “Jackson’s Empirical Assumptions,” Philosophy and Phenomenological Research 62 (3): 637–43. Sytsma, J. and Livengood, J. (2011), “A New Perspective Concerning Experiments in Semantic Intuitions,” Australasian Journal of Philosophy 89 (2): 315–32. Sytsma, J., J. Livengood, R. Sato, and M. Oguchi (2015), “Reference in the Land of the Rising Sun: A Cross-cultural Study on the Reference of Proper Names,” Review of Philosophy and Psychology 6 (2): 213–30.



Intuitive Diversity and Disagreement

123

Talbot, B. (2014), “Why So Negative? Evidence Aggregation and Armchair Philosophy,” Synthese 191 (16): 3865–96. Turri, J. (2013), “A Conspicuous Art: Putting Gettier to the Test,” Philosophers’ Imprint 13 (10). Weinberg, J. M. (2014), “Cappelen Between Rock and a Hard Place,” Philosophical Studies 171 (3): 545–53. Weinberg, J., J. Alexander, C. Gonnerman, and S. Reuter (2012), “Restrictionism and Reflection: Challenge Deflected, or Simply Redirected?” The Monist 95: 20–2. Weinberg, J. M., C. Gonnerman, C. Buckner, and J. Alexander (2010), “Are Philosophers Expert Intuiters?,” Philosophical Psychology 23 (3): 331–55. Weinberg, J., S. Nichols, and S. P. Stich (2001), “Normativity and Epistemic Intuitions,” Philosophical Topics 29 (1/2): 429–59. Williamson, T. (2007), The Philosophy of Philosophy, Malden, MA: Blackwell. Williamson, T. (2013), “Review of J. Alexander, Experimental Philosophy: An Introduction,” Philosophy 88 (3): 467–74.

7

Intuitions and the Theory of Reference1 Jennifer Nado and Michael Johnson2

Do philosophers rely on intuitions as their primary evidence for philosophical theories? Many philosophers have long simply assumed that they do. Some have gone even further and asserted that intuitions are the data of philosophy and that the goal of a philosophical theory is to “capture” our intuitions. Recently, however, this picture of philosophy has come under question. Authors like Timothy Williamson, Max Deutsch, and Herman Cappelen have challenged the thesis that Cappelen calls “Centrality”: the thesis that philosophical evidence consists chiefly of intuitions. Further, these philosophers have suggested that the falsity of Centrality undermines the primary project of experimental philosophy. Experimental philosophy, according to experimental philosophers themselves, mainly involves conducting surveys of folk intuitions; if intuitions are not a source of evidence in philosophy, or only a very peripheral source, then it wouldn’t appear very fruitful to spend lots of resources empirically studying them. Although Williamson, Deutsch, and Cappelen have set their sights on philosophical methodology as a whole, other philosophers have leveled strikingly similar challenges at one particular experimental study. In that study, Edouard Machery, Ron Mallon, Shaun Nichols, and Stephen Stich (hereafter, MMNS) uncovered cross-cultural differences in responses to one thought experiment drawn from Saul Kripke’s Naming and Necessity, and suggested that these differences undermined Kripke’s methodology and conclusions. Subsequently, the relevance of MMNS’ finding to Kripke’s project has been vigorously challenged by Michael Devitt, Genoveva Martí, and others, on the grounds that MMNS had over-exaggerated the role intuitions about reference play in Kripke’s arguments and in theories of reference in general.

126

Advances in Experimental Philosophy and Philosophical Methodology

Interestingly, points of contact between these two debates have been few.3 At first glance, one might expect that the arguments presented by Devitt, Martí, and others, could be taken by fans of the “anti-Centrality” position to be just more fuel for their fire. But, as we shall argue, the relationship between the two debates is more complex. Philosophy is a diverse field, and what is true for the theory of reference may not be true for other lines of philosophical inquiry. Indeed, what is true for Kripke’s particular theory of reference and the arguments he gives for it may not be true more generally for other theories of reference. Thus we think the question raised by Williamson, Deutsch, and Cappelen—whether philosophers rely centrally on intuitions as evidence—is unlikely to have an unequivocal answer across all philosophical disciplines. In this paper, we will examine the role that intuitions and responses to thought experiments play in confirming or disconfirming theories of reference, using insights from both debates as our starting point. Our view is that experimental evidence of the type elicited by MMNS does play a central role in the construction of theories of reference. This, however, is not because such theory construction is accurately characterized by “the method of cases.” First, experimental philosophy does not directly collect data about intuitions, but rather about people’s responses to thought experiments, which may reflect their intuitions but may well not. Second, unusually in the case of the theory of reference, experimental prompts involve elicitation of the phenomenon under investigation—that is, referring—and it is the reference facts rather than the intuitions that a theory of reference should capture. Finally, “best fit” models like the method of cases are inconsistent with the actual practice of semantic theorists, who appeal to general principles of theory construction (e.g. beauty and simplicity) as well as considerations native to the theory of reference (such as Grice’s razor), often in the service of rejecting intuitions. Indeed, Kripke himself viewed several claims of his theory as unintuitive, but felt no pressure to alter them on that account. These facts, we’ll argue, suggest that the relevance of experimental methods in the theory of reference cannot be straightforwardly extended to other fields: an experimental prompt containing Newton’s two rocks tied together in empty space may elicit referring, and may also elicit intuitions, but it doesn’t also elicit the phenomenon under investigation (rocks rotating in empty space). Furthermore, the considerations native to the theory of



Intuitions and the Theory of Reference

127

reference are not part of physics, which has its own distinct set of principles and considerations. This suggests that an evaluation of the centrality of intuitions/ responses to thought experiments in philosophical methodology must proceed in a piecemeal fashion. We conclude that the potential contributions of experimental philosophy will take different forms in different sub-fields of philosophical inquiry.

1. Intuitions in philosophy: Two debates As we mentioned, there are two separate challenges that have been leveled at experimental philosophy’s focus on intuition. The broader challenge, which, following Cappelen, we’ll call the “anti-Centrality” challenge, questions the role of intuition in philosophy as a whole.4 The narrower challenge has focused on the role of intuitions about reference in Kripke’s arguments against descriptivist theories, and the role of intuitions in confirming or disconfirming theories of reference more generally. While the two challenges are superficially similar, they differ in the specific arguments they advance. We start first with a discussion of the broader challenge. A prominent theme in Cappelen’s and Williamson’s anti-Centrality arguments—though not Deutsch’s—is a general skepticism regarding the very concept of an intuition. For Cappelen, philosophers’ usage of the term “intuition” fails to reflect a clear, commonly agreed-upon definition. Some philosophers insist that intuitions must arise from conceptual competence; others do not. Some insist that intuition has a distinctive phenomenology; others do not. And so on. Similarly, Williamson notes that the label “intuitive” can be applied to nearly any claim—even ones that are straightforwardly perceptual. There is simply no clear dividing line between philosophical intuitions and such everyday judgments as mundane cases of concept application. Williamson goes so far as to claim that “philosophers might be better off not using the word ‘intuition’ and its cognates” (Williamson 2007: 200); Cappelen would presumably agree.5 Of course, if basically everything counts as an “intuition” or if “intuition” fails to pick out a clear referent, it becomes difficult even to make sense of the claim that philosophy centrally relies on intuitions as evidence. This cuts both

128

Advances in Experimental Philosophy and Philosophical Methodology

ways, however: it becomes difficult to make sense of anti-Centrality claims as well. If there aren’t any intuitions, because the notion is ill-defined, then one doesn’t really need a case study to show that philosophers don’t appeal to them. According to the anti-Centrality camp, how then do philosophers arrive at evidence for or against philosophical theories, if not by way of an evidential appeal to intuition? The answer given is multi-faceted. Both Deutsch and Williamson, for instance, reject the view that our evidence consists of intuition facts like “I intuit that P” in favor of the view that our evidence consists of facts like p itself. In Williamson’s terms, the idea that philosophical evidence consists of intuition-facts involves an illegitimate “psychologization” of our evidence: it puts us in the undesirable position of having to argue from psychological premises to conclusions about the non-psychological (metaphysics, epistemology, etc.). Williamson holds that such a gap is “not easily bridged” (Williamson 2007: 211). However, denying that it’s appropriate to psychologize our evidence is compatible with an evidential role for intuition. Compare vision: if we assume that when we see that P, our evidence is p and not merely that we see that P, it’s still true that the reason we’re justified in believing p is that we’d seen it. Cappelen claims that what may look like an appeal to “intuition” is often really just an appeal to what is in the common ground—that is, to a proposition that needs no further support in the current dialectical context. Such common ground claims may be based on any evidential source whatsoever, and they need not be a priori. Furthermore, Cappelen and Deutsch both point out that philosophers who employ thought experiments frequently engage in careful argumentation, rather than brute appeal to intuitions about the target case. Deutsch, for instance, notes that Gettier discusses the disconnect in his counterexamples between what makes the proposition true and what makes it believed. Cappelen, meanwhile, claims that close examination of paradigm thought experiments reveals no mention of reliance on states with the sorts of features associated with intuition, such as special phenomenology or a “rockbottom” evidential status. We agree with many of these points. Like Williamson and Cappelen, we have doubts about the category of “intuition.” As one of us has previously argued (Nado 2014), the states covered by the term “intuition” aren’t



Intuitions and the Theory of Reference

129

likely to form a natural kind at the psychological level: in different domains like moral judgment or mathematical judgment, distinct mechanisms with distinct assumptions and inferential procedures will produce the spontaneous, non-consciously-inferred judgments we call “intuitions.” However, beyond the fact that such judgments are spontaneous and not consciously inferred, there will be almost nothing in general to say about them—for example, about how reliable they are or how robust they are in unusual environments. This is part of our reason for believing that the Centrality question—and the more general question of the viability of experimental philosophy—can have no simple answer. We shall return to this issue at a later point in the paper. We also agree that, when (good) philosophers present their judgments regarding thought experiments, they also provide substantial argumentation to support those judgments. For us, however, this merely shows that the “method of cases” is at best a caricature of actual philosophical practice. Philosophers present an array of evidence for their claims. A particular philosopher may be justified in believing p because a particularly reliable mental mechanism caused her to spontaneously judge p in response to a thought experiment (without needing to argue from the latter fact to the former). She might present the thought experiment and her judgment, expecting her audience to perhaps spontaneously judge as she does. It doesn’t follow that we expect philosophers to explicitly appeal to intuitions in arguing for P. First, if they expect their readers to have the same intuition, appealing to their intuitions is redundant. Compare: if A and B both see that P, A doesn’t need to tell B that A sees that p in order to get B to believe P. Second, if they expect their readers to have different intuitions, independent argumentation is necessary. Thus, we find it no wonder that Cappelen’s case studies uncovered that, in a number of classic thought experiments, a judgment was presented, and then independent argumentation was given for that judgment. That’s how one covers one’s bases. It doesn’t show that intuitions aren’t evidence in philosophy. Let’s turn now to the narrower of the two debates, which focuses on MMNS’s cross-cultural findings on subjects’ responses to thought experiments concerning reference. MMNS’s study employed two well-known thought experiments used by Kripke in his arguments against the descriptivist theory of reference—the “Gödel” case and the “Jonah” case. It will be helpful to have

130

Advances in Experimental Philosophy and Philosophical Methodology

Kripke’s versions of these thought experiments fresh in mind, so we’ll reprint them here: Suppose that Gödel was not in fact the author of [Gödel’s incompleteness theorem for arithmetic]. A man called “Schmidt,” whose body was found in Vienna under mysterious circumstances many years ago, actually did the work in question. His friend Gödel somehow got hold of the manuscript and it was thereafter attributed to Gödel. On the view in question, then, when our ordinary man uses the name “Gödel,” he really means to refer to Schmidt, because Schmidt is the unique person satisfying the description “the man who discovered the incompleteness of arithmetic”… when we talk about “Gödel,” we are in fact always referring to Schmidt. But it seems to me that we are not. We simply are not. (Kripke 1980: 83–4) Suppose that someone says that no prophet ever was swallowed by a big fish or a whale. Does it follow, on that basis, that Jonah did not exist? There still seems to be the question whether the Biblical account is a legendary account of no person or a legendary account built on a real person. In the latter case, it’s only natural to say that, though Jonah did exist, no one did the things commonly related to him. (Kripke 1980: 67)

MMNS presented simplified versions of these cases to subjects in the United States and in Hong Kong (the prompts were given in English in both groups). Their Gödel prompt, for instance, describes a man named John who knows of Gödel only that he proved incompleteness; the prompt then tells of Gödel’s “deception,” and asks subjects to judge who John refers to when he uses the name “Gödel.” They found that there was a statistically significant difference in the responses between the US subjects and the Hong Kong subjects, and that the Hong Kong subjects were more likely to report the more “descriptivist” judgment that John refers to Schmidt. They found no significant difference in the groups’ responses to the Jonah case. It’s worth remarking at the outset that while MMNS claim that these findings show that there are cross-cultural differences in intuitions regarding the Gödel case, what they really reveal is a cross-cultural difference in the judgments subjects report about the thought experiment. Even on the thinnest conception of intuitions, where intuitions are spontaneous judgments that are not the conclusions of conscious reasoning, it’s questionable whether MMNS received many intuition-reports. Their subjects were university students in an



Intuitions and the Theory of Reference

131

academic setting. Students have a proclivity to spend at least some non-zero amount of time consciously reasoning about which answers they are to give in an academic setting. Be that as it may, MMNS do take their results to show a cross-cultural difference in intuitions. Furthermore, in the paper where these findings were reported, MMNS make the following claims about the methodology underlying Kripke’s work (and theories of reference generally): There is widespread agreement among philosophers on the methodology for developing an adequate theory of reference. The project is to construct theories of reference that are consistent with our intuitions about the correct application of terms in fictional (and non-fictional) situations. Indeed, Kripke’s masterstroke was to propose some cases that elicited widely shared intuitions that were inconsistent with traditional descriptivist theories. … Even contemporary descriptivists allow that these intuitions have falsified traditional forms of descriptivism. (Machery et al. 2004: B3)

In a later paper, MMNS clarify that the methodology they have in mind is what is often known as the “method of cases.” There, they characterize the method as follows: “The correct theory of reference for a class of terms T is the theory which is best supported by the intuitions competent users of T have about the reference of members of T across actual and possible cases” (Mallon et al. 2009: 338).6 If we assume the method of cases (as characterized above) is the method philosophers employ for evaluating theories of reference, and we assume that MMNS have found a cross-cultural difference in intuitions about a particular possible case, then it’s fairly clear how those findings have problematic consequences for that project. If different demographic groups differ in their intuitions, whose intuitions are to be captured by the theory of reference? What reason could possibly be given to think that Western subjects, say, have intuitions that track the true theory of reference, whereas East Asian subjects, on the other hand, do not? It’s possible, of course, to take the line that, in Hong Kong, names are used descriptively, whereas in the United States, they are not. But even a substantial minority of the Western subjects reported descriptivist judgments. Surely students from the same culture, on the same campus, in the same classroom, speaking the same language, aren’t governed by different theories of reference.

132

Advances in Experimental Philosophy and Philosophical Methodology

MMNS’s study has been challenged on several grounds (see for instance Ludwig 2007; Deutsch 2009; Lam 2010), but our particular focus will be on challenges that have been made to MMNS’s implicit and explicit claims about the methods of research in the theory of reference. MMNS asked subjects to report which person, Gödel or Schmidt, a hypothetical speaker, John, would be “talking about” when he used the name “Gödel.” Genoveva Martí (2009) claims that the intuitions subjects have about this question reflect only their own theories about the reference of proper names, but not how they in fact use proper names. Intuitions about reference, such as those in response to MMNS’s version of the Gödel case, provide little evidence for or against theories of reference, according to Martí: subjects might easily hold mistaken theories that don’t accurately reflect their own usage. Nonetheless, Martí does think that some types of intuitions provide evidence about actual usage, and may therefore provide evidence for theories of reference. Instead of asking subjects what a name refers to, or what some person is talking about when they use a name, Martí claims that we should elicit their reactions to dialogues in which a name is used in certain ways. Here is her example: In order to determine whether users of names in the two experimental groups use or don’t use ‘Gödel’ according to what is predicted by the causal-historical picture, it would be best if the end of the story, and the question asked, went along the following lines: One day, the fraud is exposed, and John exclaims: ‘Today is a sad day: we have found out that Gödel was a thief and a liar’. What do you think about John’s reaction? (Martí 2009: 47)

This line of questioning doesn’t employ semantic terminology like “reference,” and does not aim to produce judgments about reference. Yet it can be used to get at the referential facts, according to Martí, since a subject for whom names refer non-descriptively should report finding John’s reaction deeply strange. Meanwhile, Michael Devitt (2011), and Jonathan Ichikawa, Ishani Maitra, and Brian Weatherson (2012) argue that the Gödel and Jonah cases play a minimal role at best in Kripke’s arguments against descriptivism. Both Michael Devitt and Ichikawa et al. note that Kripke targets two distinct views in Naming and Necessity.7 Devitt distinguishes between descriptivism as a theory of meaning vs. as a theory of reference determination; Ichikawa et al. use “strong descriptivism” and “weak descriptivism” for the same distinction.



Intuitions and the Theory of Reference

133

The strong “theory of meaning” version of descriptivism tells you what the meaning of a name is—it says that the meaning of a proper name is the same as the meaning of some description. The weaker “theory of reference” version of descriptivism tells you why a name refers to what it does—it says that for each name N, there is a description that speakers associate with N, and whatever uniquely satisfies that description, if anything, is the referent of N. Kripke gives arguments against both varieties of descriptivism. He notes that, since the strong thesis says that the name “Aristotle” is synonymous with some description “the D,” this predicts that “Aristotle might not have been the D” is synonymous with “the D might not have been the D.” But Kripke argues that while the latter has a necessarily false reading, the former does not, at least for any of the ordinary descriptions purported to be synonymous with “Aristotle.” For instance, it’s certainly possible that Aristotle might not have been the teacher of Plato, or the last great philosopher of antiquity, or the man named “Aristotle.” The two phrases cannot therefore be synonymous. Devitt calls this argument “Unwanted Necessity,” and, according to him8, it does appeal to an intuition—not an intuition about reference, but, rather, a modal intuition. This modal claim does not, however, undermine the weaker version of descriptivism. Weak descriptivism says not that names are synonymous with descriptions but that descriptions fix referents: for each name N, there is a description that speakers associate with N, and whatever uniquely satisfies that description, if anything, is the referent of N. Against this view, Kripke presents an argument largely based on a variety of actual cases. First, he argues that often speakers lack enough information to uniquely identify individuals that the names they use nevertheless refer to. For example, speakers who know of Feynman only that he was a famous physicist still succeed in referring to Feynman when they use the name “Feynman;” speakers who know of Cicero only that he was a famous orator still succeed in referring to Cicero when they use the name “Cicero.” But according to weak descriptivism, Feynman and Cicero cannot be the referents of these names for these speakers, since neither Feynman nor Cicero uniquely satisfies the descriptions the speakers associate with “Feynman” and “Cicero,” respectively. Devitt calls this the “Argument from Ignorance.” Moreover, Kripke argues, many speakers associate descriptions with names that are false of those names’ referents. Some people only “know” of Columbus

134

Advances in Experimental Philosophy and Philosophical Methodology

that he was the first European to discover the Americas; others only “know” of Einstein that he invented the atomic bomb. Weak descriptivism predicts that the former speak of Leif Ericson when they say things like “Columbus was European.” Kripke is right to point out that, nevertheless, they are speaking of Columbus. Devitt calls this the “Argument from Error.” In addition to the actual cases Kripke employs in service of the Ignorance and Error arguments, he also presents several non-actual cases—including the Gödel case. Contra MMNS, however, Devitt argues that the Gödel case is not central to Kripke’s arguments against descriptivism, for the evidence it provides is fairly weak when compared to the evidence provided by his other cases. The Gödel case is an unusual fictional scenario; Devitt claims that our intuitions about unusual fictional scenarios are generally not as strong as they are on actual cases, and that we have more reason to expect error. This is especially true of the intuitions of the folk, who may not be used to thinking about such cases. Devitt holds that the experimental method is appropriate, but that it is better to conduct surveys about “humdrum” cases, not intuitions on cases that are “counterfactual, hypothetical, or fictional” (p. 421). Here, Devitt emphasizes that in the actualworld case MMNS surveyed (the Jonah case), MMNS found no cross-cultural difference in subjects’ reported judgments. Devitt’s considered view is that, on balance, the evidence we have still supports Kripke’s rejection of descriptivism. Ichikawa et al. (2012) are largely in agreement with Devitt that the Gödel case plays a minimal role in Kripke’s overall argument against descriptivism, but for a different reason. Devitt expresses skepticism about the evidential value of non-expert intuitions on the Gödel case, due to its unfamiliar, counterfactual character. Ichikawa et al., however, see no reason to doubt even naïve intuitions on the case. Instead, they take the Gödel case to play a minimal role simply because its primary use is to argue against an ultra-weak version of descriptivism according to which descriptions determine reference only in cases where the speaker possesses an individuating description. They argue that, since it is largely agreed that certain names like “Jack the Ripper” do function descriptively, then if the intuitions go against Kripke on the Gödel case, “that would at most show that there are a few more descriptive names than we thought there were” (Ichikawa et al. 2012: 6). Note that Martí, Ichikawa et al., and Devitt—unlike Cappelen and Williamson—are more than happy making use of the notion of an “intuition.”



Intuitions and the Theory of Reference

135

Indeed they each appear to accept that intuitions play an evidential role in the project of theorizing about reference. Martí writes that the primary flaw in MMNS’s study is not that they tested intuitions, but that they “did not test the right kind of intuitions” (2009: 44). Ichikawa et al. claim that “most of our intuitions in this field are surely correct” (2012: 7) And Devitt goes so far as to write that “Machery et al. are surely right in claiming that theories of reference are assessed by consulting intuitions: this practice does seem to be the method of semantics” (2011: 419). Elsewhere, Devitt even expresses bewilderment at how “strangely critical” Deutsch is of the standard characterization of philosophical methodology (2011b: fn. 5).9 It might seem then that these authors are rather pro-Centrality10, despite their view that the particular results of MMNS’s surveys are of little evidential significance. However, when we look a little closer, it turns out that not all the participants in the two debates are agreed on what counts as an intuition, and thus what the method of cases is, or what Centrality really amounts to. In the end, there is more agreement between the participants in both debates than initially appears on the surface.

2. Centrality and theories of reference Centrality is a slippery thesis. It says that the primary evidence for philosophical theories consists of intuitions. But how one understands this thesis depends on how one understands “intuition,” and how one understands what role something has to play to count as evidence. On different interpretations of the thesis, the authors in the MMNS debate can be viewed either as advocates or as opponents of Centrality. Below, we’ll give our take, and discuss whether Centrality holds for the theory of reference given the arguments Devitt, Ichikawa et al., and Martí have offered.

2.1 What is an intuition? As we noted, several statements Devitt gives in his discussion of MMNS suggest a pro-Centrality view. But Devitt’s understanding of “intuition” complicates the picture. For Devitt, intuitions are simply a type of judgment. What makes a judgment an intuition is the process by which you arrive at it:

136

Advances in Experimental Philosophy and Philosophical Methodology

[I]ntuitive judgments are empirical theory-laden central-processor responses to phenomena, differing from many other such responses only in being fairly immediate and unreflective, based on little if any conscious reasoning. (Devitt 2006: 10)

Many non-philosophical sorts of judgments, even perceptual classification judgments like the cat is on the mat, plausibly turn out to be intuitions on this account. Thus Devitt’s view of intuition is in fact fairly Williamsonian, in that it countenances no sharp line between categorizing mundane objects and kinds in the world and categorizing philosophical phenomena in thought experiments. It’s also worth noting that Devitt’s definition of an intuition does not assign to it any of the features which Cappelen suggests that Centrality proponents ascribe to intuition (namely, special phenomenology, “rockbottom” evidential status, and an etiology based in conceptual competence).11 So, in a sense, Devitt likely agrees with Cappelen that philosophers make no use of “intuitions” in the sense Cappelen has in mind. Nevertheless, Devitt differs from Williamson and Cappelen in that he explicitly holds that philosophical methodology does involve an appeal to facts like X intuits that p as evidence for p itself, and that such an appeal is legitimate, when intuition is understood in Devitt’s sense. The role Devitt ascribes to intuition, however, is much more limited than a traditionalist might hope. On Devitt’s picture, intuition provides a starting point for more systematic inquiry. Various individuals in the community are more or less masters of different terms, like “grass” or “rock” or “elm” or “quartz.” Depending on the quality of their underlying theories (recall that for Devitt, intuition is theoryladen), these different individuals are better or worse sorters of objects into these categories. Insofar as they are fairly good sorters of a category F, their intuitions will be a good preliminary guide to identifying Fs, so we may inquire further into the nature of the Fs. A deep understanding of the category F, however, involves knowing what is “common and peculiar” to the Fs. In the absence of developed scientific theory, our grip on the boundaries of F-hood is largely beholden to intuition; and the intuition of experts is given greater weight, due to the superiority of their theories. However, even expert intuitions are not decisive with regard to what is and is not an F. Sometimes our science is sufficiently advanced for us to move beyond intuition—we then reason consciously from our scientific theories and our initial intuitive sorting



Intuitions and the Theory of Reference

137

to discern the underlying nature of the Fs. In such cases, scientific theory trumps even expert intuition. Devitt therefore rejects the “method of cases” as MMNS characterize it. We’re attracted to a very “thin” conception of intuition, as Devitt is. Devitt (forthcoming) makes a good case that the common, core meaning of “intuition” is “immediate judgment that’s not the result of conscious reasoning.” Given our commitment to the heterogeneity of the psychological processes that produce such immediate judgments, we think such a “thin” account is the best account, and should be adopted.12 Devitt additionally holds intuitions to be theory-laden and driven by central (as opposed to modular) processing. We, on the other hand, find it likely that some intuitions are innate, largely encapsulated, and “theory free” (cognitively impenetrable)— for instance, the effortless and instantaneous numerosity judgments we make in response to being shown a collection consisting of four or fewer objects. Other intuitions are, as Devitt says, theory-driven and acquired through experience. On our liberal understanding of “intuition,” it’s almost inevitable that philosophers are constantly relying on intuitions during their theorizing, both in the theory of reference and in other fields. After all, they surely rely on, say, immediate concept-application judgments—it’s hard to see how anyone could reason without doing so. Indeed, we’re essentially including much of Cappelen’s “common ground” in the category of “intuition.” It would be quite cheap to use such a definition in service of a rejection of Cappelen’s arguments, and we don’t intend to do so. We simply hold, with Williamson, that the states philosophers are inclined to call “intuitions” are so heterogeneous that no thicker definition will cover them all. Inevitably, our ultra-thin definition lets in quite a lot of ordinary judgment, and therefore makes it almost trivially true to say that philosophers use intuition in their theorizing. There remains the more interesting question of whether theories of reference rest on intuitions as their primary, or even their exclusive, source of support. We’ll deal with this in section 2.2. Of course, Williamson, Cappelen and Deutsch would note that, even if philosophers rely on intuitions in their theorizing (in the sense that intuitions are causally responsible for many of the judgments philosophers make), this doesn’t show that philosophers appeal to intuitions as evidence (in the sense

138

Advances in Experimental Philosophy and Philosophical Methodology

that intuition-facts serve as premises in their argumentation). And this is fair enough; we’ll return to this issue in section 2.3. But for now, it’s worth noting that we think that the former claim is enough to justify the project of experimental philosophy. If a philosopher holds premise p because she has had an intuition that p, and if there is empirical evidence that intuitions about p tend to be subject to biases, cultural variation, etc., then we have prima facie reason to be concerned that those biases may have affected the philosopher’s judgment in a problematic way. We say prima facie, for of course the judgment may well still be true and the philosopher may still have good (intuitionindependent) justification for judging as she does. We don’t, then, think that experimental findings provide sufficient reason to immediately cease relying on intuitive judgments—they merely warrant concern, and ought to prompt further investigation and reflection on our methodology. Further, since we take intuition to be utterly heterogeneous, we don’t think that any particular experiment is ever likely to cast doubt on intuition as a whole—some types of intuitions are likely to be largely bias-free, while others may be riddled with epistemological flaws. But all this means is that experimental philosophers have more work on their hands than they may have hoped.

2.2 Intuitions and the method of cases Ichikawa et al. do not provide much by way of explicit views on the overall role of intuition in the paper we’ve been discussing. However, at least two of the authors (Ichikawa and Weatherson) have written independently about intuitions and their evidential role. Here we will focus on a particular aspect of the view Weatherson advocates in his (2003). There, he puts forth a moderate view on intuition, according to which theoretical considerations often ought to trump intuition. Indeed, he argues that theoretical considerations should lead us to reject the Gettier intuition in favor of the classical JTB account of knowledge, on the grounds that JTB possesses greater simplicity and naturalness. The correct account of knowledge, Weatherson holds, is one that best balances our intuitions with various important theoretical constraints. Contrast this with the “method of cases” as characterized by MMNS, according to which the best theory is simply the one that captures the most intuitions.



Intuitions and the Theory of Reference

139

We take this contrast to be important, because one of the principal arguments offered against Centrality is that, in the contexts where philosophers have been alleged to appeal to intuitions (e.g. in Kripke’s Gödel/ Schmidt case or in Gettier’s ten coins case), the relevant philosophers present independent arguments for their conclusions. Cappelen goes so far as to claim that if a philosopher gives significant argumentation for a proposition P, that is evidence that the philosopher does not rely on intuition as evidence for P. Cappelen characterizes proponents of Centrality as ascribing to intuition a “Rock” evidential status—intuitions justify, but themselves need no further justification. Since (good) philosophy centered around thought experiments typically involves arguments, Cappelen holds that the Centrality proponent’s picture of philosophical methodology is false. We agree that sometimes, some experimental philosophers can make it seem as though (for example), Kripke and Gettier simply presented thought experiments, had some intuitions about them, and went home without providing any reasons beyond their intuitions for their conclusions. Such a picture obviously bears no resemblance to the texts in question. But, to us, this fact merely supports the claim that philosophers don’t use the method of cases; it doesn’t support the claim that intuitions aren’t evidence in philosophy. We’re inclined to think that, in actual practice, things more often approximate the ideal Weatherson suggests—that is to say, intuitions are weighed against competing theoretical considerations, and often rejected on those grounds. A good argument can lead us to reject an intuition; but, if that is so, then it stands to reason that a good argument can also be used in support of an intuition. Perhaps this undermines Centrality insofar as we understand that thesis as requiring that intuitions be the primary source of evidence for philosophy. But if experimental philosophy’s importance hinges on any version of Centrality, it certainly doesn’t hinge on that version. The actual extent to which philosophical practice approaches Weatherson’s ideal is likely to vary from field to field. In some fields, intuitions may be given heavy weight, while in others they may be easily rejected (indeed, Weatherson suggests as much himself). So let’s consider how theoretical considerations and intuitions interact when it comes to the theory of reference. Consider the thesis that semantic reference is speaker reference—that a name N refers to object O in virtue of the fact that a speaker intends to refer

140

Advances in Experimental Philosophy and Philosophical Methodology

to O by his use of N. This view is likely to line up quite well with many people’s judgments about particular cases, since even relative experts (like undergraduates who’ve completed a philosophy of language course) still struggle making the distinction. So if we surveyed a wide array of native speaker judgments, and we had good reason to think those judgments corresponded to their intuitions, and we used the method of cases, we might well arrive at the view that semantic reference is after all speaker reference. There are, however, strong reasons—specific to semantic theory—that lead us to reject the theory that best fits speakers’ intuitions. Suppose that, up until now, S has used “Bob” to refer to some individual X. On one occasion, S speaker-refers to some Y ≠ X using “Bob.” Accepting the identification of speaker reference with semantic reference would lead us to posit that S has two homophonic names “Bob”—one of which semantically refers to X, and the other of which refers to Y. This might be seen as at odds with Grice’s razor: don’t postulate senses beyond necessity. Additionally, it might be seen as at odds with certain Fregean principles, e.g. that semantic content should be preservable and communicable; for instance, S’s morning diary entry about “Bob” won’t be about the same person as his evening diary entry about “Bob.” Finally, the proposal will introduce massive indeterminacy into our theories. Suppose S wonders aloud “What’s Bob doing now?,” without any additional commentary and without any associated mental imagery. Which “Bob” is he wondering about, and, in virtue of what is it, say, X rather than Y13? Semantic theorists invoke a plethora of principles—Grice’s razor, semantic innocence, charity, compositionality, theoretical simplicity, and so on—both in arguing for and against certain theories, and in deciding what to say about different cases. It seems false, then, that the accepted method of theory construction among semanticists is anything like the “method of cases” whereby theorists simply try to fit the theory to the intuitions. Theorists of reference are often quite happy to reject intuitions in favor of theoretical considerations like those just mentioned. Consider, for instance, Kripke on unicorns. Kripke writes that “Some of the views that I have are views which may at first glance strike some as obviously wrong,” and that his “favorite” such view is his denial of the claim that “under certain circumstances, there would have been unicorns” (pp. 22–3). This wouldn’t be coherent if Kripke were just trying to capture the intuitions.



Intuitions and the Theory of Reference

141

Kripke himself would probably predict that, if you did an experiment and tested people’s reactions to unicorn cases, the intuitions would not support him. And yet he does not on this basis abandon his theory or modify it so that it does not have this consequence. Kripke never says why he ignores the intuitions in this case, but there are several responses available to him: he could say, for instance, that his theory has enough intuitive support, and that modifying it to handle specific cases would violate various principles of theoretical beauty and simplicity. Alternatively, he could say that, as reliable as intuitions are in central cases, they aren’t of any (or much) evidential weight in arcane cases. Such moves are inconsistent with the method of cases as MMNS characterize it, but they aren’t in our view inconsistent with intuitions playing a role in semantic theory-building.

2.3 Intuition facts and language facts Martí’s critique of MMNS does not provide an explicit picture of the role of intuition in the theory of reference, but we can extract a bit from what she does say. Consider Martí’s suggestion that we employ experiments to elicit reactions to uses of the name “Gödel,” rather than intuitions about what the name “Gödel” refers to. On one reading, Martí is suggesting that we simply elicit a different type of intuition—and indeed, this seems to be supported by certain lines in the original text: “[MMNS’s prompt] does not test the right kind of intuitions. It does not test the intuitions that could allow us to tell whether or not the participants in the experiment use names descriptively” (Martí 2009: 44). In fact, this is how Machery, Olivola and De Blanc (2009) (hereafter MOD) interpret her in a paper responding to Martí’s criticisms. They characterize Martí as rejecting “metalinguistic” intuitions (intuitions about semantic properties such as reference) in favor of “linguistic” intuitions (intuitions about individuals and their properties). In other words, they characterize Martí as rejecting intuitions about “Gödel” in favor of intuitions about Gödel. In this vein, they performed a follow-up experiment in which a speaker in the vignette makes an utterance using the target name—e.g. “Gödel was a great mathematician.”14 Rather than asking who the speaker refers to, MOD instead asked subjects whether the speaker’s utterance was true or false. MOD found no significant difference between responses to the “linguistic”

142

Advances in Experimental Philosophy and Philosophical Methodology

vignettes and the responses to the corresponding “metalinguistic” vignette in the original prompt. Martí (2012) objects that MOD’s case asks subjects to reflect on another person’s use of a name; it does not prompt them to use the name themselves. What is really needed, Martí claims, is data about the subjects’ own uses of the name in question. Insofar as this is what Martí has in mind, this perspective fits quite closely with Deutsch’s and Williamson’s suggestions that our evidence consists of facts about P, rather than intuitions about P. Arguments in support of a given theory of reference should invoke premises concerning reference-facts, not intuition-facts; thus the appropriate data for a theory of reference consists of cases of referring, not intuitions about cases of referring. And here we are fully in agreement. When one is embarked upon construction of a theory of, say, porcupines, it is obviously preferable to observe porcupines rather than merely asking folks how many spines something must have to count as a porcupine. Similarly for a theory of reference; ideally, one gathers data by observing cases of referring. Yet, Martí also suggests that (the right kinds of) intuitions are evidence; and this at least prima facie conflicts with the position just outlined. Here, though, we should note that the case of reference is somewhat unique, in that most cases of referring also happen to be cases of humans expressing judgments15— often, spontaneous judgments that we would classify as intuitions. We might, then, make sense of Martí’s suggestion by saying that intuitions do have an evidential role to play in theorizing about reference—not qua intuitions, but qua instances of the phenomenon under observation. But in fact, we still think this isn’t quite right. The data in question aren’t intuitions themselves, but verbal expressions of intuitions. What we’re really interested in is the linguistic behavior; it’s neither here nor there whether the mental state which caused the linguistic behavior has whatever features it must have to count as an intuition. Thus, while we think there’s something very right about Martí’s suggestion, we also think such surveys involve an “evidential role” for intuition only in a fairly minimal, roundabout way. From the perspective of the theory of reference, it doesn’t matter whether surveys elicit intuitions or reasoned judgments. If you’ve gotten your subjects to use the name in question, you’ve got perfectly good evidence in the form of speakers applying the name to a thing. That’s the core data for the theory of reference. The method is called elicited production, and it’s a tool that linguists frequently make use of.



Intuitions and the Theory of Reference

143

It is important to keep in mind, too, that straightforwardly applying a type of “method of cases” to such survey data would be laughably bad methodology. We can’t just collect a bunch of instances of speakers applying names to things and find the theory that best fits those applications. This is because, patently, speakers incorrectly apply names to things, and they do so sometimes systematically. Sometimes they misapply names because they have inferred who their bearers are from false principles; other times they misapply names because they spontaneously judge the name applies, but would realize it did not, had they had more information or spent some time thinking about the case. As always, we take our data and our theoretical considerations, and then find the best set of theories that explain some of the data, explain away other of the data, and satisfy best our theoretical considerations. It’s hard work. So what of the philosopher in her armchair—must she refrain from claims about reference facts until the surveys have come in? Not necessarily. It’s not like the armchair philosopher has no evidence whatsoever; in fact, more so than in the survey case, intuitions really are a form of evidence here—though perhaps not ideal evidence. Suppose the philosopher intuits in response to a case that she would say thus-and-so. That’s very good evidence that she would indeed say thus-and-so. She could even say it aloud to the empty room and have an instance of her applying a name to a (hypothetical) thing. This is evidence that, in her idiolect, the name applies to the thing. Of course, there may be theoretical reasons for rejecting this intuition-caused bit of verbal behavior. But, being a theorist of reference, she is in an excellent position to reflect on this data point’s fit with various theories, and those theories’ fit with various theoretical considerations, and to reach a non-intuitive, reasoned conclusion about the case. And this is even better evidence concerning what refers to what. Is it sufficiently strong evidence to wholly obviate the need for systematic survey data? It seems to us not, but there are complicated epistemological considerations that will hinge on the details of the case and the centrality of the case and the extent of the disagreement and the available methods for explaining away certain data points. To sum up: Martí wants empirical studies that are designed to elicit speakers’ uses of names in response to vignettes. This is a wholly appropriate methodology for the theory of reference, called elicited production. It’s

144

Advances in Experimental Philosophy and Philosophical Methodology

appropriate not because it elicits the “right” type of intuitions, but because it elicits the central data for the theory of reference: speakers applying terms to things.

3. Making room for thought experiments So far, we have endorsed what is, from a traditionalist point of view, a fairly Centrality-unfriendly position. We’ve agreed that the term “intuition” is a bit of a mess, and we’ve agreed that there’s no obvious way to carve off “philosophical intuitions” from the swaths of mundane categorization judgments that comprise much of our ordinary cognition. We’ve agreed that actual philosophical practice is a far cry from MMNS’s characterization of the “method of cases,” involving as it does a great deal of argumentation and weighing of various theoretical considerations and constraints. And we’ve agreed that when studying some phenomenon P, the most immediately relevant data consists of facts about P, not facts about intuitions about P. Nonetheless, we do have points of disagreement with the opponents of Centrality. Unlike Cappelen, we hold that intuitions are best defined as spontaneous, unreflective judgments. And unlike all opponents of Centrality, we hold that such judgments are frequently relied upon (if not appealed to as evidence) by philosophers when they defend their views. Finally, we hold that empirical evidence about such judgments can at least in principle have important consequences for assessments of our current methods. With regard to theories of reference in particular, we have argued that survey methods are in fact the best possible source of evidence, with the method of elicited production being the appropriate model for such data collection. We’ve also (contra both opponents of Centrality and many experimentalists) argued that armchair appeal to intuition is in fact a viable source of evidence for theories of reference, though elicited-production evidence is generally superior when available. We’d hazard a guess that both Martí and Devitt, at least, would be more or less amenable to these conclusions; in fact, our considered position on the role of intuition in theories of reference comes closest, we think, to Devitt’s. However, contra Devitt, and contra at least some proponents of negative experimental philosophy (e.g. Weinberg 2007),



Intuitions and the Theory of Reference

145

we want to argue in favor of the value of eliciting intuitions about unusual, far-fetched, non-“humdrum” hypothetical scenarios. Skepticism about cases that are merely counterfactual, hypothetical, or fictional seems, at first glance, a tad strange. The Peano/Dedekind case is none of these things, yet it’s hard to see how that alone makes intuitions about it evidentially superior to those elicited by the Gödel case. In a fictional case you make up characters (like Schmidt) and sometimes ascribe counterfactual properties to other characters (e.g. Gödel being a thief). But most English speakers haven’t heard of Peano, Peano’s axioms, Dedekind, or Dedekind’s original formulation of Peano’s axioms. For them, those things might as well be made up. The difference between the two cases is really the word “suppose” out front, and, for the vast majority of people you might hand a prompt to, they wouldn’t notice if that word went missing. Of course, presenting cases to naïve subjects wouldn’t be Devitt’s preferred way of attempting to resolve metasemantic questions. Instead, he would presumably appeal to the simple fact that people who learn first about Peano’s axioms, then later discover that Dedekind originally formulated them, don’t spontaneously blurt out things like, “Peano was German!” The observed absence of such utterances in ordinary conversation is more relevant than the linguistic intuitions of confused neophytes filling out philosophy surveys. Here we have the linguistic reality itself, available for direct investigation. And so it might be thought that people’s utterances about the actual world and actual objects and events within it are of central importance, and people’s judgments about hypothetical cases more peripheral. However, hypothetical and counterfactual cases or, at least, cases involving telling people a lot of stuff they might currently believe the opposite of, are arguably central to adjudicating questions in the theory of reference—and, if this is so, we may need to crucially rely on judgments in response to cases in addition to simple observation of linguistic behavior “in the wild.” A brief survey of the cases Kripke and Putnam adduce against descriptivism for natural kind terms reveals bizarre scenarios like duplicate planets full of chemically impossible substances, pencils that are secretly animals, atmospheric disturbances that make blue gold appear yellow, or three-legged tigers appear four-legged, and so on. This isn’t philosophers taking shortcuts with

146

Advances in Experimental Philosophy and Philosophical Methodology

convenient thought experiments rather than investigating real world cases. We can’t just look at how people use words in actual scenarios. Why not? Consider the case of fish. Let’s suppose that English speakers use “fish” to refer to whatsoever kind, in the historical chain of usage of that word, samples of which were baptized “fish.” Actual usage of “fish,” however, is going to be wildly divergent, and not just because sometimes a boot on your line might tug like a fish. Some individuals are going to call whales “fish,” because they are under the impression that whales and fish form a natural kind that excludes things like pigs and lemurs. Others are going to not call whales “fish,” because they think whales are of a kind with pigs and lemurs, but not rays and lungfish. Still others might call whales “fish,” because they think any kind including rays and lungfish must include whales, pigs, and lemurs (e.g. if they think biological kinds are clades). Without knowledge of the beliefs that drive a person’s applications of “fish,” the data they produce—those applications— are at best murky with regards to their relevance. Hypothetical scenarios are a means of controlling for what people know or believe about a case. If you make up a case and then stipulate all the facts of it, then you know precisely what information is driving subjects’ applications of terms in the case. You might have stipulated impossible things, or you might have failed to stipulate enough things to decide the questions you’ll be asking your subjects. But this form of inquiry—asking people to respond to hypothetical scenarios—is of paramount importance in the theory of reference.16 This is not to say that hypothetical cases trump “the linguistic reality itself,” or that people’s actual applications are irrelevant. Both matter. What a word does refer to is different from what people think it would refer to, were they to know all the relevant facts. People might think that they’d call mammals “fish” were cladism to be true, but then fail to do so, even after accepting cladism— in such a case their intuitions would clash with the observed linguistic reality. In such cases linguistic reality ought to trump intuition; however, for many scenarios of interest little to no linguistic reality exists to be observed, and thus we must fall back on the use of hypothetical cases. In other words, Devitt is right about the decisiveness of various humdrum cases; the Peano case is so compelling because it’s actual, and because it’s unlikely that the widespread use of “Peano” to refer to Peano is the result of some misconception among



Intuitions and the Theory of Reference

147

users of that name. But especially for natural kind terms, how people actually use the terms can be highly equivocal and extremely misleading. The theory of reference requires a balance between real cases and hypothetical cases (which is precisely what Kripke provides us). Judgments about what refers to what, or what applies to what, or what is true, depend on a wide range of ordinary factual beliefs, like (possibly) who wrote what theorem when, or whether whales are more closely related to lungfish than the latter are to rays. In syntax, you only have to look at what sentences people produce. You assume there’s some noise in the channel, but you don’t assume that the quality of the evidence varies greatly from time to time, circumstance to circumstance, and person to person, as the knowledge bases of those people in those circumstances at those times vary. That a sentence was uttered is basically all you need to know about it for purposes of syntactic theorizing—not who uttered when and what they believed at that time. This is patently not true for semantics. Whenever traits are hidden at all behind the appearances, we have to fallibly infer from the appearances to the traits, and such inferences are infected with our beliefs, which vary wildly from person to person, time to time, and circumstance to circumstance. At one time, all whales were called “fish.” In no sense does it follow from this that at that time, “fish” applied to whales. Compare syntax: all English sentences have subjects. This is very good evidence that English isn’t a pro-drop language. No one seriously considers the possibility that English is really like Spanish, but, due to some false belief, English speakers put in subjects when they don’t have to. It is a live option, however, that, due to some false belief, everyone misapplies “fish.” This doesn’t have anything particular to do with natural kind terms either. Everyone quite frequently misapplies “smart” and “friendly” and “Tuesday” as well. It’s not always obvious whether it’s Tuesday or not. As we’ve noted, devising theories of reference and application requires controlling for what people believe. It’s not requisite that people have true beliefs. You can test whether Kripke and Putnam’s judgments about natural kinds are right without completing science and then explaining it all to your undergraduate subjects. All you need to do is stipulate, in a particular case, which things are the kinds, and which things satisfy which descriptions, and then see how people apply terms to things so-described.

148

Advances in Experimental Philosophy and Philosophical Methodology

In sum, MMNS are right that getting people to report their judgments about cases—sometimes strange hypothetical cases—is the primary methodology in the theory of reference. What they’re wrong about is their claim that we have to take these data at face value (i.e. we have to employ the method of cases). Improperly controlled thought experiments aren’t particularly telling. And, as Martí points out, MMNS’s prompts are improperly controlled, as they don’t control for which theory of reference (if any) the subjects explicitly hold. Data gathering for the theory of reference from naïve subjects is extremely tricky. Whatever details you don’t supply to subjects are ones they might fill in in different ways. To be clear, this case is tricky. You don’t control for differing explicitly held theories of reference by stipulating that one of them is true in the thought experiment. If you did that, you’d only get out of your surveys what you put into them. But it’s not better to get out of them what your subjects willynilly put into them. What you want is to instruct the subjects to give you a judgment on the case without deducing it from their explicitly held theory of reference (if any), and that’s a hard instruction to convey to naïve subjects. As we mentioned earlier, Martí seemed to think that one could obviate the need for such an instruction by getting subjects to use the name, but we don’t see why that would be. This is one positive of armchair theorizing about reference: philosophers at least understand that, when considering these cases, they ought not to be deducing what refers to what from their prior commitments in the theory of reference. Not if what they’re trying to do is to determine whether those commitments are correct.

4. Conclusion On our view, verbal reports of judgments about appropriately elaborate hypothetical cases are the central data for the theory of reference, for they are instances of speakers applying terms to things that have been generated under controlled conditions to test the predictions of different theories. Philosophers frequently substitute their own judgments for the more general “what we would say” about the case. This has some virtues, as we’ve argued, for philosophers have a good grasp on what information can and can’t be brought to bear



Intuitions and the Theory of Reference

149

in this activity, but it does run the risk that philosophers are a poor barometer of “what we would say,” or perhaps that different groups would say different things. Additionally, verbal responses to hypothetical cases are not the only relevant data—all cases of referring count, whether or not they were prompted by intuition; and responses to actual cases are in some respects better than hypothetical cases (although in some respects worse, as well). Finally, the theory that best fits that data is not necessarily the best theory overall, for there are various theoretical considerations that may necessitate rejection of certain judgments as unreliable. Given that verbal reports of judgments about hypothetical cases are undoubtedly the sort of thing that experimental philosophers study, experimental studies are relevant to evaluating theories of reference, and may indeed be necessary, if it turns out that different demographics would say different things about different cases, even under the appropriately controlled conditions. Nevertheless, because theoretical constraints play a role, and because of the difficulty in adequately controlling naïve subjects’ beliefs, armchair argumentation and seasoned philosophical judgment are also necessary components of theory construction. Thus we have provided a limited defense, with respect to the theory of reference, of (a) relying on armchair judgments about cases and (b) using experimental methods. But this defense does not straightforwardly extend to other domains of philosophy. As we noted earlier, the primary data for a theory of reference consists of cases of people applying terms to things—and such applications are standardly caused by judgments, many of which are (or perhaps are themselves caused by) intuitions. It makes sense, then, that intuitions would provide some evidence about the nature of reference—since, when one intuits that a term T refers to P, one is also highly likely to apply the term to P. But this is a peculiarity of studying human language; most other philosophical fields plausibly lack such a straightforward link between intuition and their target phenomena. Intuitions frequently cause applications of terms to things, but they do not frequently cause instances of other philosophically relevant phenomena like time, consciousness, scientific progress, or God. Nonetheless, there is a complication here. Consider epistemology: specifically, suppose someone reads a Gettier case and intuits that it is a case of knowledge. On our account this is at least some evidence for the theory

150

Advances in Experimental Philosophy and Philosophical Methodology

of reference, supporting the claim that the individual would likely apply “knowledge” to that case, and thereby the claim that “knowledge” in fact applies to that case. But then it follows that the intuition is also equally good evidence that that case is knowledge, because anything that “knowledge” applies to is knowledge. So doesn’t the evidential link between intuition and referring have to be exactly as strong as the evidential link between intuition and literally anything else? In a sense yes, and in a sense no. As we’ve noted, though instances of persons applying terms to things are the primary data for a theory of reference, such data have a greater potential to mislead than, e.g. the data for a theory of syntax. A given piece of linguistic data can of course fail to reflect the grammatical facts due to performance error; but there is no possibility that English is in fact a subject-object-verb (SOV) language despite the fact that every native English speaker employs subject-verb-object (SVO) word order. By contrast, “fish” fails to apply to whales despite the fact that there was a time when (more or less) every native English speaker applied “fish” to whales. The facts about what refers to what are hostage to certain extra-linguistic facts in the way that the facts about what is grammatical are not. Importantly for our purposes, our access to these extra-linguistic facts varies dramatically. Contrast, for instance, “water” with “Saul Kripke.” Before anyone could determine the correct application of the term “water,” quite a lot of science had to be done—not so for “Saul Kripke.” Note that this is not because it’s an analytic, a priori truth that “Saul Kripke” refers to Saul Kripke, or anything like that; it’s simply that the relevant empirical facts are (assuming no Gödel-type case holds for Professor Kripke) more readily available. Note also that, prior to doing the actual science, probing people’s intuitions with hypothetical twinearth scenarios would not have told us what water in fact refers to. It would at best have told us what water would refer to if the scientific facts happen to turn out to be such-and-so. There continue to be a large number of terms—for instance, many terms from the sciences—for which we simply cannot yet determine their correct application. But we don’t need to determine the correct application of every single term in order to develop a successful theory of reference. We can plausibly develop our theory using only the cases (like that of “Saul Kripke”) where our access is reasonably good, and our confidence that we’ve gotten



Intuitions and the Theory of Reference

151

things right is therefore reasonably high. For the theory of reference, we (quite fortunately) probably have enough non-misleading data to make progress. Notice that we’re not just making the general claim “use uncontroversial, central cases to develop a theory” here—prior to modern chemistry we had a plethora of uncontroversially correct judgments about water, but this failed to be sufficient for determining the nature of water. What, then, does any of this imply about the use of intuitions in other fields of philosophy? Our intuitions will fail to be a good guide to the philosophical phenomena in cases where the correct application for the relevant term is dependent on extra-linguistic facts to which we have poor access—be these facts about the direction of future science, facts about the linguistic usage of others in our community, or even facts about undesirable consequences of a given theory that no philosopher has managed to suss out yet. It’s quite likely, then, that different areas of philosophy will vary in the degree to which intuition is a good source of evidence. We’d be willing to bet, for instance, that the correct application of “time” and “consciousness” are quite heavily dependent on highly non-accessible extra-linguistic facts—and that intuition is therefore of little evidential use in the study of time and consciousness. All of this also implies that the relevance of at least some types of experimental philosophy will differ from field to field. We’ve argued that experimental work of broadly the sort that MMNS embarked upon is crucial to a successful theory of reference; by contrast, recent studies of folk notions of consciousness (e.g. Knobe and Prinz 2008; Sytsma and Machery 2010) tell us essentially nothing about consciousness. Such studies do have value, but their value largely consists in contributing to the inherently interesting project of characterizing our everyday cognition about philosophical phenomena— which in turn may (or may not) reveal something interesting about how philosophers have come to the theories they hold. Of course, much of the controversy over experimental philosophy concerns not “positive” projects like the one just characterized, but the “negative” project of casting doubt on intuition. On our view, such projects are more or less generally legitimate—but they must be qualified in certain ways, and the conclusions they warrant will likely differ for different fields. As we’ve already noted, casting doubt on “intuition” in our ultra-thin sense is a hopelessly overbroad goal. But there’s quite a lot of sense in, say, studying the psychological

152

Advances in Experimental Philosophy and Philosophical Methodology

processes underlying intuitive moral judgments and using those findings to critique the practices of ethicists; and mutatis mutandis for other fields or sub-fields. The arguments of the anti-Centrality folks and of the critics of MMNS show, however, that the arguments here must be careful—evidence that intuitive moral judgments are affected by emotional state, e.g. don’t imply that the entire project of ethics is doomed to failure. Yet insofar as such judgments causally influence philosophers’ arguments—and we think it is clear that they do, at least to some degree—there’s cause for concern. Our painfully unexciting prediction: for some areas of philosophy, some amount of modification to some of our methods will likely be needed. Given the complexity, subtlety, and diversity of the methods and tools of philosophy, no broader conclusions are likely warranted.

Notes   1 Thanks to Max Deutsch and Michael Devitt for helpful comments on an earlier draft.   2 The work described in this chapter was partially supported by a grant from the Research Grants Council of the Hong Kong Special Administrative Region, China (Project No. LU 359613).   3 Though Deutsch has also argued directly against MMNS, he was writing before the publication of, e.g. Devitt’s and Martí’s arguments and thus does not engage with them. We have categorized him as a participant in the former debate due to the broad scope of his argumentation.   4 For the central works here, see Williamson (2007), Deutsch (2009, 2010), and Cappelen (2012).   5 Deutsch does not—he is perfectly happy to speak of intuitions, he simply thinks they do not play an evidential role.   6 This is not the only possible characterization of the “method of cases.” Cappelen’s characterization is as follows: “T is a good theory of X only if it correctly predicts our intuitions about X-relevant cases.” This imposes only a necessary condition, and makes no claim that the “best fit” theory is automatically the winner.   7 Deutsch (2009: 446) also makes this point.   8 Of course, Deutsch and Cappelen would disagree that Kripke’s argument here relies on intuition. However, in this section our goal is merely to articulate



  9 10 11

12

13

14

15 16

Intuitions and the Theory of Reference

153

various positions in the debate over the Gödel case, and not to argue for or against the claim that Kripke relies on intuitions as evidence. See also Devitt (forthcoming) for an extended critique of Cappelen and Deutsch. At least with regard to the theory of reference. To clarify the dialectic—Cappelen’s position is that “intuition,” as used by non-philosophers, does not serve to pick out a special kind of mental state. It is instead used, e.g. for hedging one’s claims. Insofar as “intuition” is used in a technical sense in philosophy, it is a defective technical term lacking a well-agreed upon definition. Of course, regardless of the status of the term ‘intuition,’ it may well be that philosophers make evidential use of states possessing some of the features commonly ascribed to intuition. To argue against this possibility, Cappelen outlines the above-mentioned diagnostic criteria for intuition and argues that philosophical practice shows no reliance on states possessing those features. The account might be refined a little: an intuition is an unreflective judgment that is not produced by memory, not produced by vision, not produced by proprioception … This isn’t merely the result of the existence of homophonic names. In standard cases of homophones, there’s a perfectly good way to say which of them is being used on each occasion—look at what the speaker intended. But in this case, S doesn’t realize he has two names ‘Bob’ referring to X and Y. So nothing about his mental states can tell us who he’s referring to on any occasion. The actual prompts used by MOD concerned the Chinese astronomer Tsu Ch’ung Chih rather than Gödel—this was an alternate case used in MMNS’s original experiment. Some are not—instances of referring also occur in, e.g. questions or exclamations. One could get at the beliefs driving a person’s applications more directly—say, by asking them whether they believe in cladism, etc. But use of hypothetical scenarios is a more efficient way of getting at the data we need, for it has the same virtues of controlled experimentation over mere observation. Suppose we are interested in the effect of caffeine on concentration; we could go search for people who have had exactly three cups of coffee in the past eight hours and measure their concentration, or we could take matters into our own hands and give subjects coffee. Similarly, by presenting a hypothetical scenario we are asking subjects to adopt certain beliefs (about the hypothetical scenario), rather than searching for subjects who already have those beliefs (about reality).

154

Advances in Experimental Philosophy and Philosophical Methodology

References Cappelen, H. (2012), Philosophy without Intuitions, Oxford: Oxford University Press. Deutsch, M. (2009), “Experimental Philosophy and the Theory of Reference,” Mind and Language 24.4: 445–66. Deutsch, M. (2010), “Intuitions, Counter-Examples, and Experimental Philosophy,” Review of Philosophy and Psychology 1.3: 447–60. Devitt, M. (2006), “Intuitions in Linguistics,” The British Journal for the Philosophy of Science 57.3: 481–513. Devitt, M. (2011a), “Experimental Semantics,” Philosophical and Phenomenological Research 82.2: 418–35. Devitt, M. (2011b), “Whither Experimental Semantics?,” Theoria 72: 5–36. Devitt, M. (forthcoming), “Relying on Intuitions: Where Cappelen and Deutsch Go Wrong,” Inquiry. Ichikawa, J., Maitra, I., and Weatherson, B. (2012), “In Defense of a Kripkean Dogma,” Philosophical and Phenomenological Research 85.1: 56–68. Kripke, S. (1980), Naming and Necessity, Cambridge, MA: Harvard University Press. Lam, B. (2010), “Are Cantonese Speakers Really Descriptivists? Revisiting CrossCultural Semantics,” Cognition 115.2: 320–9. Ludwig, K. (2007), “The Epistemology of Thought Experiments: First Person versus Third Person Approaches,” Midwest Studies in Philosophy 31.1: 128–59. Machery, E., Mallon, R., Nichols, S., and Stich, S. (2004), “Semantics, Cross-Cultural Style,” Cognition 92.3: B1–12. Machery, E., Olivola, C. and De Blanc, M. (2009), “Linguistic and Metalinguistic Intuitions in the Philosophy of Language,” Analysis 69.4: 689–94. Martí, G. (2009), “Against Semantic Multi-Culturalism,” Analysis 69.1: 42–8. Nado, J. (2014), “Why Intuition?,” Philosophical and Phenomenological Research 89.1: 15–41. Weatherson, B. (2003), “What Good are Counterexamples?,” Philosophical Studies 115.1: 1–31. Williamson, T. (2007), The Philosophy of Philosophy, Hoboken, NJ: John Wiley.

8

Intuitive Evidence and Experimental Philosophy Jonathan Jenkins Ichikawa

1. Introduction No discipline that affords a central evidential role to an unreliable source is in good standing. So if analytic philosophy affords a central evidential role to intuitions, then if intuitions are unreliable, then analytic philosophy is not in good standing.1 So much, I think, is uncontroversial. Some of the early critiques of traditional analytic methods by experimental philosophers assumed that analytic philosophy gives intuitions central roles, and, citing data that suggested that intuitions may be unreliable, argued on these grounds that analytic philosophy might not be in good standing. For example, in their famous paper alleging cross-cultural diversity with respect to epistemic intuitions, Weinberg et al. describe their target thus: The family of strategies that we want to focus on all accord a central role to what we call epistemic intuitions. Thus we will call this family of strategies Intuition Driven Romanticism (or IDR). As we use the notion, an epistemic intuition is simply a spontaneous judgment about the epistemic properties of some specific case—a judgment for which the person making the judgment may be able to offer no plausible justification. To count as an Intuition Driven Romantic strategy for discovering or testing epistemic norms, the following three conditions must be satisfied: (i) The strategy must take epistemic intuitions as data or input. (It can also exploit other sorts of data.) (Weinberg, Nichols, and Stich 2001: 432)

In recent years, some defenders of traditional philosophical methodology have argued that critiques like this one are mistaken in assuming that

156

Advances in Experimental Philosophy and Philosophical Methodology

intuitions play central evidential roles in traditional philosophical methods. According to this kind of response, experimental philosophers attack a straw man; it doesn’t matter whether intuitions are reliable, because philosophers don’t use intuitions in the way assumed. Deutsch (2010), Williamson (2007), and Cappelen (2012) all defend traditional methods in something like this way. I also endorsed something like this line in Ichikawa (2014a). In this contribution, I will follow up on this sort of defense of traditional philosophical methods in three ways. In section 2, I will rehearse and extend some of my reasons for challenging the idea that traditional methods depend on intuitions in an evidential role. (My reasons are very different from those discussed in Cappelen 2012.) I will also engage with some recent more sophisticated attempts to establish the idea that intuitions play evidential roles in philosophy, such as that of Chudnoff (2013). In section 3, I will consider and argue against a dismissive response to such positions from experimental philosophers, who consider the question of philosophical reliance on intuitions to be irrelevant to the experimentalist critique. But, in section 4, I will argue that it would also be a mistake to conclude (as Herman Cappelen does) that the critique is rendered totally irrelevant by the denial of the evidential role of intuitions; I defend a more moderate view, on which the bearing of experimental studies of philosophical intuitions is relevant for philosophical methodology, but only in a relatively limited way.

2. Intuitions and evidence What would it be for intuitions to play evidential roles in philosophy? According to a psychologistic model, propositions of the form I have the intuition that p, typically known by way of introspection, play fundamental evidential roles. Jessica Brown (2011) defends such a model. According to her version of the view, in paradigmatic cases one achieves philosophical knowledge by beginning with psychological evidence concerning what intuitions one has, and bases one’s nonpsychological philosophical conclusions on such evidence. Since, Brown thinks, the psychological episodes that are intuitions are in fact reliable indicators of the truth of their contents, those contents can end up justified, or even known, on such grounds.2



Intuitive Evidence and Experimental Philosophy

157

The psychologistic model is perhaps the clearest instance of a view on which intuitions play central evidential roles, but it is not the only one. One may also hold that the mental episodes that are intuitions play important evidential roles without themselves being essential evidence—one might hold, for instance, that intuitions provide their contents as evidence. In general, the contents of intuitions are not psychological. For example, if I have the intuition that a statue would cease to exist when smashed, the content of the intuition is the nonpsychological claim about the statue. One might hold that intuitions always provide their contents as evidence, but I do not consider this view plausible; I think there are good reasons to think that (a) some intuitions have false propositions as contents,3 and (b) false propositions are never among one’s evidence.4 So a more plausible version of the nonpsychologistic view of intuitions is that in some suitable cases, one possesses a proposition as evidence by virtue of having an intuition with that proposition as its content. Even so, however, this claim must be interpreted carefully—in particular, that “by virtue of ’” must be read in a suitably strong way. Assuming that evidence can be established inferentially, for example, anyone should admit that there are some possible cases where the intuition that p plays a role in establishing p as evidence. For example, if one knows antecedently that my intuition about whether p will be a correct intuition, then establishing that I have the intuition that p will conclusively establish p.5 The interesting epistemic claim in the area is the one that there is a stronger connection between intuition and evidence—for instance, that it is in the nature of intuition that it provide, at least in central cases, its content as evidence. Or perhaps that there is a law or a general epistemic principle to the effect that intuitions’ contents count as evidence. It is not enough to establish merely that, combined with some possible background knowledge, an intuition may provide its content as evidence. (Combined with some possible background knowledge, anything can be evidence for anything logically compatible with it.) Benjamin Jarvis and I have called the view that intuitions play such evidential roles “experientialist rationalism.” We argue that all such views are false.6 Since this question plays a central role in the question of the relevance of experimental data about intuitions, it is worthwhile to summarize some of the considerations in favor of this claim here.

158

Advances in Experimental Philosophy and Philosophical Methodology

The acquisition of evidence in favor of p can play an important role in coming to a justified belief that p, but it isn’t the only relevant factor. In addition to having the evidence, a subject must respond appropriately to it. Bruce may have figured out that the thief is after the diamonds, while Dick still has no idea what’s going to happen next—this, even though they share all the same evidence. Acquiring the evidence gives one some of what one needs, but not everything; rational skill dictates how the evidence will be used. Jarvis and I use the term propositional justification for the state of having sufficient evidence to draw a particular conclusion, whether or not the person has responded to the evidence in the rational way. A belief that is properly formed on the basis of the relevant evidence is doxastically justified. The terminological decisions aren’t the important part—the point to emphasize is that evidence is relevant at that first stage of inquiry. Evidence for p gives one reason to believe that p, whether or not he or she actually believes it (or indeed, if he or she does actually believe it, whether or not he or she believes it for a good reason). With this background in place, we are now in a position to express a central argument against the idea that intuitions provide evidence in a priori cases. Consider this case, which I introduced in Ichikawa (2014b: 201–2): Boris and Natasha … are playing chess. Natasha is playing white; it is her move. The position of the pieces is as given [below]: Natasha has available a move to win the game: Rh3# (moving her rook to the far right of the board) places the black king in checkmate. Bullwinkle knows the rules of chess, but he is nevertheless unaware that Natasha has available a move to win the game. His ignorance shouldn’t be too surprising, because, it turns out, he is blindfolded, and can’t see the board. He has no idea where the pieces are, so of course he doesn’t know that Natasha is in a position to win. But what happens next is slightly more interesting. Bullwinkle’s blindfold slips off, and for the first time, he sees the board and the position of each piece. Now he knows a lot more about the game. But, let’s suppose, he still doesn’t know that Natasha is in a position to win. You see, Bullwinkle isn’t a very experienced chess player, and, although he knows where each piece is and knows all the rules of chess, he hasn’t noticed that Rh3 would put Boris in checkmate.

Boris and Natasha’s Game



Intuitive Evidence and Experimental Philosophy

159

The key moment to consider is the one after the blindfold slips. Now, Bullwinkle has the relevant perceptual experiences—he can see the board. Although he still doesn’t know that Natasha can win on this turn, he has conclusive evidence that she can. In the terminology above, he now has propositional justification for the proposition that Natasha will win, even though he doesn’t know it, or even believe it, even if he is contemplating the question. This constitutes a rational shortcoming; by remaining agnostic about that for which he has conclusive reason to believe, Bullwinkle demonstrates that he is a less than ideally rational agent. Importantly, Bullwinkle’s position manifests rational failure only after his blindfold slipped; even an ideal rational agent will be agnostic about whether Natasha can win this turn if he can’t see the board. But once the perceptual experiences obtain, Bullwinkle faces rational pressure towards accepting this conclusion. It is sufficient for this kind of rational pressure that Bullwinkle has the experiences that justify true beliefs about the state of the game—the rules of chess, the position of the pieces, that it is Natasha’s turn, and that she is playing white. We do not need to stipulate that Bullwinkle has an intuition with the content that Natasha can win this turn. One could consider such a version of the case—perhaps Bullwinkle has enough chess experience to have developed subpersonal intuitions of this sort, and he’s failing to take advantage of them in this instance—but it is not necessary to suppose that he has this kind of chess ability in order to establish the verdict that he is manifesting rational limitations. This means that there is rational pressure in this case that does not depend on the presence of intuition.7 And this seems to apply to all rational necessities—the truths of logic and arithmetic, so-called “conceptual truths,” the parts of philosophy people have called “analytic,” etc. This is my first argument against intuitions in evidential roles—the argument from “opaque irrationality.”8 There is rational pressure in favor of accepting these truths, even if subjects are unaware of such pressure, and don’t even feel an inclination to adopt it.9 But, once we think that in cases like Bullwinkle’s there is rational pressure that does not depend on intuitions, there seems to be little motivation to suppose that intuitions are playing central roles in those cases in which they are present.10 I suppose there’s no contradiction in the idea that, in a counterfactual version of the case where Bullwinkle does have the intuition, intuition

160

Advances in Experimental Philosophy and Philosophical Methodology

is playing justificatory roles, but in the version where it is lacking, something else plays those roles instead. Yet such a view seems more than a little ad hoc; I see nothing to recommend it. If the rational pressure is present regardless of the presence of the intuition, that’s excellent reason to suppose that the intuition doesn’t explain the rational pressure. Any role intuitions have in cognition, therefore, is quite unlike the role of perceptual experience. (The counterfactual version of the Bullwinkle story, in which Bullwinkle undergoes no visual experiences of the chessboard, is one in which he has no propositional justification for any particular claims about the state of the game.) Here is a second, closely related, argument for the same conclusion. Take a subject like the subjects just discussed—someone who fails to believe something for which he has conclusive reason to believe, and thereby manifests a rational deficiency, even though he has no intuition that clues him into this fact. Now suppose that, over time, his cognitive capacities change. For example, suppose that Bullwinkle gets better at imagining various possible chess moves, and recognizing important patterns of pieces. Now, if confronted like a board like the one above, Bullwinkle will have the strong intuition, which he’d go on to accept, that white can win on this move. Such a change constitutes a rational improvement; if so, the standards of rationality are not relative to these kinds of intuitions and cognitive abilities in the way they are relative to perceptual experience. Elijah Chudnoff (2013) has given the most thorough recent defense of a strong analogy between intuition and perceptual experience of which I am aware. It is useful to consider the argument above in the context of his defense of an evidential role for intuitions. Chudnoff argues for this principle: (DIJR) If it basically intuitively seems to you that p, then you thereby have some prima facie justification for believing that p. (2013: 94)

Chudnoff writes that (DIJR) is motivated primarily by reflection on examples: Take the proposition that every concave figure can be rounded out to a convex figure that bounds a greater area in a smaller perimeter. This is a good example because it is likely not something you have thought about before. Initially it neither seems true nor seems false to you, and you neither have justification for believing it nor have justification for doubting it. After thinking about it a bit, however, it should seem true to you. (2013: 94)



Intuitive Evidence and Experimental Philosophy

161

This “seeming true,” Chudnoff thinks, is an intuition experience, and it is what provides the justification for accepting that conclusion. I recognize something like the phenomenology Chudnoff is describing here—initially, at t1, one might have no idea whether a claim is true, and then at t2, in a flash of insight, one realizes that it must be true. And I also agree with Chudnoff that at t3, after this experience, one has justification to think that every concave figure can be rounded out to a convex figure that bounds a greater area in a smaller perimeter. But I do not agree with Chudnoff that the intuitive experience is the explanation for this justification. I think that considerations arising from opaque irrationality and rational improvement give us good reason to reject Chudnoff ’s claim that one has no justification at t1. Recall that the relevant notion of justification here is propositional justification, not doxastic justification.11 It is obvious that I had no doxastic justification at t1, since doxastic justification is a property of beliefs, and I didn’t have the belief at t1. The question is whether I had sufficient reason at t1 to believe that every concave figure can be rounded out to a convex figure. In other words, was my ignorance more like Bullwinkle’s ignorance before his blindfold slipped, or afterwards? Once the question is thus clarified, it seems to me clear that my failure at t1 is much more like Bullwinkle’s failure after he gains the relevant sensory experiences. My problem at t1 isn’t that I am lacking evidence—it’s that I am engaging with the question with less than fully ideal rationality. I need better reasoning, not better evidence. If so, the intuitive experience is playing a role very unlike that of perceptual experience in perceptual belief. It is not providing propositional justification.12 I consider it an open question as to which domains of philosophy have this status. I am open to the idea that there are some subfields in which psychological elements do occupy central evidential roles. For example, Regina Rini (2013) argues that moral psychology proceeds by way of an abstraction from one’s own moral commitments—the idea is that one engages in a theoretical process of reflection to determine which of one’s moral commitments are idiosyncratic and which more plausibly represent fundamental norms. Rini very sensibly observes that, given this conception, empirical evidence about the sources of one’s intuitions might be very relevant. What I want to insist is that many cases of a priori reasoning do not rely on such psychological inputs.

162

Advances in Experimental Philosophy and Philosophical Methodology

(I will return in section 4 to the question of whether psychological evidence can be relevant for the evaluation of arguments in these areas.) So I do agree with the central conclusion of Herman Cappelen (2012): it is a mischaracterization of how philosophical and other a priori reasoning works to suppose that it generally relies in a central evidential way on intuitions. (I defend this claim further in Ichikawa 2014a.) In the remainder of this paper, I will treat this as established, and turn to the question of its implications for the experimentalist critiques of philosophical methods that are often expressed in terms of intuitions. My view is intermediate between that defended by Jonathan Weinberg and Joshua Alexander—that it’s completely irrelevant for this experimentalist project—and that defended by Herman Cappelen—that it justifies totally ignoring this experimentalist project. I discuss these two extreme views in the next two sections, respectively.

3. Relevance to experimental philosophy In a series of influential papers, Timothy Williamson has also argued against the assumption that intuitions play important evidential roles in philosophy.13 He suggests that the reason some philosophers tend to think that philosophical conclusions are based on intuition derives from an erroneous principle of “evidence neutrality”—the idea that for anything to be evidence for someone it has to be the sort of thing that all parties to the debate must recognize as evidence. He argues forcefully against this principle, and suggests on this basis that, in the kinds of philosophical cases in which one might be tempted to say that an intuition is the ultimate evidence, it is actually the philosophical fact itself that is playing the evidential role, even though it is a controversial philosophical fact. Williamson’s argument against intuitions as evidence differs from mine in several important ways. Examining some of the responses to Williamson, and how they bear with respect to my approach, is one way to bring out these important differences. Joshua Alexander and Jonathan Weinberg (2007) give the most direct and explicit response I have seen to the Williamsonian approach. They write:



Intuitive Evidence and Experimental Philosophy

163

Timothy Williamson has also developed a more radical response to the restrictionist threat: rejecting the picture of philosophical practice as depending on intuitions at all! He argues that our evidence [in the relevant cases] is not any sort of mental seeming, but the facts in the world. He compares philosophical practice to scientific practice, where we do not take the perceptual seemings of the scientists as our evidence, but the facts about what they observed. Similarly, then, we should construe Gettier’s evidence to be not his intellectual seeming that his case is not an instance of knowledge, but rather the modal fact itself that such a case is not an instance of knowledge. We retreat from talk of the world to talk of percepts when we (mistakenly) attempt to accommodate the skeptic; so, too, do we retreat to talk of intuitions only under the pressure of skeptical arguments. And since Williamson is himself antiskeptical, emphasizing the continuity between ordinary modal cognition and philosophical cognition, he concludes we should give up thinking of our philosophical evidence in the thinly psychological terms of intuitions. But we do not think that Williamson’s arguments can provide much solace for traditional analytic philosophers. For the results of experimental philosophers are not themselves framed in terms of intuitions, but in terms of the counterfactual judgments of various subjects under various circumstances. Although the results are often glossed in terms of intuitions to follow standard philosophical usage, inspection of the experimental materials reveals little talk of intuitions and mostly the direct evaluation of claims. (2007: 72)

(This is one of two responses they give to Williamson; the other is suggestive of the line I will press in section 4.) Alexander and Weinberg are giving a version of what Cappelen (2014) calls the “X-Phi-Doesn’t-Need-Intuitions Reply.”14 Although, in their presentations of their critiques of armchair methods, experimental philosophers often fall into the language of “intuition” talk, they are not ultimately relying on any substantive claims about intuitions in important evidential roles; after all, the experimental data themselves do not typically involve the word “intuition.” For example, although Weinberg et al.’s seminal “Normativity and Epistemic Intuitions” (2001) makes frequent use of “intuition” language (including in its title), the application to intuition and to philosophical methodology comes in the interpretation of their data. Their data concern the correlation between demographic variables and answers to questions like this one: One day Charles is suddenly knocked out by a falling rock, and his brain becomes re-wired so that he is always absolutely right whenever he estimates

164

Advances in Experimental Philosophy and Philosophical Methodology

the temperature where he is. Charles is completely unaware that his brain has been altered in this way. A few weeks later, this brain re-wiring leads him to believe that it is 71 degrees in his room. Apart from his estimation, he has no other reason to think that it is 71 degrees. In fact, it is at that time 71 degrees in his room. Does Charles really know that it was 71 degrees in the room, or does he only believe it? (Weinberg, Nichols, and Stich 2001: 439)

Weinberg et al. observed that students’ answers to this question were correlated with demographic variables. This by itself does not straightforwardly imply anything about intuition, unless one assumes that answers to questions invariably express intuitions.15 So to some degree, I think it’s clear that Alexander & Weinberg are correct: the data that experimental philosophers emphasize does not by itself presume any kind of evidential role for intuition. I do not think, however, that this observation is enough to represent a very substantial defense of the experimentalist critique. For, just as it requires substantive philosophical interpretation to move from the experimentalists’ data to claims about intuition, so too does it require substantive philosophical interpretation to move from the data to any kind of challenge to traditional methodology. It is not an explicit tenet of armchair philosophical methods that demographic variables are probabilistically independent from judgments about thought experiments; the experimental results do not directly show a problem. Of course the experimentalist critics are well aware of this fact, which is why they produce extensive arguments concerning the interpretation of these results. Unfortunately for these critics in the present context, many of these arguments do rely heavily and explicitly on intuition language. In arguing from (a) demographic correlations with thought-experiment judgments to (b) problems for traditional philosophical methodology, critics often commit to contentious claims about how philosophers use intuitions. For example, as highlighted in section 2, Weinberg et al. (2001) explicitly target the view they call “Intuition-Driven Romanticism,” which is defined as a view that takes intuitions as substantive input into a process that yields normative conclusions. Their conclusion—that such views face a serious challenge—does not bear on any method that does not take intuitions as substantive input; if I am right that traditional armchair philosophical methods on the whole do not use intuitions in this way, then the conclusion of their paper does not bear on traditional armchair methods. So, even if Alexander and Weinberg are right



Intuitive Evidence and Experimental Philosophy

165

that experimental philosophy results do not essentially make use of intuitions, this doesn’t show that their analysis doesn’t. Experimental philosophy—even the subset that is its negative “restrictionist” program—is a diverse movement, and we shouldn’t assume that every argument against traditional methods makes the same assumption. We must consider the arguments one by one. But it does seem that many experimentalist arguments, at least, do commit to questionable assumptions about intuitions. The argument of the Weinberg et al. paper just mentioned, for example, carries ineliminable commitments to the psychologistic approach to philosophical evidence that Williamson challenges in the passage above, and that I challenged in section 2. In Ichikawa (2014a), I argue that quite a lot of the experimentalist critique carries such problematic commitments. It is worth noting that, contra Kirk Ludwig (2007), whether one uses the word “intuition” is immaterial; the argument that the epistemology does not depend on intuitions in an evidential role does not turn on verbal questions about how to use the word “intuition.” It just doesn’t matter whether “intuition” is used broadly to refer to all or most judgments or inclinations to judge, or more narrowly to refer only to judgments with a special etiology, or accompanying a particular phenomenology. Williamson argues that no sort of psychological claim is the central evidence; my argument in section 2 has the stronger conclusion, that no psychological state is or provides the central evidence. So, while I agree with Alexander and Weinberg that one can express the relevant experimental results without the language of “intuitions,” I do not see a clear case that could be made that it’s a straightforward matter to modify experimentalist arguments to avoid appeal to problematic assumptions about intuition. This is a point that Cappelen (2014) emphasizes—he considers various simple strategies, such as replacing the word “intuition” with “philosophical judgment” or “philosophically relevant judgments about cases” in presentations of these arguments, and argues that the result is not a compelling argument. I think that Cappelen is right about this, at least with respect to many cases. For example, the view that traditional armchair epistemology proceeds by way of judgments about cases as central evidential inputs is no more plausible than the view that it proceeds by way of intuitions; the argument of section 2 counts equally against this view. (Someone who does not judge that wishful thinking isn’t a route to justified belief thereby manifests a rational failure, even without

166

Advances in Experimental Philosophy and Philosophical Methodology

what is suggested to be the central evidence.) I conclude that if, as many of us have recently argued, intuitions do not play evidential roles in traditional philosophical methodology, this fact is relevant for experimentalist critiques; it cannot straightforwardly be ignored or translated away.16 I say that the denial of evidential roles has relevance. I do not say with Herman Cappelen that it is definitive against experimental philosophy. In section 4, I explain why.

4. Relevance of experimental philosophy Herman Cappelen thinks, as I do, that the role of intuitions in philosophical methodology has been greatly exaggerated. He also thinks, as I do not, that this renders experimental philosophy irrelevant: The Big Objection to experimental philosophy is easy to state and should be obvious: philosophers don’t rely on intuitions about thought experiments, so studies of intuitions people have about thought experiments have no direct relevance for philosophical arguments or theorizing. … Negative experimental philosophy attacks a practice that doesn’t exist. Positive experimental philosophers attempt to support a practice that doesn’t exist. In short: If philosophers don’t rely on intuitions, then the project of checking people’s intuitions is philosophically pointless. (Cappelen 2012: 221–2)

Although I do think, as I suggested in section 3, that some experimental philosophy is subject to this criticism, it seems to me that Cappelen goes too far. Not all experimental philosophy is irredeemably based in a misconception about the role of intuitions in traditional philosophical methods. Not even all of the restrictionist critique relies on such misconceptions. It is consistent with the claim that intuitions do not play evidential roles in philosophical methodology that the empirical investigation of intuitions may be relevant for philosophical methodology—even that in theory, they may show deep problems with it. In the abstract, this should be relatively easy to see; there’s no contradiction in the idea that a method doesn’t make central evidential use of X, but that the study of X may nevertheless be methodologically important for that practice—evidential roles aren’t the only roles of epistemic significance. For example, I don’t think that judgments about gender play evidential roles



Intuitive Evidence and Experimental Philosophy

167

in hiring decisions regarding university biology laboratory manager positions; biology professors base such hiring decisions on scientific qualifications. The gender of the candidate isn’t evidence regarding one’s suitability for such a job; moreover, scientists recognize that gender isn’t evidence regarding one’s suitability for such a job. (A careful study of the discussions biologists have in their hiring decisions, similar to the case studies in Cappelen (2012), I’d conjecture, would not yield evidence of reliance on gender.) None of this is any reason to ignore Corinne Moss-Racusin et al.’s (2012), which finds evidence that scientists display systematic subconscious bias against women. Even setting aside the moral and social justice issues here, Moss-Racusin et al.’s findings, if true, demonstrate a significant epistemic source of bias—empirical methods have demonstrated a respect in which scientists’ beliefs about who would make good laboratory managers go systematically awry. It would be a mistake for a scientist to ignore this critique on the grounds that gender does not play evidential roles in scientific hiring—not because it does, but because evidential roles aren’t the only things that matter, even from an epistemic point of view. One has epistemic reason to consider sources of epistemic defeat, even independently of considerations about what is and isn’t evidence. Some experimentalist critiques of traditional philosophical methods seem naturally interpretable along just these kinds of lines. When Stacey Swain et al. (2008) demonstrate that epistemic intuitions are subject to an order effect, this is of epistemic significance to philosophers who are making epistemic judgments—not because the intuitions, or the judgments themselves, or indeed anything psychological, are the evidence they’re relying upon, but because these empirical findings constitute a possible reason to doubt one’s own ability to respond rationally to the available evidence. Recall the case of Bullwinkle and the chess board. After his blindfold slips, he has all the evidence necessary to know that Natasha can win this turn, but he doesn’t know it. Suppose he believes it anyway; although he can’t see what the winning move would be, he just has a feeling that this looks like a board where white is in a position to win. The fact that Bullwinkle isn’t very good at chess is the sort of thing that ought to give him reason to doubt his judgment—this, even though he does not rely on any claim about his chess skill as evidence. In the same way, order effects in intuitions about philosophical thought experiments are epistemically significant, regardless of whether one is using intuitions in

168

Advances in Experimental Philosophy and Philosophical Methodology

evidential roles. If I learn through empirical means that I tend to be influenced in my judgments by irrelevant factors, this gives me good reason to take extra care that I’m not going wrong.17 Similarly, when Tamara Horowitz (1998) argues that certain controversial deontological moral intuitions are the result of a systematic tendency to overestimate the value of what one thinks of oneself as already having, these kinds of psychological data can be relevant for deciding which arguments to accept. Such, of course, is continuous with the traditional armchair project of explaining away intuitions.18 Cappelen recognizes a version of the experimentalist critique along these lines. In Cappelen (2014), he presents this fictionalized bit of dialogue between himself and a defender of experimental philosophy: X-P: So is your view that we should not care about or try to find biases in philosophical practice? We should just ignore them? Isn’t that an irresponsible intellectual attitude? HC: That’s not my view. There are lots of biases in philosophy. Where you and I differ is on how to discover and evaluate those. I think we discover those by doing philosophy—by finding assumptions and presuppositions that bias philosophical reflection. Construed in that way, I see myself as spending a great deal of my time as a philosopher discovering philosophical biases. The discussion at the core of this paper is a good illustration: I think much metaphilosophy is biased in favour of the view that philosophical practice is intuition-based. That bias has distorted metaphilosophical reflections. But discovering that bias, and justifying opposition to it, doesn’t involve surveys. It is about doing philosophy like we were trained to do it: it involves thinking hard about arguments, uncovering hidden assumptions and then questioning them. (Cappelen 2014: 284)

But one needn’t—and, in my view, shouldn’t—think of the most plausible versions of the experimentalist critique as in tension with Cappelen’s selfattribution. Cappelen here admits that philosophical biases are epistemically relevant, and that it is worthwhile for philosophers to expend energy in the attempt to detect them; the contrast he draws is between the idea that one should investigate them with surveys, or by “doing philosophy like we were trained to do it.” If what I have said in this section is right, then this is a false dichotomy (and not just because some philosophers were trained to do philosophy with surveys!). We should think of the experimentalists’ surveys, in at least some instances, as a part of “philosophy like we were trained to



Intuitive Evidence and Experimental Philosophy

169

do it.” It is, I think, incumbent upon a defender of a restrictionist challenge to explain why her data casts doubt on traditional methods—and I agree with Cappelen that, too often, the proffered explanations make unrealistic assumptions about the role of intuitions in those traditional methods—but this isn’t always so. As I suggested in Ichikawa (2014a), I think that experimental philosophy tends to be perceived, by many of its practitioners and most of its critics, as a radical departure from standard methods; once we dispense with the assumption of anything like “intuition-driven romanticism” as a descriptive thesis of traditional methods, we should should consider that perception an error. And the arguments of Cappelen and of Williamson are, I think, very helpful in showing why that is an error. But most sensitive contemporary experimental philosophers expressing concerns about traditional methodology are more careful to express their concerns in ways that do not call for such a radical departure from traditional methods; they are calling instead for a more systematic and scientific treatment of the psychological questions concerning philosophers’ biases. Perhaps traditional methods have the resources to satisfy that call;19 but it is not one it is legitimate wholly to ignore.20

Notes   1 Some experimentalist critics of traditional philosophical methods do not charge that intuition is unreliable, but that it is epistemically deficient in some weaker way—e.g. the ‘hopelessness’ of Weinberg (2007).   2 Brown argues (pp. 512–13) that an externalist approach like the reliabilist one described here is necessary to defend the psychologistic model from skeptical worries. But other defenders of similar models, such as Chudnoff (2013), develop their views in internalist ways.   3 Ludwig (2007) argues that “intuition” is factive; nothing counts as an “intuition” unless it is the successful exercise of a conceptual competence, which will always issue into truth. This seems to me rather clearly to be a somewhat revolutionary sense of the term. Peter Singer knows full well that many people have the intuition that one has special moral obligations to one’s kin; he thinks that intuition is mistaken. Whether or not his ethical view is correct, he is not contradicting himself by thinking that these are false intuitions.

170

Advances in Experimental Philosophy and Philosophical Methodology

  4 Rizzieri (2009) gives dissent; I agree with Unger (1975), Williamson (2000), and Littlejohn (2013) that evidence cannot be false. I am particularly impressed by this argument, which is related to the discussion of Williamson (2000: Ch. 9): if one could have false evidence, then one’s evidence could entail a false hypothesis, which would mean that one has conclusive evidence for something false. But by definition, if I have conclusive evidence for something, it must be true.   5 Whether p could thereby be evidence depends on one’s theory of evidence; if Williamson (2000: Ch. 9) is correct, then it certainly could.   6 Ichikawa and Jarvis (2013: Ch. 11)   7 One might wonder, as Jennifer Nado did in personal correspondence, whether it’s necessary for the kind of rational pressure I’m describing that Bullwinkle could form the intuition under the circumstances. I make three observations in response. First, even if it were true that Bullwinkle only has propositional justification if he is capable of forming the relevant intuition, this would not suffice to defend experimentalist rationalism from the critique; according to experimentalist rationalism, it is something about the experience of an intuition itself that does justificatory work. (For example, on the view of Chudnoff (2013), the phenomenology of intuitions plays the relevant role.) Positing a kind of counterfactual intuition as a necessary condition for propositional justification does not support any central epistemic role for actual intuitions. Second, in the plausible ways of spelling out this case, there is a straightforward sense in which Bullwinkle could have the relevant intuition—this is what would happen if he thought a little harder and more carefully. Third, Jarvis and I argue that the counterfactual necessary condition is false. See Ichikawa and Jarvis (2013: 282–3). So even if Bullwinkle couldn’t have the intuition, that doesn’t prevent his situation from constituting a rational failure. Indeed, it is reasonably natural to think that his inability to have such intuitions contributes, causally or constitutively, to his rational failure.   8 In previous work, Jarvis and I have called these cases of ‘blind irrationality’.   9 Note that to say there is rational pressure here leaves open many normative questions. In particular, it does not imply that one is criticizable for failing to believe what one has conclusive reason to believe. It also does not imply that if one does believe what one has such reason to believe, one has formed a doxastically justified belief. One can believe what one has reason to believe, but do so for the wrong reasons. Thanks to Michael Johnson for discussion here. 10 Thanks to Jennifer Nado for reminding me of the need for this discussion. 11 I do not think that this is a point of contention between Chudnoff and myself; see, e.g. Chudnoff (2013: 85), where he clarifies that the notion of justification



12

13 14

15

16 17

18 19 20

Intuitive Evidence and Experimental Philosophy

171

at issue in the corresponding claim about perceptual justification concerns propositional, rather than doxastic, justification. Since his broad project is to argue that intuitive experience and perceptual experience play similar roles, and since he doesn’t say otherwise, I assume that Chudnoff ’s view is that intuitions establish propositional justification. (Ichikawa and Jarvis [2013: 298] gives arguments against intuitions’ playing a special role in establishing doxastic justification.) The very natural question of what does provide propositional justification is beyond my present scope. Ichikawa and Jarvis (2013) argues at length that in cases like this, it is a brute fact that all subjects (or at least those with the relevant concepts who are considering the question) always have such justification, regardless of experiences. E.g. Williamson 2004; Williamson 2007. Alexander himself may have changed his mind about this; in Alexander (2010: 382) he writes of the idea that “intuitions neither need nor ought to be treated as evidence in philosophy” that “[i]f true, this claim would seem to threaten the significance of experimental philosophy.” He goes on to that this claim is not true, but his argument seems to me to conflate persuasive ability with evidence. Indeed, Chudnoff (2013: 109–12) uses this observation to defend an evidential role for intuitions from experimentalist arguments; Chudnoff (2013) and Alexander and Weinberg (2007) are each defending, from opposite sides of the divide, a kind of independence between questions about intuitions and questions about the experimentalist critique of traditional methodology. Ichikawa (2014a) also contains arguments to this effect. It is of course controversial whether experiments like those of Swain, Alexander, and Weinberg (2008) do give professional philosophers reason to doubt themselves—for example, one might think that philosophers are sufficiently different from the folk who are the targets of these studies. I suspect that, in many cases, responses like this on behalf of traditional methods may be defensible. (In Ichikawa [2011] I argue that traditional methods are generally able to meet the experimentalist critiques.) But this is to engage directly with the experimentalist critics; my claim here is that, in such cases, this is what is necessary. We cannot sidestep them by denying an evidential role for intuitions. See Ichikawa (2009) for further discussion of this point. Ichikawa (2011) argues that it is. Thanks to Michael Johnson and Jennifer Nado for helpful comments on a draft of this paper; thanks to Jessica Brown, Herman Cappelen, Yuri Cath, Carrie Jenkins, Jonathan Weinberg, and Timothy Williamson for important

172

Advances in Experimental Philosophy and Philosophical Methodology

conversations over the years that helped develop my thinking on these matters. A version of this paper was presented at a May 2015 workshop on Philosophical Methodology in Hong Kong; thanks also to the audience there for helpful discussion. I owe a particular intellectual gratitude to Benjamin Jarvis, with whom I first developed many of the ideas discussed here.

References Alexander, J. (2010), “Is Experimental Philosophy Philosophically Significant?,” Philosophical Psychology 23 (3): 377–89. Alexander, J. and Weinberg, J. (2007), “Analytic Epistemology and Experimental Philosophy,” Philosophy Compass 2 (1): 56–80. Brown, J. (2011), “Thought Experiments, Intuitions and Philosophical Evidence,” Dialectica 65 (4): 493–516. Cappelen, H. (2012), Philosophy Without Intuitions, Oxford: Oxford University Press. Cappelen, H. (2014), “Experimental Philosophy Without Intuitions,” in A. Booth and D. Rowbottom (eds), Intuitions, Oxford: Oxford University Press, 269–86. Chudnoff, E. (2013), Intuition, Oxford: Oxford University Press. Deutsch, M. (2010), “Intuitions, Counter-Examples, and Experimental Philosophy,” Review of Philosophy and Psychology 1 (3): 447–60. Horowitz, T. (1998), “Philosophical Intuitions and Psychological Theory,” Ethics 108 (2): 367–85. Ichikawa, J. J. (2009), “Explaining Away Intuitions,” Studia Philosophica Estonica 2 (2): 94–116. Ichikawa, J. J. (2011), “Experimentalist Pressure Against Traditional Methodology,” Philosophical Psychology 25 (5): 743–65. Ichikawa, J. J. (2014a), “Who Needs Intuitions? Two Experimentalist Critiques,” in A. Booth and D. Rowbottom (eds), Intuitions, Oxford: Oxford University Press, 232–56. Ichikawa, J. J. (2014b), “Intuitions in Contemporary Philosophy,” in L. Osbeck and B. Held (eds), Rational Intuition: Philosophical Roots, Scientific Investigations, Cambridge: Cambridge University Press. Ichikawa, J. J. and Jarvis, B. (2013), The Rules of Thought, Oxford: Oxford University Press. Littlejohn, C. (2013), “No Evidence is False,” Acta Analytica 28 (2): 145–59. Ludwig, K. (2007), “The Epistemology of Thought Experiments: First Person Versus



Intuitive Evidence and Experimental Philosophy

173

Third Person Approaches,” in P. French and H. Wettstein (eds), Midwest Studies in Philosophy 31: 128–59. Hoboken, NJ: Blackwell Publishing. Moss-Racusin, C., Dovidio, D., Brescoll, V., Graham, M., and Handelsman, J. (2012), “Science Faculty’s Subtle Gender Biases Favor Male Students,” Proceedings of the National Academy of Sciences 109 (41): 16474–9. doi:10.1073/pnas.1211286109 Rini, R. (2013), “Making Psychology Normatively Significant,” The Journal of Ethics 17 (3): 257–74. Netherlands: Springer. doi:10.1007/s10892-013-9145-y Rizzieri, A. (2009), “Evidence Does Not Equal Knowledge,” Philosophical Studies 153 (2): 235–42. Swain, S., Alexander, J., and Weinberg, J. (2008), “The Instability of Philosophical Intuitions: Running Hot and Cold on Truetemp,” Philosophy and Phenomenological Research 76 (1): 138–55. Unger, P. (1975), Ignorance: A Case for Scepticism, Oxford: Oxford University Press, 105. Weinberg, J. (2007), “How to Challenge Intuitions Empirically Without Risking Skepticism,” Midwest Studies In Philosophy 1 (31): 318–43. Weinberg, J., Nichols, S., and Stich, S. (2001), “Normativity and Epistemic Intuitions,” Philosophical Topics 29 (1–2): 429–60. Fayette, AR: University of Arkansas Press. Williamson, T. (2000), Knowledge and Its Limits, Oxford: Oxford University Press. Williamson, T. (2004), “‘Philosphical Intuitions’ and Scepticism About Judgement,” Dialectica. Williamson, T. (2007), The Philosophy of Philosophy, Hoboken, NJ: Blackwell Publishing.

Index */?/?? system 29 abduction 83–6 Alexander, J. 73, 108–9, 162–5, 171 n.14 Alston, W. P. 36 Americans 101, 130–1 anti-Centrality position 126–8 anti-intuitionism 19–20 anti-luck argument 73, 75–7, 94 n.11 anti-luck epistemology 81 a priori reasoning 161–2 argument from diversity 6, 101–3 Argument from Error 133–4 Argument from Ignorance 133 argument from opaque irrationality 159, 161 argument from rational improvement 160, 161 armchair 6–7, 19–22, 45, 99, 143–4, 148, 149, 163–5, 168 armchair philosophy, extra-mentalist interpretations of 106 Aristotle 35, 133 Austin, J. L. 35 Bealer, G. 2, 19, 73 belief evaluation truth–insensitive approach to 42–3 truth–sensitive approach to 43–4 Bishop, M. 57 blind irrationality 170 n.8 Bonjour, L. 42 Boyd, K. 12 brain-in-a-vat 42–5 Brogaard, B. 81–2, 95 n.17 Brown, James 57–8 Brown, Jessica 156, 169 n.2 Cappelen, H. 3–4, 53, 87–8, 95 n.19, 109, 112–13, 115–17, 125–9, 136–7, 139, 152 n.6, 153 n.11, 156, 162–9

Centrality 3, 125–9 relevance to experimental philosophy 162–6 and theories of reference 135–44 central processing 137 Chalmers, D. 87–8 Chisholm, R. 43 Chudnoff, E. 160–1, 170 n.11, 171 n.15 Cicero 133 Clark, M. 78–80 cluster concepts 24 Cohen, S. 39–40 Cokely, E. 101 Columbus 133–4 common ground 116–18, 128, 137 common sense 35, 82 commonsense epistemology 5, 35–6, 40–1, 44–7 concept mastery 19–20 concepts 19–20, 46–7, 103–8 conceptual analysis 77 the mentalist move to 103–8 conceptual truths 159 conciliationists 103 Condorcet Jury Theorem 13, 107 Conee, E. 43 confirmation bias 27, 30 consciousness 63–4, 149, 151 contextualism 37 low/high cases 37–40 controlled experiments 46, 153 n.16 Cooper, R. 57 counterfactuals 23 David and Bathsheba 61–2 De Blanc, M. 141 deferral hypothesis 39–40 demographic variables 163–4 DeRose, K. 37–9 descriptive metaphilosophy 88

176 Index descriptivism 110, 132–4, 145 strong vs weak 132–3 theory of meaning vs theory of reference 132–3 Deutsch, M. 53, 109–12, 114–17, 119, 125–8, 135, 137, 142, 156 Devitt, M. 108, 132–7, 144–6 disagreement 17, 29, 65, 81 among epistemic peers 6–7, 102–3, 116 and future philosophy 118–19 Dorr, C. 116–17 doxastical justification 158 Dretske, F. I. 36 Dunaway, B. 107 E=K 20 East Asians 7, 101–2, 131 Edmonds, A. 107 Einstein 54, 134 Elga, A. 119 n.4 elicited production 142–4 epistemic closure principle 40–2 epistemic defeat 167 equal weight view 119 n.4 error-detection 21–2 error profile 31 n.2 esoteric cases 18, 20, 25 evidence 158, 166–7 defeasible 112–13 evidence neutrality 162 experientialist rationalism 157 experiential states attribution 64 experimental philosophy master argument of 2 negative program in 1, 26, 99, 151 positive projects 118, 151 relevance of 166–9 results vs analysis 165 restrictionist program 165 expertise defense 3–4, 102–3, 136–7 fake-barn cases 44–5 family resemblances 24 Feltz, A. 101 Feynman 133 fiction 65 fictional narratives 58–9, 63–6 first-person exceptionalism 61 Fodor, J. 23

folk beliefs 107 folk psychology 19 formal models 23 free will 101 Fumerton, R. 40, 43 fundamental attribution error 102 Galileo 55 gender bias 166–7 Gendler, T. 57–8, 61–2, 65 Generality Problem 12 generics 24–5, 57 Gettier, E. 6, 56, 69 Gettier cases 6, 24–6, 29, 56–7, 128, 138, 149–50 10 Coins Case 72, 75, 90, 111–12, 114 Brown in Barcelona Case 72, 75–6, 91, 93 n.5, 113–14 content problem 93 n.6 logical form of 93 n.6 no false lemmas solution 78–9 textbook renditions 87–9, 96 n.22 Gettierology 77–81 Gettier problem 77 Gettier’s original cases 90–1 Gödel case 110, 129–30, 132–4, 141, 145 Goldman, A. 12, 36, 44, 79–80, 104, 119 n.5 Grice’s razor 126, 140 Hawthorne, J. 40 hindsight bias 102 Hong Kong 101, 130–1 Horowitz, T. 168 Horvath, J. 2 hopelessness 169 n.1 hypothetical cases 144–8, 153 n.16 Ichikawa, J. 132, 134–5, 138 identity, claims of 23–4 “I do know” 39–40 “I don’t know” 38–40 imagination 6, 59–62 imaginative resistance 65 interim assumption 117–18 introspection 5, 45–6, 156 intuition a posteriori 108–9 a priori 99, 108–9

Index vs argumentation 3–4, 70, 128 baseline reliability of 12–13, 17, 21 classical philosophical appeal to 19 conceptual competence 127, 169 n.3 cultural diversity in 7, 101–2, 130–1, 134, 155 definition of 100 distinctive phenomenology 127 and evidence 156–62 explaining away 16–7 heterogeneity 128–9, 137–8 innate 137 linguistic vs. metalinguistic 141–2 as noninferential 100, 109 nonpsychologistic view of 157 perceptual judgments 127, 136 rational pressure and 159–60 as seeming 81, 100, 161 as a source of evidence in philosophy 19 and theoretical considerations 138–41 theory free 137 theory-laden 108, 136–7 “thin” conception of 137 Intuition Driven Romanticism 155, 164, 169 “intuition”-talk 115–18, 163 intuitive error 16–7, 21–2 intuitive verdict 71 “it is intuitive” claims 117 Jackson, F. 104, 106 “Jack the Ripper” 134 Jarvis, B. 157–8 Jonah case 129–30, 132, 134 judgment skeptic 2, 4 justification, and truth 42–4 knowledge causal theory of 79–80 justified true belief (JTB) theory of 23–4, 29, 69, 138 necessary conditions on 76 knowledge attribution 36, 40, 44, 45 knowledge-preventing luck 77, 79 Kripke, S. 7, 110–11, 125–6, 129–34, 139–41 Laird, J. 59 linguistic behavior 7, 37, 142, 145

177

Livengood, J. 101 Locke, J. 35 Ludwig, K. 165, 169 n.3 Machery, E. 7, 101, 125, 131, 141 Maitra, I. 132 Mallon, R. 125, 131 Malmgren, A. 71, 92 n.3, 93 n.6 Manley, D. 107 Martí, G. 132, 134–5, 141–4, 148 mental content 60 mental models 59 mental representations 59 metatheory 18, 20–1 method of cases 21, 69, 92 n.3, 108, 126, 129, 131, 152 n.6 cooperative relationships between experimental philosophy and 63–6 intuitions and 138–41 usual construal of 71–2 modal argument 56–7 model-selection Akaike information criterion 28–9 holistic criteria of 26–9 philosophical theory-selection and 28 modes of inference 18 moral objectivism 101 moral psychology 161 moral responsibility 101 Moss-Racusin, C. 167 Nagel, J. 12, 19, 76, 107 Naming and Necessity 111, 125, 132 narrative eliminativism 57 narrative incompleteness 58, 65 natural kind 129, 145–7 Nersessian, N. 59–60 Neta, R. 73 Nichols, S. 101, 120 n.7, 125, 155, 163–4 Nisbett, R. 101–2 Norton, J. 54–60 numerosity judgments 137 Olivola, C. 141 order effect 167 order of explanation 83–4 ordinary language 35 overfitting data 28

178 Index paradox 22–3 parsimony 26–7 Peano axioms 22, 145 perceptual experience 159–61 analogy between intuition and 160–1 personality traits extroversion 101 openness to experience 101 phenomenal consciousness 63–4 phenomenological worry 57–8 philosophical cognition 6, 63–4 philosophical exceptionalism 60 philosophical expertise 4, 102–3 philosophical intuition 2, 53 philosophical methods experimentalist critiques of 162–6 Platonic armchair picture of 6–7, 99 philosophical starting points 82 philosophical truths, modal force of 23 philosophy of science 6, 54 Philosophy Without Intuitions 87 physics 54–7 prescriptive epistemology 47 presuppositions 118 principle of charity 94 n.9 proper names 101, 110, 132–3 propositional justification 158–61 prototypes 25 psychologistic model 156–7 psychologization of evidence 4, 128 Pust, J. 2, 104, 119 n.5 Ramsey-sentence 113 rational necessities 159 Rawls, J. 113 reconstruction thesis 57–8 reductio argument 55 reflection 3, 21, 106, 161 reflective equilibrium 112 Reid, T. 35, 46 reliabilism 36 proto-reliabilist hypothesis 36–7 reliability 155 knowledge and 36–7 methodological inadequacy of 11–17, 21 Rini, R. 161 “Rock” evidential status 113, 139

Sellars, W. 35 semanticists 140 semantic reference 139–40 semantics 147 Fregean principles 140 shared or common concepts 106 sheep-in-the-field case 95 n.14 simplicity 27, 138, 140–1 skeptical pressure cases 25 skepticism about the concept of intuition 2–3, 127 about counterfactual cases 134, 144–8 Sosa, E. 36, 65 source selection 17–18 speaker reference 139–40 stakes effects 25, 37–40 standard justificatory procedure 53 steadfast theorists 103 Steup, M. 40–1 Stich, S. 101, 107, 125, 155, 163–4 Stroud, B. 40 subconscious bias 167 subject-object-verb (SOV) language 150 subject-verb-object (SVO) language 150 suppositions 93 n.6, 93 n.8, 116 Sutton, J. 43 Swain S. 167 syntax 17–18, 147 Sytsma, J. 101 Talbot, B. 107 theories exception-intolerant 5, 20, 22 exception-tolerant 5, 22–3 structure 18 Thomson, J. 62 thought experiments argument view of 6, 54–8 and experimental philosophy 63–6 mental model view of 6, 58–62 producing vs. consuming 84–6 Ulatowski, J. 120 n.7 unicorn 140–1 universal generalization 20, 57 universals 104 “Unwanted Necessity” argument 133

Index variation argument 1–4 Violinist Case 62 Weatherson, B. 16, 132, 138–9 Weinberg, J. 73, 101, 108–9, 155, 162–5 Westerners 7, 101–2 “what we would say” 148–9

179

Williamson, T. 2, 4, 19, 23, 53–4, 56–7, 60–1, 63, 66, 100, 106, 108, 125–8, 136–7, 156, 162, 165, 169 x-phi 5–6, 21, 25–6, 29–30 see also experimental philosophy X-Phi-Doesn’t-Need-Intuitions Reply 163